framework,version,device,op_name,kernel_source,gemm_dtype,m,n,k,latency
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,12288,7.421629163953994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,16384,9.612958272298178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,12288,9.984360588921442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,16384,13.309599982367622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,10240,5.034554799397786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,10240,8.17531246609158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,8192,6.564392937554254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,8192,3.858886294894748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,16384,6.656256781684028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,12288,4.977041880289714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,10240,4.016924540201823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,7168,5.592669169108073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,7168,3.4057831234402127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,8192,3.09119012620714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,6144,4.720499674479167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,6144,2.9603936937120228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,7168,2.755605273776584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,6144,2.4111760457356772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,5120,3.901288774278429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,5120,2.350407918294271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,5120,1.8990569644504125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,4096,1.4867075814141169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,4096,3.108396530151367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,4096,1.7762799792819555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,4096,0.8679964277479383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,6144,1.347256024678548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,5120,1.1262630886501737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,7168,1.5725901921590169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,8192,1.6646862030029297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,10240,2.292424096001519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,3584,1.6132835812038844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,16384,3.922161950005425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,12288,2.8349492814805775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,3584,2.691184785630968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,3072,1.4973866144816081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,3072,2.156425688001845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,2560,1.8522656758626301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,2560,1.2620400322808158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,2048,1.3872079849243164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,2048,0.9351128472222222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,3584,0.7855582237243652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,3072,0.6856808662414551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,2560,0.6008613374498155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,1536,1.067005369398329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,1536,0.750091552734375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,2048,0.5023226737976074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,1024,0.7350933286878797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,1024,0.8182160059611002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,1536,0.401234679751926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,1024,0.3083706696828206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,768,0.5673208766513401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,768,0.7503217591179742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,3584,1.2828302383422852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,768,0.2657048967149523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,512,0.4341768953535292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,512,0.5072453286912706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,2048,0.7662862141927084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,3072,1.170405281914605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,2560,0.9526106516520182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,256,0.38103199005126953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,256,0.5320240126715766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,1536,0.6154186460706923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,128,0.2165608935885959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,128,0.3544035487704807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,512,0.22541599803500703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,64,0.19890666007995605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,64,0.3249155680338542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,65536,32,0.19409510824415419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,65536,32,0.3313902219136556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,256,0.20500177807278105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,1024,0.4643920262654622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,65536,128,0.20964622497558594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,65536,10.373445298936632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,65536,13.163692050509981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,768,0.3627315627204047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,16384,3.0954666137695312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,16384,2.55803468492296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,65536,5.096800062391493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,16384,1.1521857579549153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,12288,2.3037565019395614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,12288,2.0819369422064886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,12288,0.8263760142856174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,10240,1.8383074866400824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,10240,1.2281804614596896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,10240,0.6865013440450033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,512,0.3069777753618029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,8192,1.3894613054063585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,8192,0.9785618252224393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,256,0.24812976519266763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,8192,0.4973715676201715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,7168,1.204127099778917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,7168,0.8742124239603678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,65536,128,0.27090133561028373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,7168,0.4236728880140517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,6144,1.0524329079522026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,6144,0.7930062082078723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,6144,0.3652702172597249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,16384,1.5896533330281575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,65536,7.937711079915364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,5120,0.87118042839898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,5120,0.698422220018175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,12288,1.164446194966634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,4096,0.6741937531365289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,4096,0.514944871266683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,5120,0.3045857747395833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,4096,0.2452497747209337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,10240,0.9676888783772787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,3584,0.6190151108635796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,3584,0.4309253427717421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,3584,0.2167208857006497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,3072,0.5313999917772081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,3072,0.37808889812893337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,3072,0.191064887576633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,8192,0.7782630920410156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,2560,0.42905955844455296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,2560,0.295289781358507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,2560,0.16547644138336182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,7168,0.6867493523491753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,2048,0.3424657715691461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,2048,0.28178222974141437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,6144,0.5894719759623209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,2048,0.13814577791425917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,1536,0.2735208935207791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,1536,0.2025688886642456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,1536,0.10936711231867473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,5120,0.4970533582899306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,1024,0.18427111042870414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,1024,0.1613644493950738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,1024,0.08444977468914455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,4096,0.40726576911078555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,768,0.1502702236175537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,768,0.1453146669599745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,3584,0.35761245091756183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,768,0.072507553630405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,512,0.11128444141811794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,512,0.10957688755459255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,3072,0.31155021985371906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,512,0.06384266747368707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,256,0.07669333616892497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,256,0.0985671083132426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,256,0.06131111250983345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,2560,0.2635128762986925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,128,0.05636622508366903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,128,0.08993955453236897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,2048,0.21860267056359184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,16384,128,0.06186222367816501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,64,0.04753777715894911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,64,0.08722844388749863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,16384,32,0.04872444603178236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,16384,32,0.0884515576892429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,1536,0.17385244369506836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,65536,10.937650892469618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,1024,0.13003822167714438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,65536,7.7466684977213545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,65536,3.7902115715874563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,16384,1.6626248889499242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,16384,2.4673298729790583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,16384,0.8601217799716525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,768,0.10627644591861302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,12288,1.7995351155598958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,12288,1.1943768395317924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,12288,0.5622106658087836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,10240,1.3052008946736653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,10240,0.9257164001464844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,10240,0.45508797963460285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,512,0.0916080011261834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,8192,1.0693635940551758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,8192,0.9309946695963541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,8192,0.358171542485555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,256,0.07159288724263509
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,7168,0.984791119893392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,7168,0.6666426658630371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,7168,0.31490310033162433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,16384,128,0.06254133250978258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,6144,0.8581733173794217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,6144,0.6769839922587076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,6144,0.27400533358256024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,65536,6.148663838704427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,5120,0.6575457784864638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,5120,0.4241155518425836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,5120,0.2316088941362169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,16384,1.2399733861287434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,4096,0.5116044680277506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,4096,0.3904675642649333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,12288,0.9364408916897244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,4096,0.18773155742221406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,3584,0.4520088831583659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,3584,0.31016000111897785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,3584,0.16722310913933647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,10240,0.8004533449808756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,3072,0.39762311511569554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,3072,0.2716133329603407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,3072,0.1488800048828125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,8192,0.6053199768066406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,2560,0.3232426643371582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,2560,0.22449154324001738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,2560,0.12815645005967882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,7168,0.5319022072686089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,2048,0.26377778583102757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,2048,0.1878417730331421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,6144,0.46218310462103945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,2048,0.10769244697358872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,1536,0.20724799897935656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,1536,0.15074755085839167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,1536,0.08551466464996338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,5120,0.39213066630893284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,1024,0.14556977483961317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,1024,0.12314132849375407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,1024,0.06584622462590535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,4096,0.3149813281165229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,768,0.11255021890004475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,768,0.10780621899498834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,768,0.05609244770473904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,3584,0.27847645017835826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,512,0.08157600296868218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,512,0.08908622132407294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,512,0.05111111203829447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,3072,0.24129067526923287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,256,0.05679644478691948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,256,0.07817422019110785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,256,0.049878223074807056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,2560,0.20745600594414604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,128,0.04379466507169935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,128,0.06964177555508085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,12288,128,0.050275554259618126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,64,0.03791555431154039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,64,0.06752000252405803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,12288,32,0.039447112215889826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,12288,32,0.06847822003894381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,2048,0.1700026724073622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,65536,8.745628356933594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,65536,6.646870930989583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,65536,3.33094851175944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,1536,0.13454755147298178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,16384,1.9796346028645833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,16384,1.439393785264757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,1024,0.10072088903850979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,16384,0.6910888883802625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,12288,1.4548898273044164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,12288,1.063886218600803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,768,0.08331644535064697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,12288,0.45745065477159286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,10240,1.1672746870252821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,10240,0.9196373621622721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,10240,0.3738853401607937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,512,0.07322577635447185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,8192,0.8658275604248047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,8192,0.6960870954725477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,8192,0.29845955636766225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,256,0.0592764417330424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,7168,0.7959866523742676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,7168,0.6003973219129775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,7168,0.2640168931749132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,12288,128,0.05191555619239807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,6144,0.678551091088189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,6144,0.5134515762329102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,6144,0.2296382321251763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,65536,5.128420511881511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,5120,0.573452419704861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,5120,0.40223556094699436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,5120,0.19946844047970244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,16384,1.1103297339545357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,4096,0.45416620042588973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,4096,0.3387306796179877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,4096,0.16167110866970485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,12288,0.8077573246426053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,3584,0.3881839911142985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,3584,0.3009377850426568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,3584,0.1441511048210992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,10240,0.6540257665846083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,3072,0.32437154981825084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,3072,0.24660089280870226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,3072,0.12857422563764784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,8192,0.522125350104438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,2560,0.2729502254062229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,2560,0.1891066630681356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,2560,0.1111875507566664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,7168,0.4568062358432346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,2048,0.2227271132998996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,2048,0.15807910760243735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,2048,0.09268089135487874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,6144,0.39576265547010636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,1536,0.17117867204878065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,1536,0.12546133995056152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,1536,0.07310755385292901
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,5120,0.3311137888166639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,1024,0.11944533718956842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,1024,0.09183644586139256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,1024,0.05579288800557455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,4096,0.27342221471998424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,768,0.0947822199927436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,768,0.08180799749162462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,768,0.0485173331366645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,3584,0.23905332883199057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,512,0.06936355431874593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,512,0.07168710894054837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,512,0.044694221682018705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,3072,0.20879556073082817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,256,0.04792177677154541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,256,0.06514044602711995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,256,0.04444977641105652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,2560,0.1781991057925754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,128,0.03773955504099528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,128,0.059177776177724205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,10240,128,0.04424444503254361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,64,0.03472088774045309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,64,0.0576346649063958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,10240,32,0.03654844562212626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,10240,32,0.05827644136216906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,2048,0.14591911103990343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,65536,6.201783921983506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,65536,5.591719309488933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,65536,2.5167937808566623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,1536,0.11431822511884902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,16384,1.4777190950181749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,16384,1.0619191063774955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,16384,0.5037608676486546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,1024,0.08735377920998467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,12288,1.0108150906032987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,12288,1.0512755711873372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,12288,0.3638586733076308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,768,0.07421777645746867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,10240,0.8651128874884711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,10240,0.7495475345187717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,10240,0.3018248875935872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,512,0.06450221935908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,8192,0.7512488894992404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,8192,0.5883955425686306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,8192,0.24464888042873809
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,256,0.052935110198126904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,7168,0.6359475453694662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,7168,0.48475556903415257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,7168,0.2169333299001058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,10240,128,0.046190222104390465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,6144,0.532152016957601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,6144,0.4523590935601129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,6144,0.19047733147939047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,65536,4.137847052680121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,5120,0.4690044191148546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,5120,0.30148887634277344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,5120,0.16660000218285456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,16384,0.9213582144843208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,4096,0.34123733308580184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,4096,0.25293511814541286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,4096,0.13458400302463108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,12288,0.6615120040045844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,3584,0.29975912306043834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,3584,0.2204524411095513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,3584,0.11944354905022515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,10240,0.5412800047132704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,3072,0.2612604565090603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,3072,0.18395911322699654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,3072,0.1053217781914605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,8192,0.43434755007425946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,2560,0.21703199545542398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,2560,0.1643822193145752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,2560,0.09108444717195298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,7168,0.38541245460510254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,2048,0.17908088366190592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,2048,0.1299742195341322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,2048,0.0773457752333747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,6144,0.3304906686147054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,1536,0.13953510920206705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,1536,0.10370044575797187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,1536,0.06091644366582235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,5120,0.2792729006873237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,1024,0.09664977921379937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,1024,0.07933155695597331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,1024,0.04836888776885139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,4096,0.2285155455271403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,768,0.07596800062391493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,768,0.06730844577153523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,768,0.040948443942599826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,3584,0.20002667109171549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,512,0.05567466550403171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,512,0.05990044275919596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,512,0.03862133291032579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,3072,0.173071993721856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,256,0.03803199860784743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,256,0.05135200089878506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,256,0.03835377759403653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,2560,0.14710577328999838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,128,0.031009776724709406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,128,0.04654755526118808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,8192,128,0.038487109873029925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,64,0.025962667332755193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,64,0.04446311129464043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,8192,32,0.026517333255873785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,8192,32,0.04580710993872749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,2048,0.12107643816206191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,65536,6.011308458116319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,65536,4.658684624565972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,65536,2.1288782755533853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,1536,0.09702133470111424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,16384,1.35754574669732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,16384,0.9702097574869791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,16384,0.4439848793877496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,1024,0.07528977923923068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,12288,0.9975262747870551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,12288,0.7233875592549642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,12288,0.3237440056271023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,768,0.06544355551401775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,10240,0.8645386695861816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,10240,0.6697466638353137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,10240,0.27141067716810435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,512,0.05585777759552002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,8192,0.6957306861877441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,8192,0.4807510905795627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,8192,0.22164532873365614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,256,0.046663112110561795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,7168,0.5700773133171929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,7168,0.38387110498216415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,7168,0.1963520050048828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,8192,128,0.04042844308747186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,6144,0.49634043375651044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,6144,0.36212534374660915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,6144,0.17185244295332167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,65536,3.6935369703504772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,5120,0.44404978222317165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,5120,0.3044168949127197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,5120,0.14926044146219888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,16384,0.83660888671875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,4096,0.3301795588599311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,4096,0.24096179008483887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,4096,0.12127999464670818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,12288,0.591586642795139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,3584,0.28475112385219997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,3584,0.20219111442565918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,3584,0.10922666390736897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,10240,0.4870159890916612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,3072,0.2456728882259793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,3072,0.16308532820807561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,3072,0.09583466582828098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,8192,0.392303122414483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,2560,0.20688533782958984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,2560,0.15022310945722792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,2560,0.08296800322002836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,7168,0.34881244765387637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,2048,0.16962933540344238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,2048,0.1194559997982449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,2048,0.06994489166471693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,6144,0.29942043622334796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,1536,0.12808088461558023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,1536,0.09336621893776788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,1536,0.05554133322503832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,5120,0.25049156612820095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,1024,0.08823911348978679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,1024,0.06828977664311726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,1024,0.04391022192107307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,4096,0.20252088705698648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,768,0.07362133264541626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,768,0.060998221238454185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,768,0.037412444750467934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,3584,0.1781484418445163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,512,0.05694222450256348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,512,0.05233600073390537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,512,0.035106665558285184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,3072,0.15619999832577175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,256,0.04045866595374213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,256,0.045921779341167875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,256,0.03526666760444641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,2560,0.13200978438059488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,128,0.0299893319606781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,128,0.042472889026006065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,7168,128,0.03536622060669793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,64,0.02550044490231408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,64,0.04032622112168206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,7168,32,0.026633777552180823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,7168,32,0.041327999697791204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,2048,0.10973778035905625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,65536,4.83403566148546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,1536,0.08775111039479573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,65536,4.121746487087674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,65536,1.8141795264350042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,16384,1.197339587741428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,16384,0.9151146146986219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,16384,0.38737421565585667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,1024,0.06925777594248454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,12288,0.8096666865878634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,12288,0.6135208871629503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,12288,0.2824968761867947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,768,0.059826665454440646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,10240,0.6933315594991049
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,10240,0.5230177773369683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,10240,0.23748710420396593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,512,0.0516844458050198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,8192,0.5458399984571669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,8192,0.45015912585788304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,8192,0.19271111488342285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,256,0.04324266645643446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,7168,0.4775102403428819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,7168,0.3495448960198297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,7168,0.17124177349938285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,7168,128,0.037646220790015325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,6144,0.4146844546000163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,6144,0.2742435667249891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,6144,0.152547558148702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,65536,3.279500537448459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,5120,0.33840534422132706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,5120,0.2641804483201769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,5120,0.13197955820295545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,16384,0.7574328846401639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,4096,0.26217510965135365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,4096,0.18786933686998156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,4096,0.10758311218685573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,12288,0.5493777592976888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,3584,0.23606488439771864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,3584,0.16468800438774958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,3584,0.09701599677403767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,10240,0.4558417532179091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,3072,0.20010133584340414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,3072,0.14595199955834284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,3072,0.08529511425230239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,8192,0.36469777425130206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,2560,0.17200088500976562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,2560,0.12253244717915852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,2560,0.07414755556318495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,7168,0.319015105565389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,2048,0.13872799608442518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,2048,0.10076089037789239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,2048,0.061887111928727895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,6144,0.2747102313571506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,1536,0.10616799857881333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,1536,0.08037333356009589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,1536,0.04998133248753018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,5120,0.23134133550855848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,1024,0.07425866524378459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,1024,0.05974666939841377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,1024,0.039219554927613996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,4096,0.1867457760704888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,768,0.05730488565233019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,768,0.054321779145134814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,768,0.03354933195643955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,3584,0.16514489385816786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,512,0.04394222299257914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,512,0.046661333905326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,512,0.03227377931276957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,3072,0.1438008944193522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,256,0.030191110240088567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,256,0.04082311193148295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,256,0.031994667318132185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,2560,0.1229884492026435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,128,0.024313777685165405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,128,0.037754668129815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,6144,128,0.03234666585922241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,64,0.021029333273569744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,64,0.03604355454444885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,6144,32,0.021613332960340712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,6144,32,0.036815110180113055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,2048,0.10271288951237996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,65536,4.115105099148221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,65536,3.6435830858018665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,65536,1.4880871242947047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,1536,0.0828017791112264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,16384,0.9400089051988391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,16384,0.6783768865797254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,16384,0.3225991196102566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,1024,0.06438666582107544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,12288,0.6689288881089953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,12288,0.5134986771477593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,12288,0.24270311991373697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,768,0.05748533540301853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,10240,0.5863288773430718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,10240,0.4454711278279622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,10240,0.208478225602044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,512,0.04890399840142992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,8192,0.4665786425272624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,8192,0.3077084488338894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,8192,0.16862222883436415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,256,0.041548444165123835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,7168,0.40366400612725156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,7168,0.2988933457268609
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,7168,0.15061599678463405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,6144,128,0.03599288728502061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,6144,0.34430755509270566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,6144,0.25752798716227215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,6144,0.13454222679138184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,65536,2.8059618208143444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,5120,0.2761760022905138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,5120,0.21925689114464653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,5120,0.11687200599246556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,16384,0.6298195521036783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,4096,0.22012710571289062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,4096,0.1696231100294325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,4096,0.09489333629608154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,12288,0.4642373190985785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,3584,0.19429422749413383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,3584,0.14308799637688532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,3584,0.08489155769348145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,10240,0.3854737811618381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,3072,0.1707493331697252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,3072,0.12148266368442112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,3072,0.07579377624723646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,8192,0.30998312102423775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,2560,0.14459910657670763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,2560,0.10415644115871853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,2560,0.066392891936832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,7168,0.27022311422559947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,2048,0.11691110663943821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,2048,0.08637066682179768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,2048,0.055530667304992676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,6144,0.23356623119778105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,1536,0.08871822224722968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,1536,0.06904177533255683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,1536,0.045369777414533824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,5120,0.1963946686850654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,1024,0.06253688865237765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,1024,0.051402668158213295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,1024,0.03488889005449083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,4096,0.15982755025227866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,768,0.04823911190032959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,768,0.04751733276579115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,768,0.030011554559071858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,3584,0.14159733719295925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,512,0.03511022196875678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,512,0.03984533415900336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,512,0.028627554575602215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,3072,0.12377333641052246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,256,0.02472177810139126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,256,0.03512800070974562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,256,0.029167999823888142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,2560,0.10608000225490993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,128,0.021372444099850122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,128,0.03257066673702664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,5120,128,0.02979466650221083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,64,0.019685332973798115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,64,0.03086044390996297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,5120,32,0.020576889316240948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,5120,32,0.03192977772818671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,2048,0.08955466747283936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,65536,3.23907110426161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,65536,2.6011901431613498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,65536,1.211224873860677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,1536,0.0718088878525628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,16384,0.7309013472663032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,16384,0.539064884185791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,16384,0.27061955134073895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,1024,0.05801688962512546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,12288,0.5558213127983941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,12288,0.4146533277299669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,12288,0.20628621843126085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,768,0.050851556989881724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,10240,0.46901241938273114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,10240,0.3978959984249539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,10240,0.17859910594092476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,512,0.043059554364946157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,8192,0.37066043747795957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,8192,0.270980437596639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,8192,0.14827111032274035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,256,0.03611733184920417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,7168,0.3235199981265598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,7168,0.2239262262980143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,7168,0.1284897724787394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,5120,128,0.03214577833811442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,6144,0.27419911490546334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,6144,0.19455022282070586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,6144,0.1140257782406277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,65536,2.601531558566623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,5120,0.21607289049360487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,5120,0.17853600449032256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,5120,0.10304711262385051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,16384,0.5736453268263075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,4096,0.1762497822443644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,4096,0.138208892610338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,4096,0.08279911014768812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,12288,0.4262479941050212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,3584,0.15665688779619005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,3584,0.1182524495654636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,3584,0.0733671122127109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,10240,0.35366400082906085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,3072,0.1364648871951633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,3072,0.10151288906733195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,3072,0.06724622514512804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,8192,0.28268978330824107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,2560,0.11606489287482367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,2560,0.0874204436937968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,2560,0.05776888794369168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,7168,0.24850310219658744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,2048,0.09344177775912815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,2048,0.07254222366544935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,2048,0.048728889889187284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,6144,0.214227557182312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,1536,0.07105866405698988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,1536,0.058041777875688344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,1536,0.039354665411843195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,5120,0.1801288922627767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,1024,0.049066666099760264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,1024,0.043324444029066295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,1024,0.031344887283113264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,4096,0.14740888277689615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,768,0.03793066740036011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,768,0.040764444404178195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,768,0.026605332891146343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,3584,0.13168088595072427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,512,0.027710222535663184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,512,0.03369955552948846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,512,0.025381333298153345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,3072,0.11492444409264459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,256,0.01999200052685208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,256,0.029331554969151814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,256,0.026268444127506677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,2560,0.09970221916834514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,128,0.016563556260532804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,128,0.027943111128277246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,4096,128,0.026604443788528442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,64,0.014900444282425774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,64,0.026287110315428838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,4096,32,0.015086222026083203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,4096,32,0.026708443959554035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,2048,0.08264266782336764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,65536,3.0136390262179904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,65536,2.1476177639431424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,65536,1.0678746965196397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,1536,0.0673911107911004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,16384,0.7173022164238824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,16384,0.5029235415988499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,16384,0.24740089310540092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,1024,0.05312800076272753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,12288,0.5197235743204752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,12288,0.3407279915279812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,12288,0.18918132781982422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,768,0.04808800088034736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,10240,0.43491554260253906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,10240,0.3325795597500271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,10240,0.16862666606903076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,512,0.04016088777118259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,8192,0.3364400068918864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,8192,0.22787555058797201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,8192,0.13135022587246364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,256,0.03445600138770209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,7168,0.29683200518290204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,7168,0.22358666525946724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,7168,0.12353511651357015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,4096,128,0.030790223015679255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,6144,0.25034666061401367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,6144,0.17469599511888292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,6144,0.1060568888982137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,65536,2.16831292046441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,5120,0.2028799984190199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,5120,0.16006488270229763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,5120,0.09604355361726548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,16384,0.49948798285590273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,4096,0.16706311702728271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,4096,0.1192764441172282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,4096,0.07629333602057563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,12288,0.37259822421603733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,3584,0.14755554993947348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,3584,0.10479822423723008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,3584,0.06896622313393487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,10240,0.3095635573069255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,3072,0.12731911076439753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,3072,0.09143200185563828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,3072,0.06162133481767443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,8192,0.24828354517618814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,2560,0.10652177863650852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,2560,0.07742844687567817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,2560,0.053689777851104736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,7168,0.21757422553168404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,2048,0.0859617789586385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,2048,0.06389777527915107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,2048,0.04553600152333578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,6144,0.18906844986809623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,1536,0.06628888845443726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,1536,0.052240888277689614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,1536,0.036878221564822726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,5120,0.1592951085832384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,1024,0.04612355430920919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,1024,0.03907377852333917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,1024,0.028918223248587713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,4096,0.1310675541559855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,768,0.0362062222427792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,768,0.03791911072201199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,768,0.02515911062558492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,3584,0.11588088671366374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,512,0.026986666851573523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,512,0.030472887886895075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,512,0.023665777511066858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,3072,0.1022755569881863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,256,0.019670221540662978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,256,0.027531555957264368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,256,0.02500177754296197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,2560,0.08848889006508721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,128,0.01686844395266639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,128,0.02516533268822564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3584,128,0.025390222668647766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,64,0.01846222248342302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,64,0.024157333705160353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3584,32,0.01899111105336083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3584,32,0.02405244443151686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,2048,0.07484622134102716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,65536,2.3798035515679254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,65536,2.058912913004557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,65536,0.9821608861287435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,1536,0.060434665944841176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,16384,0.5424462424384223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,16384,0.4621724552578396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,16384,0.22353598806593153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,1024,0.04846133457289802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,12288,0.4131768809424506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,12288,0.30349598990546334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,12288,0.1726906696955363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,768,0.04396977689531115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,10240,0.3663511011335585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,10240,0.25207198990715873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,10240,0.15409421920776367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,512,0.03633155425389608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,8192,0.2940124405754937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,8192,0.20609777503543428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,8192,0.12017956044938828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,256,0.03094489044613308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,7168,0.2593093448215061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,7168,0.18245600329505074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,7168,0.10831733544667561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3584,128,0.02786488996611701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,6144,0.2182248830795288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,6144,0.15629243850708008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,6144,0.0992506676250034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,65536,1.9651627010769313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,5120,0.17659644285837808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,5120,0.14025244447920057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,5120,0.08884444501664902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,16384,0.4650408956739638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,4096,0.14469956027136907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,4096,0.10401778088675605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,4096,0.07056711117426555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,12288,0.3468160099453396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,3584,0.12683910793728298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,3584,0.0915786690182156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,3584,0.06363288561503093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,10240,0.29045067893134224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,3072,0.1099662250942654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,3072,0.08073955774307251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,3072,0.056913779841529004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,8192,0.2335564560360379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,2560,0.09163200192981297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,2560,0.06828622023264568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,2560,0.04996800091531542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,7168,0.20591199398040771
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,2048,0.07429422272576226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,2048,0.05849599838256836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,2048,0.0424542228380839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,6144,0.17702221870422363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,1536,0.05589866638183594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,1536,0.04667466547754076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,1536,0.033624887466430664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,5120,0.15069244967566595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,1024,0.038266665405697294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,1024,0.035418666071361966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,1024,0.026703110999531213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,4096,0.12431732813517253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,768,0.02928000026279026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,768,0.033653332127465144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,768,0.023016000787417095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,3584,0.11090133587519328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,512,0.02134399943881565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,512,0.027266666293144226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,512,0.021941334009170532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,3072,0.09825511111153497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,256,0.016217778126398723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,256,0.024417777856190998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,256,0.023629332582155865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,2560,0.08575733502705891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,128,0.01349066694577535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,128,0.02222844461599986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,3072,128,0.0236791107389662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,64,0.01239377756913503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,64,0.02144977781507704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,3072,32,0.012723555167516073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,3072,32,0.021735999319288466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,2048,0.0721013347307841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,65536,2.0581573910183377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,65536,1.5742346445719402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,1536,0.05993688768810696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,65536,0.8411528799268934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,16384,0.4460346433851454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,16384,0.38395200835333926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,16384,0.2219484382205539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,1024,0.045943111181259155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,12288,0.3474284542931451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,12288,0.28128088845147026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,768,0.041215999258889094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,12288,0.16467555363972983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,10240,0.3111964331732856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,10240,0.22307199902004668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,10240,0.1341973278257582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,512,0.03407911128467984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,8192,0.24762845039367676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,8192,0.18021066983540854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,8192,0.10827466514375474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,256,0.029465778006447688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,7168,0.21591911051008436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,7168,0.15744000011020237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,7168,0.09912000099817912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,3072,128,0.026156443688604567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,6144,0.17979021867116293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,6144,0.1352586613761054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,6144,0.08858399920993382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,65536,1.8890062967936199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,5120,0.15018133322397867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,5120,0.1184613307317098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,5120,0.0804471108648512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,16384,0.44723823335435653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,4096,0.12321066856384277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,4096,0.09158400032255386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,4096,0.06554755237367418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,12288,0.3347804546356201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,3584,0.10743644502427842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,3584,0.08005333609051175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,3584,0.05782666471269396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,10240,0.2784799999660916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,3072,0.09287733501858181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,3072,0.07005688879224989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,3072,0.051903999514049955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,8192,0.22321067916022408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,2560,0.07812533113691542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,2560,0.0601466695467631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,2560,0.04563377632035149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,7168,0.19829244083828398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,2048,0.062260442309909396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,2048,0.050233778026368886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,2048,0.03846222162246704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,6144,0.171832005182902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,1536,0.04718222220738729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,1536,0.04119911127620273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,1536,0.03056710958480835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,5120,0.14540533224741617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,1024,0.03219111098183526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,1024,0.031137777699364558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,1024,0.024320888850424025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,4096,0.12013778421613906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,768,0.024528000089857314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,768,0.02992266747686598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,768,0.02123822271823883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,3584,0.10769510931438869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,512,0.018304889400800068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,512,0.0243520008193122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,512,0.020592888196309406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,3072,0.09414666891098022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,256,0.013764444324705334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,256,0.021320889393488567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,256,0.021817778547604878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,2560,0.08173155784606934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,128,0.012095111111799875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,128,0.019719999697473314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2560,128,0.02199199961291419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,64,0.01036177741156684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,64,0.018923555811246235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2560,32,0.010656000011497073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2560,32,0.019270223047998216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,2048,0.0684551133049859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,65536,1.5207431581285265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,65536,1.3647724787394206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,65536,0.7247449027167426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,1536,0.05583466423882378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,16384,0.3656186792585585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,16384,0.31924978892008465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,16384,0.20088444815741646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,1024,0.04455022348297966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,12288,0.2758328914642334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,12288,0.22761334313286674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,12288,0.13597955968644884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,768,0.04164088765780131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,10240,0.24426666895548502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,10240,0.1981502241558499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,10240,0.12245421939425999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,512,0.03354310989379883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,8192,0.19826132721371123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,8192,0.1550284491644965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,8192,0.09746044211917454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,256,0.02917511264483134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,7168,0.17249777581956652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,7168,0.13524177339341906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,7168,0.09021955728530884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2560,128,0.02641777859793769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,6144,0.14725866582658556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,6144,0.1156346665488349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,6144,0.08113955789142184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,65536,1.6609617869059246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,5120,0.12007110648685032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,5120,0.10389333301120335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,5120,0.07586755355199178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,16384,0.3892506758371989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,4096,0.0990648865699768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,4096,0.07816711399290296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,4096,0.05893066856596205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,12288,0.2945715586344401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,3584,0.08657511075337727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,3584,0.06822577450010511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,3584,0.05263911022080315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,10240,0.24364089965820312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,3072,0.07442222038904826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,3072,0.05950044261084663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,3072,0.04706844356324938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,8192,0.1979315545823839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,2560,0.06192621919843885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,2560,0.05117600162823995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,2560,0.041702220837275185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,7168,0.17557156085968018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,2048,0.05057155423694187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,2048,0.04333600070741442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,2048,0.035255998373031616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,6144,0.15309154987335205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,1536,0.03812088900142246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,1536,0.03638311227162679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,1536,0.028975112570656672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,5120,0.12965244717068142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,1024,0.026320889592170715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,1024,0.027375110321574744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,1024,0.022290666898091633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,4096,0.1071466671095954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,768,0.020850666695170935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,768,0.02660977840423584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,768,0.019329778022236295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,3584,0.09523377815882365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,512,0.015631111131774057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,512,0.020809777908855014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,512,0.01887022289964888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,3072,0.0851386653052436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,256,0.012607999973826937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,256,0.018458666072951425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,256,0.020669332808918424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,2560,0.07404444615046184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,128,0.010497777826256223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,128,0.01699022286468082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,2048,128,0.020429333051045735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,64,0.010012444522645738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,64,0.01651822196112739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,2048,32,0.010291555689440833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,2048,32,0.016543999314308167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,2048,0.06284266710281372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,65536,1.1034115685356987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,65536,1.1265839470757377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,65536,0.6518017980787489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,1536,0.05170489019817776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,16384,0.28498488002353245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,16384,0.26956354247199166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,16384,0.1760631137424045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,1024,0.04084711273511251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,12288,0.2147679991192288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,12288,0.1903608904944526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,12288,0.12264266279008652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,768,0.03782400157716539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,10240,0.18808889389038086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,10240,0.1590071121851603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,10240,0.1076622207959493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,512,0.029800001117918227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,8192,0.15155111418830022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,8192,0.1290124389860365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,8192,0.08779022428724502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,256,0.026268444127506677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,7168,0.13405688603719076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,7168,0.11329332987467448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,7168,0.0801795588599311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,2048,128,0.022727999422285292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,6144,0.11361332734425862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,6144,0.09621422158347236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,6144,0.07356622483995226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,65536,1.4078381856282551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,5120,0.0935208863682217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,5120,0.08071911334991455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,5120,0.06781333022647433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,16384,0.3515306578742133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,4096,0.07516088750627306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,4096,0.06491466363271077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,4096,0.05388977792527941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,12288,0.266698678334554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,3584,0.06628177563349406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,3584,0.05747999747594198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,3584,0.04870577653249105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,10240,0.22326132986280653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,3072,0.05725066529379951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,3072,0.05104355679617988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,3072,0.04390844371583727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,8192,0.18064799573686388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,2560,0.047016001409954496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,2560,0.042973332934909396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,2560,0.037779556380377874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,7168,0.15992355346679688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,2048,0.03856177793608771
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,2048,0.036679112248950534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,2048,0.03177600105603536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,6144,0.13808622625139025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,1536,0.029469334416919287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,1536,0.03030222323205736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,1536,0.02629511058330536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,5120,0.11873688962724473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,1024,0.020618667205174763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,1024,0.023235556152131822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,1024,0.021003555920388963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,4096,0.09805422359042698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,768,0.016709332664807636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,768,0.022666666242811415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,768,0.017916444275114272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,3584,0.08852799733479817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,512,0.01297688898113039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,512,0.017519111434618633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,512,0.017498667041460674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,3072,0.07869422435760498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,256,0.011061333119869232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,256,0.01511111193233066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,256,0.01920977731545766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,2560,0.06829866435792711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,128,0.009384888741705153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,128,0.0148044443792767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1536,128,0.01940355532699161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,64,0.009986666341622671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,64,0.01442577772670322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1536,32,0.00865955568022198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1536,32,0.01443199978934394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,2048,0.05820000171661377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,65536,0.7408249113294813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,65536,0.8913937674628364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,65536,0.5145049095153809
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,1536,0.047989332013660006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,16384,0.18952266375223795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,16384,0.2040853367911445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,16384,0.13652711444430882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,1024,0.03750755720668369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,12288,0.14502488242255315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,12288,0.15457599692874485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,12288,0.10548444588979085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,768,0.03490044342146979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,10240,0.12461154990726048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,10240,0.1280871099895901
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,10240,0.09376977549658881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,512,0.027959111664030287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,8192,0.099945776992374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,8192,0.1031564474105835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,8192,0.07612355550130208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,256,0.025056888659795124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,7168,0.08722488747702704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,7168,0.08937866820229425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,7168,0.06982666916317411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1536,128,0.02264444364441766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,6144,0.07622933387756348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,6144,0.07715111308627658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,6144,0.06497866577572294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,65536,1.2812844382392037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,5120,0.06217066446940104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,5120,0.06754755311542086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,5120,0.061153776115841336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,16384,0.31610221332973903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,4096,0.05084355672200521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,4096,0.05281688769658407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,4096,0.047352890173594155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,12288,0.2408817874060737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,3584,0.045019557078679405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,3584,0.04691466689109802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,3584,0.04279999931653341
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,10240,0.20034755600823295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,3072,0.039240890079074435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,3072,0.04022044605678982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,3072,0.03809600075085958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,8192,0.1656631098853217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,2560,0.0325271122985416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,2560,0.03523111012246873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,2560,0.03336266676584879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,7168,0.14599733882480198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,2048,0.02642044425010681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,2048,0.029765334394243028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,2048,0.02832533253563775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,6144,0.12766488393147787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,1536,0.0206666671567493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,1536,0.02477866742346022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,1536,0.023601777023739282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,5120,0.10889599720637004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,1024,0.015565334094895257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,1024,0.018962666392326355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,1024,0.018431110514534842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,4096,0.09046222103966607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,768,0.013262222210566202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,768,0.01919644408755832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,768,0.015816888875431485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,3584,0.08105244239171346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,512,0.009927111367384592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,512,0.01369599997997284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,512,0.015487111277050443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,3072,0.07176711161931355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,256,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,256,0.012200000385443369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,256,0.01738755570517646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,2560,0.06186311112509834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,128,0.00686488880051507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,128,0.01204355557759603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,1024,128,0.01737422247727712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,64,0.007261333366235097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,64,0.011677333050303988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,1024,32,0.007325333025720384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,1024,32,0.011771555576059552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,2048,0.05352355705367195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,65536,0.5633449024624294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,65536,0.8844915495978461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,65536,0.5960631370544434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,1536,0.04479555620087517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,16384,0.14793511231740317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,16384,0.18056621816423205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,16384,0.13419200314415827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,1024,0.03552444444762336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,12288,0.11114754941728379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,12288,0.13583821720547146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,12288,0.10468710793389215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,768,0.03319199879964193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,10240,0.10114933384789361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,10240,0.1141973336537679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,10240,0.09326399697197808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,512,0.025813332862324182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,8192,0.08094133271111383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,8192,0.09267644749747382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,8192,0.0761715571085612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,256,0.02404444416364034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,7168,0.07082221905390422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,7168,0.07959555255042182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,7168,0.06920178069008721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,65536,1.0983111063639324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,6144,0.061221334669325084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,6144,0.06907200151019625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,6144,0.06446399953630236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,1024,128,0.02070399953259362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,5120,0.05140533381038242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,5120,0.05745066536797417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,5120,0.059677335951063365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,16384,0.28751111030578613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,4096,0.04112799962361654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,4096,0.04719110992219713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,4096,0.04689155684577095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,12288,0.2184142271677653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,3584,0.036008887820773654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,3584,0.04065955678621928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,3584,0.042282667424943715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,10240,0.18429956171247694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,3072,0.031062222189373438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,3072,0.0358586675590939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,3072,0.03769066598680284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,8192,0.15112800068325466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,2560,0.0266302227973938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,2560,0.030571556753582422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,2560,0.03312799996799893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,7168,0.13368799951341417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,2048,0.02198133369286855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,2048,0.026663111315833196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,2048,0.028288000159793433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,6144,0.11603111690945095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,1536,0.017454221844673157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,1536,0.02200177808602651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,1536,0.023014222582181294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,5120,0.09788978099822998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,1024,0.01369066701995002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,1024,0.017107556263605755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,1024,0.01812622282240126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,4096,0.08269421921836005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,768,0.011050666371981302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,768,0.017351999878883362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,768,0.015499555402331881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,3584,0.07457066906823052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,512,0.008289777570300633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,512,0.012090666426552666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,512,0.015439111325475903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,3072,0.06592355834113227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,256,0.006573333508438534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,256,0.011053333679835001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,256,0.01715022159947289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,2560,0.05796533160739475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,128,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,128,0.010699555277824402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,768,128,0.017312889297803242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,64,0.007931555310885111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,64,0.010678222609890832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,768,32,0.005220444252093633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,768,32,0.010666666759385003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,2048,0.04968444506327311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,65536,0.41528977288140195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,65536,0.6856151156955295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,65536,0.49701065487331814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,1536,0.04128088884883457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,16384,0.1062248879008823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,16384,0.16034221649169922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,16384,0.11242577764723037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,1024,0.03281689021322463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,12288,0.07994755771425036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,12288,0.12103733751508926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,12288,0.08816888597276475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,768,0.030710223648283217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,10240,0.06798577970928617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,10240,0.10139377911885579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,10240,0.07965777979956733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,512,0.024401777320437964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,8192,0.05423022309939066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,8192,0.08197866545783149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,8192,0.06441422303517659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,256,0.021611554755104914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,7168,0.047452443175845675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,7168,0.07054844167497423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,7168,0.05992533100975884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,768,128,0.01958755486541324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,6144,0.04085066583421495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,6144,0.06048711140950521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,6144,0.05594222413169014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,65536,1.0399235619439018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,5120,0.03409511182043288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,5120,0.05009066727426317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,5120,0.0525493323802948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,16384,0.27376000086466473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,4096,0.02812888887193468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,4096,0.040384001202053495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,4096,0.04079466727044847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,12288,0.2081377771165636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,3584,0.024887111451890733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,3584,0.03602933221393161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,3584,0.03704444567362467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,10240,0.17506844467586943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,3072,0.021549334128697712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,3072,0.030966222286224365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,3072,0.032460444503360324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,8192,0.1416391134262085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,2560,0.018897778458065458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,2560,0.027052443888452318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,2560,0.028493334849675495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,7168,0.12605867120954725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,2048,0.015964445140626695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,2048,0.023380445109473333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,2048,0.024303999212053087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,6144,0.10983110798729791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,1536,0.013265777793195514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,1536,0.019254222512245178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,1536,0.020006222857369315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,5120,0.09330577982796563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,1024,0.01071911139620675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,1024,0.014696000350846184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,1024,0.015609777635998197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,4096,0.07774488793479072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,768,0.008906666603353288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,768,0.015286222100257874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,768,0.013521778086821238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,3584,0.06977066728803846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,512,0.007260444263617198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,512,0.011032889286677042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,512,0.013501333693663279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,3072,0.06167733007007175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,256,0.005542222410440445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,256,0.009701333112186855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,256,0.0151573336786694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,2560,0.054139554500579834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,128,0.004936000125275718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,128,0.009357333183288574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,512,128,0.015101333459218344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,64,0.004517333375083076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,64,0.009344889058007134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,512,32,0.004520888957712385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,512,32,0.009304000271691216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,2048,0.046253333489100136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,65536,0.2479768859015571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,65536,0.5594595803154839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,65536,0.4351129002041287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,1536,0.03872888949182298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,16384,0.06287022431691487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,16384,0.1421297788619995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,16384,0.11158400111728245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,1024,0.030713776747385662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,12288,0.047687109973695546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,12288,0.10717422432369655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,12288,0.08775111039479573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,768,0.02947288751602173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,10240,0.039819555150138006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,10240,0.08968533409966363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,10240,0.0791155563460456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,512,0.022799111074871484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,8192,0.03245866629812453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,8192,0.07186933358510335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,8192,0.06406666835149129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,256,0.020232000284724765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,7168,0.0289066665702396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,7168,0.06127733654446072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,7168,0.059495111306508384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,65536,0.9405706193712023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,6144,0.025009777810838487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,6144,0.05231111248334249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,6144,0.055604444609748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,512,128,0.016152888536453247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,5120,0.021485333641370136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,5120,0.04296711087226868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,5120,0.05225777626037598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,16384,0.24594399664137098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,4096,0.017831999394628737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,4096,0.0346666673819224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,4096,0.04055022199948629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,12288,0.19004800584581164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,3584,0.01571111049917009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,3584,0.030326220724317763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,3584,0.03657333387268914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,10240,0.16060533788469103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,3072,0.0140337778462304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,3072,0.026535110341178045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,3072,0.032702220810784235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,8192,0.1305102242363824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,2560,0.01202488856183158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,2560,0.02345866627163357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,2560,0.028805332051383123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,7168,0.11579110887315537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,2048,0.010795555180973478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,2048,0.0199306673473782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,2048,0.0244862222009235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,6144,0.10124799940321182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,1536,0.009131555755933126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,1536,0.016768889294730294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,1536,0.020212444994184706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,5120,0.08611377742555405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,1024,0.007609777980380588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,1024,0.013047110703256396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,1024,0.015795555379655626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,4096,0.0712640020582411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,768,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,768,0.013737777868906656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,768,0.013430222041077085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,3584,0.06305955515967475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,512,0.004869333157936732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,512,0.009311999711725447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,512,0.012506666282812754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,3072,0.056827558411492236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,256,0.0041848888827694785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,256,0.008303110798199972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,256,0.014839111102951897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,2560,0.04982666505707634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,128,0.0038337777886125776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,128,0.008044444852405125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,256,128,0.015094222293959724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,64,0.003547555456558863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,64,0.008282666405042013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,256,32,0.003658666792843077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,256,32,0.008216888540320927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,2048,0.04337244563632541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,65536,0.18309333589341906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,65536,0.5502444373236762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,65536,0.39757688840230304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,1536,0.03603911068704393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,16384,0.046562665038638644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,16384,0.14162577523125544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,16384,0.11231822437710232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,1024,0.028896888097127277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,12288,0.037138667371537946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,12288,0.10559288660685222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,12288,0.08859733078214858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,768,0.027094221777386133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,10240,0.030637333790461223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,10240,0.08832266595628525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,10240,0.07965866724650066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,512,0.01956888867749108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,8192,0.02573066618707445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,8192,0.07094044155544706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,8192,0.06480889187918769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,256,0.016999999682108562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,7168,0.022410665949185688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,7168,0.06049333016077677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,7168,0.06084711021847195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,65536,0.9340746137830945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,6144,0.019659555620617337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,6144,0.051663110653559365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,6144,0.05583022038141886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,256,128,0.014176888598336114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,5120,0.017051556044154696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,5120,0.04251377781232198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,5120,0.052468445565965444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,16384,0.2445280022091336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,4096,0.0148053334818946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,4096,0.033972445461485125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,4096,0.04052266809675429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,12288,0.18662222226460776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,3584,0.013000888956917657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,3584,0.029608001311620075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,3584,0.03701333204905192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,10240,0.1572764449649387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,3072,0.011744889120260874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,3072,0.026034666432274714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,3072,0.03300889001952277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,8192,0.1284115579393175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,2560,0.010289777484205034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,2560,0.023344889283180237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,2560,0.02863911125395033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,7168,0.11354755030737983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,2048,0.009002666506502364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,2048,0.020188444190555148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,2048,0.024716445141368445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,6144,0.0976488921377394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,1536,0.0075786668393347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,1536,0.016571554872724745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,1536,0.020239111449983384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,5120,0.08401155471801758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,1024,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,1024,0.013064000341627332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,1024,0.016116445263226826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,4096,0.07022755675845675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,768,0.004875555634498596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,768,0.014295111099878946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,768,0.013423110875818463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,3584,0.0632915563053555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,512,0.0038862224254343244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,512,0.009372444616423713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,512,0.012400888734393649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,3072,0.05584444602330526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,256,0.0035048888789282907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,256,0.008283555507659912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,256,0.015018666783968607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,2560,0.04911466770701938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,128,0.0031991110493739447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,128,0.008206222620275285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8192,128,128,0.015117333994971381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,64,0.0031804444475306403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,64,0.007975999679830339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,128,32,0.0032035555276605818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,128,32,0.00797333319981893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,65536,0.17148799366421172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,65536,0.5423297882080078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,16384,0.05004444387223986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,16384,0.14084533850351968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,12288,0.03523644473817613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,12288,0.10532355308532715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,10240,0.02936799989806281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,10240,0.08748088942633735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,8192,0.024780443973011438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,8192,0.0703164471520318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,7168,0.02166844407717387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,7168,0.059960001044803195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,6144,0.019194665882322524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,6144,0.0513573322031233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,5120,0.016799110505315993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,5120,0.04183822207980686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,4096,0.01404800017674764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,4096,0.033985777033699885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,3584,0.012423111332787408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,3584,0.02949777907795376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,3072,0.011238222320874533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,3072,0.025842666625976562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,2560,0.010390222072601318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,2560,0.023001778456899855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,2048,0.008939555949634975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,2048,0.019606222709019978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,1536,0.007790221936172909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,1536,0.01958666741847992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,1024,0.005877333382765452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,1024,0.014808000789748298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,768,0.00453244439429707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,768,0.014862222803963555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,512,0.0038764443662431505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,512,0.009331555830107795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,256,0.003510222252872255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,256,0.008608000146018134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,128,0.0031271111220121384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,128,0.007972444097201029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,64,0.0028693332440323303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,64,0.007977777885066139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,64,32,0.003152888889114062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,64,32,0.007937777373525832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,65536,0.1707377831141154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,65536,0.5427315500047472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,16384,0.04685066805945503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,16384,0.14097332954406738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,12288,0.03631110986073812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,12288,0.10543555683559841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,10240,0.02789777848455641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,10240,0.0870328876707289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,8192,0.02363377809524536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,8192,0.06989777750439115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,7168,0.021003555920388963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,7168,0.05990755558013916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,6144,0.01830044388771057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,6144,0.051198220915264554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,5120,0.015864888827006023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,5120,0.04214222232500712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,4096,0.013561777770519257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,4096,0.03355377912521362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,3584,0.012236444486512078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,3584,0.03673511081271701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,3072,0.010996444357766045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,3072,0.03183022141456604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,2560,0.00960977789428499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,2560,0.02847466535038418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,2048,0.008275555239783393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,2048,0.023696889479955036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,1536,0.007632888853549957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,1536,0.019447111421161227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,1024,0.005919111271699269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,1024,0.014783110883500842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,768,0.004508444418509801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,768,0.01479111115137736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,512,0.0038657776183552216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,512,0.009329777624871995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,256,0.003257777748836411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,256,0.008492444124486711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,128,0.002862222285734283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,128,0.00796888851457172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,64,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,64,0.00796622203456031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8192,32,32,0.0031724443866146933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8192,32,32,0.007928888830873702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,16384,3.2010169559054904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,16384,6.5480906168619795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,16384,3.9946267869737415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,2048,0.042500442928738065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,16384,1.5915706422593858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,12288,4.763848198784722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,12288,2.8762071397569446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,12288,2.2887946234809027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,1536,0.03525333272086249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,10240,1.842624028523763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,10240,4.151568942599826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,10240,2.285698572794596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,12288,1.2630533642239041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,8192,3.187898635864258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,1024,0.02776266634464264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,8192,1.790546629163954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,7168,1.3105075624254015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,7168,3.1214105818006725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,7168,1.5878275765313044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,8192,1.471409797668457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,10240,0.9905671013726128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,6144,2.5253092447916665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,768,0.024159999357329473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,6144,1.3202453189425998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,8192,0.8265920215182834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,512,0.01886666648917728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,5120,1.1160169177585177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,5120,1.9265386793348525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,6144,1.130192862616645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,4096,1.4842400021023219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,4096,0.9590586556328667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,7168,0.7265369097391764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,3584,0.6675653457641602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,256,0.01591022147072686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,3584,1.209323565165202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,3584,0.7607279883490669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,5120,0.9062035878499349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,3072,1.0416249169243705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,3072,0.8372702068752713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,4096,0.754626644982232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,6144,0.6299297544691298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,2560,0.8641573588053385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,2560,0.6750177807278104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,5120,0.542151133219401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,3072,0.5677137904696995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,2048,0.6998453140258789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,2048,0.6255920198228624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,4096,0.4442675643497043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,3584,0.39871822463141543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,2560,0.4873973528544108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,1536,0.5354417694939507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,1536,0.42995288636949325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,1024,0.2945173316531711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,1024,0.3749528990851508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,1024,0.28869422276814777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,3072,0.3489644527435303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,2048,0.40330134497748477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,768,0.28241244951883954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,768,0.2783973217010498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,2560,0.3036373456319173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,1536,0.3236791027916802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,512,0.22108356157938638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,512,0.2574640115102132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,768,0.19007021851009795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,2048,0.2554711235894097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,256,0.15148977438608804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,256,0.20308266745673287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,1536,0.20644711123572457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,1024,0.1628053320778741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,512,0.1763377719455295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,128,0.11043910847769843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,128,0.18172800540924072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,64,0.1020488871468438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,64,0.17743910683525932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,65536,32,0.09989155663384332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,65536,32,0.18005777729882133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,768,0.14090043968624538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,256,0.13044267230563694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,512,0.12570756011539036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,65536,128,0.1349440018335978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,65536,4.87153583102756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,65536,6.323292626274957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8192,128,128,0.014087999860445658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,16384,0.9007999632093641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,16384,1.354729758368598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,12288,1.222472932603624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,256,0.10990933577219646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,12288,0.7774631182352701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,65536,1.9508622487386067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,16384,0.42439821031358504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,10240,0.9563937717013888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,10240,0.5587209065755209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,65536,3.884996626112196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,12288,0.3225342167748345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,8192,0.7514444457160102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,65536,128,0.10887910922368367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,8192,0.42676978641086155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,7168,0.6085591316223145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,7168,0.4447129037645128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,16384,0.7910479969448514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,10240,0.2719759941101074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,6144,0.5317386521233453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,6144,0.3531626595391168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,12288,0.5840649074978298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,8192,0.22210666868421766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,5120,0.43169599109225804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,5120,0.3019333415561252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,10240,0.4873395495944553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,7168,0.19837866889105904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,6144,0.17342755529615614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,4096,0.3553626537322998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,8192,0.3923119968838162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,4096,0.22512711418999565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,3584,0.3118355539109972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,3584,0.202820446756151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,5120,0.14964621596866184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,7168,0.35079465972052676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,3072,0.27089065975613064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,3072,0.19257599777645537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,6144,0.30245688226487905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,4096,0.12224888801574707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,2560,0.2246168984307183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,2560,0.14993422561221653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,3584,0.10940000083711411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,2048,0.11688266860114203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,5120,0.2549368805355496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,2048,0.18347644805908203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,2048,0.12688622209760878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,3072,0.09722933504316543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,1536,0.1418186691072252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,1536,0.1025200022591485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,4096,0.2076124482684665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,2560,0.08368266953362359
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,1024,0.09700089030795628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,1024,0.07392799854278564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,2048,0.07060977485444811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,3584,0.18437777625189888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,768,0.07573244306776258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,768,0.06708444489373101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,1536,0.05749333567089505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,3072,0.16256977452172175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,512,0.055769774648878306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,512,0.06019466453128391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,2560,0.13999022377861872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,1024,0.0466213325659434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,256,0.03808622227774726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,256,0.05371733175383674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,768,0.04200977749294705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,512,0.039719998836517334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,128,0.03143289022975498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,128,0.05110133356518216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,1536,0.09534577528635661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,64,0.02596355477968852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,64,0.04955910974078708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,16384,32,0.026510222090615168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,16384,32,0.049908445941077344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,256,0.035411556561787925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,1024,0.07738488912582397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,16384,128,0.03523111012246873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,768,0.06254133250978258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,512,0.057088891665140785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,65536,3.8256437513563366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,16384,1.1793973710801866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,65536,4.8682810465494795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,256,0.04532088836034139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,16384,0.8475644323560926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,12288,0.8809884389241537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,12288,0.5767928759256998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,16384,128,0.0399217771159278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,65536,1.5243138207329645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,10240,0.7004462348090278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,10240,0.44549422793918186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,16384,0.3344115681118435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,8192,0.5385795699225532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,12288,0.2511395613352458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,8192,0.4126879904005263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,10240,0.21135555373297799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,7168,0.4810098012288411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,7168,0.29190044932895237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,65536,2.8208185831705728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,8192,0.17280711068047416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,6144,0.3044026692708333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,6144,0.41292444864908856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,6144,0.2733955648210314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,16384,0.6224835713704427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,5120,0.3248506651984321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,12288,0.45495022667778867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,5120,0.22943645053439668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,4096,0.17926222748226592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,7168,0.15370933214823404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,4096,0.17536355389489067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,4096,0.2658897770775689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,6144,0.13553422027164036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,3584,0.24038132031758627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,3584,0.15554932753245035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,10240,0.3805110984378391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,8192,0.30776800049675834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,3072,0.20498400264316133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,5120,0.11606044239468044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,3072,0.13656532764434814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,4096,0.0953635573387146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,2560,0.17336889108022055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,2560,0.11618844668070476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,7168,0.2742462158203125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,3584,0.08603288729985555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,2048,0.14135911729600695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,2048,0.09707022375530666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,3072,0.0747706691424052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,2560,0.06578666634029813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,1536,0.10792266660266453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,1536,0.07744889126883613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,5120,0.19993689325120714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,2048,0.05536977781189812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,1024,0.07391644186443753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,1024,0.05731733640034994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,1536,0.045078221294615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,3584,0.1454755597644382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,768,0.057738668388790555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,768,0.05310222175386217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,3072,0.12740443812476263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,1024,0.03752711084153917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,2560,0.10998488797081842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,512,0.0425600012143453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,512,0.04736888739797804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,2048,0.09268711010615031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,256,0.02977688776122199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,256,0.042466666963365346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,768,0.034240000777774386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,512,0.032256000571780734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,128,0.024127999941507976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,128,0.04000355468855964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,1536,0.0767315559917026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,64,0.0211191111140781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,64,0.03892799880769517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,12288,32,0.021470222208235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,12288,32,0.039048890272776283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,256,0.029151111841201782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,1024,0.06150844362046984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,65536,2.3188915252685547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,12288,128,0.02870933214823405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,768,0.051757332351472646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,65536,4.0913717481825085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,65536,3.1450311872694225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,16384,0.9690959718492297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,16384,0.6010222434997559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,12288,0.709769778781467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,512,0.04753066764937507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,256,0.03889688849449158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,12288,0.5024746788872613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,65536,1.3102551566229927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,10240,0.5589057604471842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,10240,0.3817031118604872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,16384,0.28545599513583714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,12288,128,0.034210665358437434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,8192,0.4505297872755263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,12288,0.21443466345469156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,8192,0.28227734565734863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,7168,0.39669866032070583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,7168,0.2776702245076497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,10240,0.184096892674764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,8192,0.14851644304063585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,6144,0.33985066413879395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,6144,0.2109982172648112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,7168,0.1319297817018297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,16384,0.5259537696838379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,5120,0.27013156149122447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,5120,0.19359644254048666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,6144,0.11677422788408066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,12288,0.3918035560184055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,4096,0.22483732965257433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,4096,0.16782755321926543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,10240,0.3282159964243571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,5120,0.09973422023985122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,3584,0.19807467195722792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,3584,0.13054666254255506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,8192,0.266595549053616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,4096,0.08228888776567247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,3072,0.171780440542433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,3072,0.11461510923173691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,7168,0.23367288377549914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,3584,0.07337599992752075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,2560,0.14374221695794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,6144,0.20524888568454316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,2560,0.0977226694424947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,3072,0.06516266531414457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,2048,0.11717600292629665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,2048,0.08159377839830187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,2560,0.056870222091674805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,5120,0.17170043786366782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,1536,0.08978755606545342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,1536,0.06600266695022583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,2048,0.048115554783079356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,4096,0.14189778433905706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,1024,0.06212177541520861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,1024,0.049288001325395375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,3584,0.12609955999586317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,1536,0.03955644369125366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,768,0.04823288983768887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,768,0.04598044355710348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,3072,0.11091822385787964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,512,0.04262666569815742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,512,0.03566933340496487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,512,0.04013777772585551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,1024,0.03254755669169956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,768,0.030415111117892798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,256,0.02478933334350586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,256,0.03631110986073812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,512,0.02861422300338745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,2048,0.08097600274615817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,2560,0.09512711233562893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,128,0.022788445154825848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,128,0.03451022174623277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,64,0.01941422290272183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,64,0.033223999871148005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,10240,32,0.020337777005301583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,10240,32,0.03347200155258179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,256,0.026072000463803608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,1536,0.06698044141133626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,10240,128,0.025635555386543274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,1024,0.05417955583996243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,65536,2.277338663736979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,65536,3.2076560126410594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,768,0.047462221648958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,16384,0.924680921766493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,16384,0.49507644441392684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,65536,1.1540497673882377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,12288,0.5462391111585829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,12288,0.4388773441314697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,10240,0.3561582300398085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,10240,0.44053154521518284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,16384,0.25871645079718697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,10240,0.2981493208143446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,256,0.03547822104560004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,12288,0.18725777996910942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,8192,0.3689102331797282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,8192,0.2318169011010064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,7168,0.2082773314581977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,10240,128,0.03139200144343906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,7168,0.31636354658338756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,7168,0.20185066594017875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,10240,0.15551555156707764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,6144,0.27213867505391437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,6144,0.1909253332349989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,8192,0.12531733512878418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,7168,0.11204889085557725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,5120,0.22601689232720268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,5120,0.15356443987952337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,6144,0.09802933533986409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,65536,1.8795856899685328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,4096,0.17740355597601998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,4096,0.13127556112077501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,5120,0.08520444234212239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,16384,0.43966044320000547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,3584,0.15810044606526694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,3584,0.10722933212916057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,12288,0.33231645160251194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,4096,0.06988800234264798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,3072,0.13611643844180638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,3072,0.09434755643208821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,3584,0.06220177809397379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,8192,0.22270756297641328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,2560,0.11486666732364231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,3072,0.05546844336721632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,2560,0.08033866352505155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,6144,0.17008088694678414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,2048,0.0926177766587999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,2048,0.0665644473499722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,2560,0.04862666792339749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,5120,0.14526577790578207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,1536,0.07145422034793429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,1536,0.05457511213090685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,2048,0.041492445601357356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,4096,0.11922755506303574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,1024,0.049294223388036094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,1024,0.0424542228380839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,1536,0.03354310989379883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,3584,0.10638844304614598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,768,0.03818577859136794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,768,0.03868800070550706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,1024,0.028059555424584284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,3072,0.09353778097364639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,512,0.027828445037206013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,512,0.03400355577468872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,768,0.025221332907676697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,2560,0.08169600036409166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,256,0.01999022232161628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,256,0.02963199880388048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,512,0.02369244396686554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,2048,0.07003288798862033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,128,0.016531555189026725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,128,0.02847822176085578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,256,0.022097777989175584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,64,0.014767999450365702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,64,0.027323555615213182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,8192,32,0.015204444527626038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,8192,32,0.026800000005298193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,1536,0.05819466378953722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,8192,128,0.022346667117542688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,1024,0.047648890150917896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,65536,2.118978712293837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,768,0.04199555516242981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,65536,3.1205919053819446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,512,0.03789866632885403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,65536,0.9906773037380643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,16384,0.481278207567003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,16384,0.6972373326619467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,12288,0.3310657872094048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,12288,0.5564177831013998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,256,0.032320890161726214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,16384,0.22612977027893066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,10240,0.4208639992607965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,12288,0.1662773291269938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,10240,0.2605200078752306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,8192,0.227384885152181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,8192,0.32295200559828013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,8192,128,0.0284222231970893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,8192,0.2023022174835205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,7168,0.28813600540161133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,10240,0.1399573352601793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,7168,0.1807031101650662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,6144,0.15490221977233887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,6144,0.24253421359592012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,6144,0.15471911430358887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,8192,0.1146426730685764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,7168,0.10095200273725723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,5120,0.20593955781724718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,5120,0.13321510950724283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,65536,1.743459595574273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,6144,0.08914044168260361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,4096,0.1648906601799859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,4096,0.10862933264838324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,5120,0.07744355334175958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,4096,0.06405155526267158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,16384,0.3979857762654622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,3584,0.14474666118621826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,12288,0.2961120075649685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,3584,0.10410310824712117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,3072,0.121998217370775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,3072,0.08435289065043132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,10240,0.24964711401197645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,3584,0.05746044715245565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,2560,0.10441066821416219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,2560,0.07206755876541138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,3072,0.050986667474110924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,7168,0.1782604455947876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,2048,0.08437244759665595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,2048,0.06001422140333387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,2560,0.04434577624003092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,5120,0.13140800264146593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,2048,0.03869955407248603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,1536,0.06525599956512451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,1536,0.04978933268123203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,1536,0.030756443738937378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,4096,0.10736888647079468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,1024,0.046069333950678505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,1024,0.03750666644838121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,1024,0.025455999705526564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,768,0.03728977839152018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,3584,0.09658755858739217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,768,0.03552444444762336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,3072,0.08559555477566189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,512,0.02825777729352315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,512,0.030400888787375555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,2560,0.07523377736409505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,768,0.023698666029506262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,256,0.02196177840232849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,256,0.026632888449562922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,512,0.022101332743962605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,2048,0.06440266635682848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,128,0.017483555608325534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,128,0.024653333756658766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,256,0.020971554848882887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,64,0.01462399959564209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,64,0.023212444451120164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,7168,32,0.01478488908873664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,7168,32,0.023041778140597876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,1536,0.0540960000620948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,7168,128,0.020999999509917364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,1024,0.04469688733418783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,65536,1.5921128590901692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,65536,2.4147262573242188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,16384,0.6217857996622721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,16384,0.40291110674540204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,768,0.0394702222612169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,65536,0.8953510920206705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,12288,0.43580089675055605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,12288,0.28698134422302246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,512,0.034908443689346313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,16384,0.19885778427124023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,10240,0.3629315429263645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,10240,0.22166311740875244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,12288,0.14613421758015951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,256,0.030478222502602473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,8192,0.294739564259847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,8192,0.1936497820748223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,7168,128,0.026707554856936138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,10240,0.12350400288899739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,7168,0.2562097708384196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,7168,0.15765866968366835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,8192,0.10070577594969009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,7168,0.08952889177534316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,6144,0.21610221597883436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,6144,0.13679644796583387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,65536,1.5596639845106337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,6144,0.07907555500666301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,5120,0.17920622560713026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,5120,0.11577867137061225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,16384,0.3614355458153619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,12288,0.27045154571533203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,4096,0.10076710912916396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,4096,0.14199288686116537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,4096,0.09478488895628188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,5120,0.06951555278566149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,3584,0.1250444385740492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,3584,0.0833466649055481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,10240,0.22926667001512316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,3072,0.08101422256893583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,3072,0.10910399754842122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,4096,0.057800889015197754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,3072,0.07391466697057088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,8192,0.18556977642907035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,2560,0.09079466925726996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,2560,0.06292533212237887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,3584,0.052143110169304736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,3072,0.046153777175479464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,7168,0.1630151139365302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,2048,0.07324533330069648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,2048,0.05198577708668179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,2560,0.04119822382926941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,1536,0.05627466572655571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,1536,0.04426222377353244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,6144,0.14310399691263834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,2048,0.034112887250052557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,1024,0.038204444779290095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,1024,0.03327288892534044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,1536,0.02880177895228068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,1024,0.023057777020666335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,768,0.03149066699875726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,768,0.03161333335770501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,5120,0.1213937732908461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,768,0.021241777473025854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,512,0.022845332821210224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,512,0.026736888620588515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,3584,0.09087377786636353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,512,0.020523554748959012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,256,0.01665599975321028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,256,0.023015111684799194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,2560,0.07142221927642822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,128,0.02513866623242696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,128,0.013746666411558786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,128,0.02090222140153249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,256,0.01925155520439148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,64,0.012721777790122561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,64,0.019889778561062284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,6144,32,0.01292888902955585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,6144,32,0.020607999629444547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,2048,0.06212888823615181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,1536,0.05231022172504001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,6144,128,0.019262222780121695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,1024,0.04212977819972568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,65536,1.3097875383165147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,65536,2.1426542070176864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,16384,0.5075404379102919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,768,0.037808001041412354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,16384,0.430418676800198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,65536,0.7615200148688422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,12288,0.343378676308526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,12288,0.2534204324086507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,16384,0.16684888468848336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,10240,0.3023680051167806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,10240,0.18989955054389107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,12288,0.12715910540686712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,10240,0.10747288995318943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,512,0.032270222902297974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,8192,0.24391465716891816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,8192,0.15286755561828613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,8192,0.08774489164352417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,7168,0.21231288380093047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,7168,0.13246933619181314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,6144,256,0.02840533190303379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,7168,0.07747199800279406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,6144,0.186408003171285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,6144,0.11494044462839763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,65536,1.2804284625583224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,6144,0.06948622067769368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,16384,0.31339022848341197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,5120,0.14863822195265028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,5120,0.09894666406843398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,12288,0.23539288838704428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,4096,0.11983911196390788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,4096,0.08071911334991455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,5120,0.062013334698147245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,10240,0.1969697740342882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,3584,0.10628889004389445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,3584,0.07236888673570421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,4096,0.052744001150131226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,3584,0.04645511176851061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,8192,0.15938222408294678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,3072,0.09160533216264512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,3072,0.06254133250978258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,3072,0.041328890456093684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,2560,0.07687289184994169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,2560,0.05424088901943631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,7168,0.14110488361782497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,2560,0.0362391107612186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,2048,0.06217510832680596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,2048,0.04514400164286295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,6144,0.12344266308678521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,2048,0.031202667289310034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,1536,0.04747555653254191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,1536,0.03771022293302748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,4096,0.08824444479412502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,5120,0.1048168871137831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,1024,0.03165510959095425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,1024,0.02936089038848877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,1536,0.025198222862349615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,768,0.034745779302385114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,768,0.02476799984773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,768,0.02752888864941067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,3584,0.0792693363295661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,1024,0.02070577773782942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,512,0.01813066667980618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,512,0.023481777972645227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,768,0.019493333167499967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,3072,0.0724453330039978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,256,0.013840888937314352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,256,0.020484444167878892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,512,0.018699554933442008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,256,0.017817777064111497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,128,0.012087110843923358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,128,0.01827111177974277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,2560,0.06289244360393949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,64,0.010640888578361936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,64,0.017851556340853374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,5120,32,0.012096000214417776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,5120,32,0.017895110779338412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,5120,128,0.017841777867741056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,2048,0.05447377761205038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,65536,1.5084622701009114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,65536,1.1189564598931205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,65536,0.6034622192382812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,1536,0.04575111137496101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,16384,0.3684488932291667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,16384,0.2746906545427111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,1024,0.03762133253945245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,12288,0.2782906691233317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,12288,0.1952435572942098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,16384,0.1419173346625434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,10240,0.18537510765923393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,10240,0.24580889277988008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,10240,0.16755910714467367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,512,0.030067556434207495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,12288,0.10884710815217759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,8192,0.19723110728793672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,8192,0.12890044848124185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,256,0.025803556044896443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,10240,0.09359555774264866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,7168,0.17131377591027153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,7168,0.11269156138102214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,8192,0.07600444555282593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,5120,128,0.024329778220918443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,6144,0.14517866240607366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,6144,0.09727644258075291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,7168,0.0691795547803243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,6144,0.06174400117662218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,5120,0.12001600530412461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,5120,0.08218933476342095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,5120,0.05503466725349426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,65536,1.1953502231174045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,4096,0.10141866736941868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,4096,0.08098755280176799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,4096,0.046500444412231445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,16384,0.2871697743733724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,3584,0.08495289087295532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,3584,0.06044889158672757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,3584,0.040707555082109236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,12288,0.21606400277879503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,3072,0.07338666915893555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,3072,0.05343822307056851
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,8192,0.14749599827660456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,3072,0.03623466690381368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,2560,0.061660442087385386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,2560,0.045957333511776395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,7168,0.1300284465154012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,2048,0.05162222186724345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,2048,0.050026665131251015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,2048,0.03878044419818454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,6144,0.11440443992614746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,2560,0.03250933355755276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,1536,0.038027554750442505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,1536,0.032447109619776406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,5120,0.09831644429100884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,2048,0.027867555618286133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,1024,0.026368889543745253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,1024,0.025064888927671645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,1536,0.023181334137916565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,1024,0.01889955500761668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,768,0.020789333515697055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,768,0.023767999476856653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,4096,0.08239822255240546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,768,0.017855111095640395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,512,0.015232000086042615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,512,0.019932443896929424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,512,0.017292444904645283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,3584,0.07519644498825073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,256,0.012320888539155325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,256,0.017083555459976196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,3072,0.06722399923536512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,2560,0.059454222520192467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,128,0.010739555789364709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,128,0.015861334072219003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,256,0.016523554921150208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,64,0.009768889182143742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,64,0.015581332974963717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,4096,32,0.010130667024188572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,4096,32,0.015503999259736804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,1536,0.04427822099791633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,4096,128,0.016201777590645682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,1024,0.035642666949166193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,65536,0.967142211066352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,65536,1.4666017956203883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,16384,0.39804532792833114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,768,0.03273244367705451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,16384,0.2762862311469184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,65536,0.5409822463989258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,12288,0.2661502096388075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,12288,0.17616355419158936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,16384,0.13052533732520208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,512,0.02797777785195245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,10240,0.21915555000305176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,12288,0.1011600030793084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,10240,0.14443022674984402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,10240,0.0865395532713996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,256,0.024622221787770588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,8192,0.1721928914388021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,8192,0.11676888995700413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,8192,0.07152089145448473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,7168,0.1490844488143921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,7168,0.10177066591050889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,7168,0.06481599807739258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,4096,128,0.022640888889630634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,6144,0.12668622864617243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,6144,0.08669866455925836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,6144,0.05751111110051473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,65536,1.0931190914577908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,5120,0.10345777538087632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,5120,0.08145688639746772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,12288,0.20065600342220732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,16384,0.2645040088229709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,4096,0.08571555879380967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,4096,0.06092444393369886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,5120,0.05266222357749939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,4096,0.043086220820744835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,3584,0.07493066787719727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,3584,0.05421688821580675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,10240,0.16812800036536324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,3072,0.06272355715433757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,3072,0.06476089027192858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,3072,0.04801333281728956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,8192,0.1372684505250719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,3584,0.038995554049809776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,2560,0.054069333606296115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,2560,0.04126133190260993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,3072,0.03455555438995361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,7168,0.12143288718329535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,2048,0.04379822148217095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,2048,0.03517955541610718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,2560,0.03046755658255683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,2048,0.026176888081762526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,1536,0.033360888560613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,1536,0.029613332615958318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,6144,0.10776711172527736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,1536,0.02159022291501363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,1024,0.023189332750108507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,1024,0.023137778043746948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,1024,0.017813333206706576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,5120,0.09198933177524143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,768,0.018449778358141582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,768,0.021926222576035395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,768,0.017087111870447796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,4096,0.07754577530754937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,512,0.014507555299335055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,512,0.01752622259987725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,512,0.016756445169448853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,3584,0.07006666395399305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,256,0.010796444283591377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,256,0.015512888630231222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,256,0.015835555063353646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,2560,0.055143998728858105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,128,0.009318222602208456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,128,0.014775110615624322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,2048,0.04785866538683573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,64,0.008649778034951951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,64,0.014339556296666464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3584,32,0.010057777994208867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3584,128,0.01572533282968733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3584,32,0.014490667316648694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,1536,0.040694223509894475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,65536,1.1418533325195312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,65536,0.9138719770643445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,1024,0.03393955694304572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,16384,0.2854435443878174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,16384,0.3799742327796088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,16384,0.2184373272789849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,65536,0.5114675627814399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,12288,0.21430844730801055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,16384,0.11944444974263509
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,12288,0.15403910477956137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,768,0.03188088867399428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,12288,0.09073599841859604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,10240,0.1856506665547689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,10240,0.12899555100335017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,512,0.027071111732059058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,8192,0.14750577343834773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,8192,0.10189333226945664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,10240,0.07790933052698772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,8192,0.06503822406133015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,7168,0.12862133979797363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,7168,0.08935644229253133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,256,0.024403555525673762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,7168,0.058712886439429395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,6144,0.1089937753147549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,6144,0.07692355579800077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,6144,0.0535715553495619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3584,128,0.022396443618668452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,5120,0.09115111165576512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,5120,0.06566933128568861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,5120,0.04862133330769009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,4096,0.07381155755784777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,4096,0.07315733035405476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,4096,0.054304000404145986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,65536,0.9746266471015083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,4096,0.03997777899106344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,3584,0.06405155526267158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,3584,0.04853777752982246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,12288,0.18919644090864393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,3584,0.03629866573545668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,3072,0.05622577667236328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,3072,0.04291911257637871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,10240,0.1579671170976427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,3072,0.03266755408710904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,2560,0.04630755715899997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,2560,0.03717333409521315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,2560,0.028918223248587713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,8192,0.12979643874698216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,2048,0.03773599863052368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,2048,0.031888888941870794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,7168,0.11531021859910752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,2048,0.02498400045765771
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,1536,0.029049777322345312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,1536,0.02712533374627431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,6144,0.10128177536858453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,1536,0.02085599965519375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,1024,0.020785777105225455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,1024,0.021173333128293354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,5120,0.08699466784795125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,1024,0.017128888103697035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,768,0.016862221890025668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,768,0.02032888929049174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,768,0.01624000072479248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,3584,0.06681511137220594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,512,0.01273333364062839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,512,0.015843555331230164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,3072,0.060212446583641895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,512,0.015621332658661736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,256,0.011043555206722684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,256,0.014145778285132514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,256,0.014817777607176038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,2560,0.05277155505286323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,128,0.009536888864305284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,128,0.01350844403107961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,3072,128,0.015007999208238391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,64,0.010039110978444418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,64,0.013067555924256643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,3072,32,0.008737777670224508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,3072,32,0.01311822235584259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,2048,0.0465848876370324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,1536,0.03956177830696106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,65536,0.9637973573472766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,1024,0.032411555449167885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,65536,0.7417493396335177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,16384,0.2527253362867567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,65536,0.4337777826521132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,16384,0.20547821786668566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,768,0.031125333574083116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,12288,0.18482222821977404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,16384,0.12094310919443767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,12288,0.1343404452006022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,12288,0.08228711287180583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,10240,0.15380799770355225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,10240,0.11220710807376438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,512,0.02539288832081689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,10240,0.07178755601247151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,8192,0.12240178055233425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,8192,0.08951910999086167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,8192,0.059564444753858775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,256,0.02299733294381036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,7168,0.10650489065382217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,7168,0.10621066888173421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,7168,0.07825777928034465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,7168,0.05483999848365784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,6144,0.0909128917588128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,6144,0.06839466757244535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,3072,128,0.021592888567182753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,6144,0.050027555889553495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,5120,0.07567377885182698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,5120,0.0588408907254537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,65536,0.9016062418619791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,16384,0.22629067632887098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,5120,0.04599022203021579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,4096,0.06051911248101128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,4096,0.04917155702908834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,4096,0.03671288821432326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,3584,0.05336622065967984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,3584,0.04302311274740431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,12288,0.17190310690138075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,3584,0.03327555457750956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,10240,0.1447884506649441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,3072,0.045800887876086764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,3072,0.038118223349253334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,3072,0.03002133303218418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,2560,0.03851911094453599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,2560,0.033168001307381526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,2560,0.026707554856936138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,8192,0.11932533317142063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,2048,0.031231999397277832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,2048,0.028667555914984807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,2048,0.02291822267903222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,6144,0.0936995546023051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,1536,0.023743111226293776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,1536,0.024654222859276667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,1536,0.019191111127535503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,5120,0.08077866501278348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,1024,0.016939555605252583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,1024,0.018783999813927543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,1024,0.015961777832772996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,4096,0.06862577464845446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,768,0.014075555735164218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,768,0.018213333355055917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,768,0.015318221516079374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,3584,0.0613466633690728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,512,0.010623999767833285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,512,0.014114666316244336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,512,0.01477599971824222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,2560,0.0495742228296068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,256,0.008039111064540016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,256,0.0123857781291008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,3072,0.05547733439339531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,256,0.014134221606784396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,128,0.007256888680987888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,128,0.012004444168673621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,2048,0.04324977927737766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,64,0.006697777658700943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2560,128,0.014189332723617554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,64,0.011728888584507836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2560,32,0.006590222318967183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2560,32,0.01202044470442666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,65536,0.7303653293185763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,1536,0.036779556009504534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,65536,0.6518453492058648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,65536,0.3643937905629476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,16384,0.18394933806525338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,16384,0.15453510814242893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,16384,0.09454844395319621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,1024,0.030416889323128596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,12288,0.13937066660987005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,12288,0.11541688442230225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,768,0.02968888812594944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,12288,0.07338666915893555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,10240,0.12415822347005208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,10240,0.09632622533374363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,10240,0.0645680030186971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,512,0.024361777636739943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,8192,0.10024355517493354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,8192,0.07766222291522555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,8192,0.05440800057517158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,256,0.022742221752802532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,7168,0.0886533326572842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,7168,0.06766844458050199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,7168,0.048680000834994845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2560,128,0.021019554800457422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,6144,0.07414755556318495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,6144,0.05836621920267741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,6144,0.04459999998410543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,65536,0.7939475377400717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,5120,0.06047288576761881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,5120,0.05340622199906243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,16384,0.201218671268887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,5120,0.040770666466818914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,4096,0.04900888933075798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,4096,0.04169155491722954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,4096,0.033406221204333834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,12288,0.15326578087276882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,3584,0.04394844505521986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,3584,0.03771911064783732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,10240,0.1291688945558336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,3584,0.030576888057920668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,3072,0.03822133276197646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,3072,0.03391022152370877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,3072,0.027315555347336665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,8192,0.1064186692237854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,2560,0.03166044420666165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,2560,0.02946399980121189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,7168,0.09540977742936875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,2560,0.024525332782003615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,2048,0.026250667042202417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,2048,0.025232000483406916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,6144,0.08323999908235338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,2048,0.021187555458810594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,1536,0.020745777421527438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,1536,0.021852445271280076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,1536,0.017975111802419026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,5120,0.0726648900243971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,1024,0.015113777584499784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,1024,0.016869333055284288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,4096,0.06172800064086914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,1024,0.014944000376595391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,768,0.012705778082211813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,768,0.015899555550681222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,768,0.014144889182514615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,3584,0.056016001436445445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,512,0.010040000081062317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,512,0.012150222228633033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,512,0.013791110780504016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,3072,0.0505617790751987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,256,0.008065777520338694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,256,0.010984888507260216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,256,0.013099555340078143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,2560,0.04558133416705661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,128,0.007000889215204451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,2048,0.03986666599909464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,128,0.010964444114102257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,64,0.006608888920810487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,64,0.01070844464831882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,2048,32,0.006616000086069107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,2048,32,0.010724444356229572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,2048,128,0.013400889105266996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,65536,0.5693084398905436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,65536,0.5389199786716038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,1536,0.033639109796947904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,65536,0.3815564314524333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,16384,0.15390222602420384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,16384,0.1543955538007948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,16384,0.08715733554628159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,1024,0.028198222319285076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,768,0.028142223755518597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,12288,0.10923466417524551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,12288,0.09692177507612441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,12288,0.06519644127951728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,10240,0.09820799695120917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,10240,0.08009511232376099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,10240,0.057822220855289035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,512,0.022787556052207947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,8192,0.07817600170771281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,256,0.021703999903466966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,8192,0.06529600090450711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,8192,0.0488319993019104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,7168,0.06730310784445868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,7168,0.057125336594051786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,7168,0.04360888732804192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,2048,128,0.01995644470055898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,6144,0.05794755617777506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,6144,0.04945422212282816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,6144,0.03975822197066413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,65536,0.6943804423014323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,5120,0.048306667142444186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,5120,0.04379733403523763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,5120,0.03673955467012193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,16384,0.1808746655782064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,4096,0.039444443252351545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,4096,0.035395556026034884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,4096,0.030023111237419978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,12288,0.14016621642642552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,3584,0.034203555848863386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,3584,0.03278399838341607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,3584,0.028011557128694322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,10240,0.11865244971381293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,3072,0.029652443197038438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,3072,0.02869510981771681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,3072,0.025222222010294598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,8192,0.09797155857086182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,2560,0.02531555626127455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,2560,0.025224000215530396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,7168,0.08756089210510254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,2560,0.02208799951606327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,2048,0.020971554848882887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,2048,0.021692444880803425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,6144,0.07721689012315539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,2048,0.019867555962668527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,1536,0.016146666473812528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,1536,0.018927999668651156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,1536,0.01643022232585483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,5120,0.0672560003068712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,1024,0.011995555626021491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,1024,0.014429334137174817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,1024,0.013574221895800697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,4096,0.05741688940260145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,768,0.010039110978444418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,768,0.013611555927329592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,768,0.012856889102194043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,3584,0.051776889297697276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,512,0.00833422193924586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,512,0.01072088877360026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,3072,0.047525333033667676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,512,0.012834666503800286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,256,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,256,0.009715555442704095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,2560,0.04243377844492594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,256,0.012161778079138862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,128,0.018727999594476488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,128,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,128,0.009502222140630087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1536,128,0.012060444388124677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,64,0.005864000154866113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,64,0.009332444104883406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1536,32,0.005881777654091517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1536,32,0.009287999735938178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,65536,0.392339547475179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,65536,0.4125813378228082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,2048,0.03749333487616645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,1536,0.03212533394495646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,16384,0.0995111134317186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,65536,0.28514933586120605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,16384,0.10269333256615533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,16384,0.07368533478842841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,12288,0.07523110840055677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,12288,0.0789582199520535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,12288,0.05592266718546549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,1024,0.02642400066057841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,10240,0.06311200062433879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,10240,0.06641244226031832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,10240,0.05063466562165154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,8192,0.08892799748314752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,8192,0.05117600162823995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,8192,0.05384711093372769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,8192,0.04156977931658427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,768,0.026155556241671245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,7168,0.04489422175619337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,7168,0.046903110212749906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,7168,0.03821333249409994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,512,0.02128444446457757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1536,256,0.01942044496536255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,6144,0.03929511043760512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,6144,0.04174222217665779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,6144,0.03544266687499152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,5120,0.03313422203063965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,5120,0.03516888949606154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,5120,0.03300889001952277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,65536,0.6212906837463379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,4096,0.026985777748955622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,4096,0.029787556992636785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,4096,0.0266275554895401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,12288,0.12691999806298151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,3584,0.02344177828894721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,3584,0.02700888944996728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,16384,0.16672533088260225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,3584,0.02472266720400916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,3072,0.020748444729381137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,3072,0.024160888459947374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,10240,0.10770577854580349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,3072,0.02204533252451155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,2560,0.017892445127169292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,2560,0.0210844443904029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,2560,0.019448000523779128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,7168,0.0796497795316908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,2048,0.01569333341386583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,2048,0.01827822294500139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,2048,0.01720533271630605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,6144,0.07055999835332234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,1536,0.013157332936922709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,1536,0.015577778220176697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,1536,0.014378666877746582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,5120,0.0614853302637736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,1024,0.010071111222108206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,1024,0.01239911135700014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,1024,0.012090666426552666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,4096,0.052637332015567355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,768,0.008206222620275285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,768,0.011742221812407175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,768,0.011389333340856763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,3584,0.0476435555352105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,512,0.0069475554757648045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,512,0.009003555609120263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,512,0.011083555718262991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,2560,0.03937777876853943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,256,0.00553955551650789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,256,0.008285333712895712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,3072,0.043756445248921715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,256,0.010947555303573608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,128,0.004910222358173794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,128,0.008275555239783393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,2048,0.03461600012249417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,64,0.004551110996140374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,1024,128,0.010760000182522668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,64,0.008217777642938826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,1024,32,0.0041991112132867174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,1024,32,0.008287111090289222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,65536,0.3070017761654324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,65536,0.4197857909732395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,65536,0.26740890079074436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,1536,0.029842668109469946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,1024,0.024939555260870192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,16384,0.07947999901241727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,16384,0.10133155849244858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,16384,0.07085333267847697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,12288,0.05990133020612928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,12288,0.07741600275039673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,768,0.023767110374238756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,12288,0.05607999695671929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,10240,0.05031466815206739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,10240,0.06610933277342054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,10240,0.05041333370738559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,512,0.018216888109842937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,8192,0.04096444447835287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,8192,0.0529386666085985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,8192,0.04118311074044969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,256,0.017860444055663217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,7168,0.03637511200375027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,7168,0.045808000697029956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,7168,0.03804533349143134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,1024,128,0.016855110724767048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,6144,0.03079644507831997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,6144,0.04020266731580099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,65536,0.5562755796644423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,6144,0.03511111272705926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,5120,0.02581688927279578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,5120,0.03460444344414605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,5120,0.03277600142690871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,16384,0.15042400360107422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,4096,0.02105955613984002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,4096,0.029548443026012842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,4096,0.026660444007979497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,12288,0.11645778020222981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,3584,0.01888088881969452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,3584,0.026752889156341553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,3584,0.024297777149412367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,10240,0.09932000107235378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,8192,0.08254488971498278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,3072,0.01682577696111467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,3072,0.023961777488390606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,3072,0.021879111727078755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,2560,0.014929778046078153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,2560,0.02090311050415039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,2560,0.019368888603316415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,7168,0.07367555300394694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,2048,0.013120888835854001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,2048,0.01805688937505086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,2048,0.016833777228991192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,6144,0.06469066937764485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,1536,0.011003555523024665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,1536,0.01574222246805827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,1536,0.014075555735164218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,5120,0.057532442940606006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,1024,0.008551999926567078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,1024,0.011719110939237805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,1024,0.01184800018866857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,4096,0.04921244581540426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,768,0.0069466663731469055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,768,0.011355555719799466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,768,0.011078221930397881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,3584,0.04529955652025011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,512,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,512,0.008964444200197855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,512,0.011064888702498542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,3072,0.04131999942991469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,256,0.004879999905824661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,256,0.00833511104186376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,256,0.010681777364677854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,2560,0.03659822212325202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,128,0.004559111264016894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,128,0.008053333395057255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,768,128,0.010856000085671743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,64,0.004176889028814104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,64,0.007983999947706858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,768,32,0.004209777961174647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,768,32,0.00794044468137953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,2048,0.03252089023590088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,65536,0.2230924500359429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,65536,0.30995111995273167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,65536,0.26803821987575954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,1536,0.02844088938501146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,16384,0.05675644344753689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,16384,0.08083199792438082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,16384,0.06809510787328084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,1024,0.023360000716315374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,12288,0.04333422250217862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,12288,0.06264355447557238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,768,0.02334311107794444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,12288,0.04850044515397814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,10240,0.03637955586115519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,10240,0.053278221024407275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,10240,0.048688001102871366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,8192,0.0765262246131897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,8192,0.02938844429122077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,8192,0.04251199960708618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,8192,0.0351164440313975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,256,0.018523555662896898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,7168,0.02534488836924235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,7168,0.03678399986690945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,512,0.018199111024538677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,7168,0.03649422195222642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,6144,0.021746666895018682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,6144,0.03195733494228787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,6144,0.03344800074895223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,768,128,0.01647555496957567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,5120,0.018696000178654987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,5120,0.028035554620954726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,65536,0.528897762298584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,5120,0.03151999910672506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,4096,0.01555999947918786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,4096,0.024043555061022442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,4096,0.023008000519540574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,16384,0.1430062188042535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,3584,0.014128888646761576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,3584,0.021976000732845728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,3584,0.02160888910293579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,10240,0.09456800089942084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,3072,0.012441777520709567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,12288,0.11053066783481175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,3072,0.019488000207477145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,3072,0.019153777096006606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,2560,0.011400889191362592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,2560,0.017557332913080852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,2560,0.016917333006858826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,7168,0.06877866718504164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,2048,0.009995555712117089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,2048,0.015159999330838522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,2048,0.015099555253982544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,5120,0.05381244421005249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,6144,0.06194400125079685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,1536,0.008584000170230865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,1536,0.013153778182135688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,1536,0.012917333179050021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,1024,0.006602666858169768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,1024,0.00997155573632982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,1024,0.01016355554262797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,4096,0.046289778418011136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,768,0.005712000032265981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,768,0.00963466697269016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,768,0.009954666925801171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,3584,0.04236355423927307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,512,0.004876444323195351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,512,0.0074311114019817775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,512,0.01000355515215132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,3072,0.0381644434399075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,256,0.004193777839342753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,256,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,2560,0.0348364445898268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,256,0.010030222435792288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,128,0.0038399998512532976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,128,0.007306666837798224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,512,128,0.010045333041085137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,64,0.003642666671011183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,64,0.0069475554757648045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,512,32,0.003563555578390757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,512,32,0.007000000112586551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,2048,0.0304097765021854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,65536,0.12824977768792048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,1536,0.026610665851169165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,65536,0.30031734042697483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,65536,0.24566488795810273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,16384,0.03965866565704346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,16384,0.07973600096172757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,16384,0.06622133652369182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,768,0.021224000387721594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,1024,0.021327111456129286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,12288,0.03042222062746684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,12288,0.061108443472120494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,12288,0.04744355546103584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,10240,0.025461332665549383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,10240,0.051743110020955406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,10240,0.048222220606274076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,512,0.01646222174167633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,8192,0.020941333638297185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,8192,0.041195554865731135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,8192,0.03529777791765001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,256,0.017156443662113614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,7168,0.018645332919226754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,7168,0.036095112562179565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,7168,0.035932444863849215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,512,128,0.014771555860837301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,6144,0.016531555189026725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,6144,0.03116977877087063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,6144,0.033230221933788724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,65536,0.5055155754089355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,5120,0.014505777094099255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,5120,0.027467555469936792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,5120,0.0310444434483846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,16384,0.1369653277926975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,4096,0.012706666356987424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,4096,0.02362666692998674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,12288,0.10424800051583184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,4096,0.023111111587948267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,3584,0.039883553981781006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,3584,0.011424000064531961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,3584,0.021235555410385132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,3584,0.021368000242445204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,3072,0.010388444695207808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,3072,0.01888799998495314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,3072,0.01923466722170512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,8192,0.07403644588258532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,2560,0.00906577789121204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,2560,0.017032000753614638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,10240,0.08978044324451023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,2048,0.029500444730122883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,2560,0.017163554827372234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,2048,0.007985778152942657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,2048,0.014775110615624322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,2048,0.015240889456537036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,1536,0.025482666161325242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,1536,0.006606222026877933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,1536,0.013006221916940479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,1536,0.012315555579132505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,1024,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,7168,0.06577599710888334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,1024,0.009463999834325578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,1024,0.010058666268984476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,768,0.004921777794758479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,768,0.009298666483826106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,768,0.009983110758993361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,512,0.015441776977645027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,512,0.004222222086456087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,512,0.007613333563009898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,512,0.010016889207892949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,6144,0.058902223904927574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,256,0.0034942221310403612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,256,0.0069324444565508105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,256,0.009714666340086196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,5120,0.05118311113781399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,128,0.0034888887570963967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,128,0.006948444578382704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,256,128,0.00997688869635264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,64,0.003165333428316646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,64,0.006897777732875612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,256,32,0.0034773333205117118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,256,32,0.006615110983451207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,4096,0.04398399922582838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,65536,0.09519200192557441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,3072,0.03609244359864129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,65536,0.3020435439215766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,65536,0.24904176923963758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,16384,0.030191110240088567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,16384,0.07880977789560954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,2560,0.03297777639495002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,16384,0.06647555695639716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,12288,0.024524443679385718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,12288,0.06087377998563978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,12288,0.048151999711990356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,1024,0.01961688862906562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,10240,0.019010666343900893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,10240,0.05136177937189738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,10240,0.04807288779152764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,768,0.020607999629444547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,256,0.014114666316244336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,8192,0.016604445046848722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,8192,0.04058400127622817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,7168,0.0644657810529073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,8192,0.036032888624403216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,7168,0.014655111564530266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,7168,0.03537955549028184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,7168,0.036269333627488874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,6144,0.01293777757220798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,6144,0.031084444787767198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,6144,0.033463110526402794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,256,128,0.01312800000111262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,5120,0.011756444142924415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,5120,0.027452444036801655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,5120,0.03075466553370158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,65536,0.4952479998270671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,4096,0.010400888820489248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,4096,0.023294222023752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,4096,0.023656888140572443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,16384,0.13400533464219835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,3584,0.009371555513805812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,3584,0.021242666575643752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,3584,0.021907554732428655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,10240,0.08612444665696885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,12288,0.1028311120139228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,3072,0.008608000146018134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,3072,0.01887911061445872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,3072,0.019322666856977675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,2560,0.0075839997993575195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,8192,0.0725368857383728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,2560,0.016751110553741455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,2560,0.017264889346228707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,2048,0.006670222100284364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,2048,0.014409777190950183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,2048,0.01533422205183241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,5120,0.050013333559036255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,1536,0.005531555662552516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,1536,0.01276177747382058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,1536,0.012088889049159156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,6144,0.05731911129421658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,1024,0.004534222185611725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,1024,0.00902844468752543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,1024,0.010022222167915767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,4096,0.04333422250217862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,768,0.0038782221575578055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,768,0.009682666924264696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,768,0.009669333696365356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,3584,0.03936977850066291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,512,0.0038195554580953387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,512,0.007277333074145847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,512,0.010040000081062317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,3072,0.03579911258485582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,256,0.0032239999208185407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,2048,0.02825511164135403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,256,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,1536,0.024264888630972967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,256,0.010052444206343757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,128,0.0031537777847713898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,128,0.012889777620633444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,2560,0.032272001107533775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,128,0.006968888971540663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,64,0.002862222285734283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4096,128,128,0.00998933364947637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,64,0.006607111129495833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,128,32,0.0032213332338465583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,128,32,0.0069413334131240845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,65536,0.08728888962003921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,16384,0.027965333726671007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,16384,0.0785964462492201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,65536,0.29304801093207467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,12288,0.022779555784331426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,12288,0.060903999540540904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,10240,0.01888533267709944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,10240,0.05113866594102648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,8192,0.01593244406912062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,8192,0.04074044360054864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,7168,0.0144213338692983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,7168,0.0355386667781406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,6144,0.01276000009642707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,6144,0.0307608875963423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,5120,0.011388444238238864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,5120,0.02728355593151516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,4096,0.010426667001512317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,3584,0.009124444590674506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,4096,0.023005333211686876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,3072,0.008319999608728621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,3584,0.0206995556751887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,3072,0.01891644464598762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,2560,0.0074746666683091064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,2048,0.006280888699822956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,2560,0.01686044368478987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,2048,0.014502222339312235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,1536,0.005084444251325396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,1536,0.013337777720557319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,1024,0.004212444441186057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,1024,0.009478222164842818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,768,0.0038720000949170855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,768,0.009669333696365356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,512,0.007326222128338284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,512,0.0035297779573334586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,256,0.0031822222388452957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,128,0.0030373332815037835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,256,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,128,0.006963555183675554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,64,0.0030293334275484085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,64,0.006791111081838608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,64,32,0.006670222100284364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,64,32,0.00286311118139161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,16384,0.02641333308484819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,65536,0.08864088853200276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,16384,0.0788266658782959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,12288,0.02110311057832506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,12288,0.06053600046369764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,10240,0.018650665879249573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,1024,0.01891288823551602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,10240,0.050973332590527005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,65536,0.2983528772989909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,8192,0.016164445214801364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,8192,0.04035288757748074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,7168,0.03500444359249539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,6144,0.013454222016864352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,7168,0.014871110518773397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,6144,0.030530667967266504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,5120,0.013056000073750814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,5120,0.027355555031034682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,4096,0.011594666375054253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,4096,0.0229404436217414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,3584,0.01000711073478063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,2560,0.008595555192894405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,3584,0.023058666123284235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,3072,0.009450666606426239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,3072,0.02065333392884996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,2560,0.018224888377719455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,2048,0.006226666685607698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,2048,0.01548266741964552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,1536,0.004917333523432414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,1536,0.01348088930050532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,1024,0.003871110992299186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,1024,0.009576000273227692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,768,0.0038382220599386427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,768,0.009703999592198266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,256,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,512,0.003245333416594399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,512,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,256,0.0032355555643637977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,768,0.018989332848125033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,128,0.0029395555870400537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,128,0.0069271110826068455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,64,0.0028604444944196274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,64,0.006846222198671765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4096,32,32,0.0028604444944196274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4096,32,32,0.006970666348934174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,512,0.015461333923869662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4096,128,256,0.01408622165520986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,12288,1.2027911080254448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,16384,1.6763928731282551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,12288,2.2168256971571183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,16384,3.1073510911729603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,10240,2.088008032904731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,10240,1.0290906694200304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,8192,1.605070220099555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,8192,0.9096693462795682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,12288,1.128893322414822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,16384,1.4952177471584742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,10240,0.9461066987779406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,7168,0.72725952996148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,7168,1.364699575636122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,8192,0.722324424319797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,16384,0.7917964193556043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,6144,1.0460319519042969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,6144,0.697806199391683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,12288,0.6092737515767416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,5120,0.8868453237745496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,5120,0.5931697951422797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,10240,0.5170417891608344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,4096,0.43582932154337567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,4096,0.6883591016133627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,4096,0.5840915573967828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,8192,0.42556532224019367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,7168,0.6398293177286783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,3584,0.5973608758714464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,3584,0.4471946822272406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,6144,0.5632675488789877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,3072,0.3199511104159885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,3072,0.5198871294657389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,3072,0.357638226615058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,5120,0.4656009144253201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,7168,0.3744506571027968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,2560,0.44556087917751735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,2560,0.31904710663689506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,6144,0.3272826671600342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,3584,0.34447820981343585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,2048,0.36356533898247617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,2048,0.2943449020385742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,5120,0.28032533327738446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,1536,0.18133510483635795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,1536,0.27583466635810006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,1536,0.1966773271560669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,4096,0.23402667045593262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,2560,0.25690934393141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,1024,0.18880977895524767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,1024,0.14806311660342747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,3584,0.2077315515942044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,768,0.10239644183052911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,768,0.1480720043182373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,768,0.12774844964345297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,2048,0.2170471085442437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,3072,0.1837057802412245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,512,0.11179021994272868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,512,0.11598043971591526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,2560,0.15968355867597792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,1024,0.15015021959940592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,256,0.07757244507471721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,256,0.10565333233939277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,2048,0.13612355126274958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,1536,0.10863377650578816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,128,0.05770044194327461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,128,0.10082399845123291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,512,0.11357333925035264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,64,0.048467556635538735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,64,0.09945155514611138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,65536,32,0.05007466673851013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,65536,32,0.09969511297014023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,1024,0.08521866798400879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,768,0.07582755883534749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,256,0.07039378086725871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,65536,3.289983961317274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,65536,2.002752939860026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,512,0.0678079989221361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,16384,0.8944746653238932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,16384,0.42472977108425564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,65536,128,0.06896889209747314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,12288,0.3074835671318902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,12288,0.5637475649515787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,12288,0.321399105919732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,10240,0.2608666684892443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,10240,0.4380355676015218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,65536,0.9229013654920789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,10240,0.26888354619344074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,16384,0.21693778038024902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,8192,0.35639111200968426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,8192,0.21889955467647979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,12288,0.16782222853766549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,7168,0.18145599630143908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,256,0.061392002635531955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,7168,0.31636977195739746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,65536,1.8865394592285156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,7168,0.1983271174960666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,6144,0.2697226736280653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,6144,0.20641777250501844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,10240,0.14135200447506374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,6144,0.17030044396718344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,5120,0.2228177653418647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,5120,0.16683644718594023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,65536,128,0.058802664279937744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,16384,0.3999795648786757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,4096,0.17441066106160483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,4096,0.12992000579833984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,8192,0.1161831087536282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,3584,0.09984711143705581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,3584,0.15848533312479654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,3584,0.1059546669324239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,7168,0.10429244571261936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,6144,0.09094133641984727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,3072,0.13716177145640054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,3072,0.09199377563264634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,5120,0.07818400197558932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,2560,0.07883999745051067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,2560,0.11523821618821885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,2560,0.07942399713728163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,8192,0.20368711153666177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,4096,0.06497155295477973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,2048,0.09318221939934625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,2048,0.06656889120737712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,3584,0.058324442969428174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,3072,0.05208888981077406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,1536,0.07167200247446696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,1536,0.05410933494567871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,2560,0.04581155710750156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,5120,0.1334302160474989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,1024,0.04979644550217522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,1024,0.0422346658176846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,2048,0.03985688752598233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,4096,0.11150133609771729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,768,0.03875822159979079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,768,0.03976622223854065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,1536,0.032503998941845365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,3072,0.08920177486207749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,512,0.02812622321976556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,512,0.035792890522215105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,1024,0.027457776996824477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,768,0.025233777032958135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,256,0.019746666153271992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,256,0.0325564444065094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,2048,0.06934755378299289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,1536,0.06025333536995781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,128,0.01862133377128177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,128,0.030277334981494482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,512,0.023171555664804246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,64,0.015729778342776828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,64,0.027814222706688776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,16384,32,0.016148444679048326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,16384,32,0.029177778297000464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,1024,0.05066755414009094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,256,0.021652445197105408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,768,0.04143822193145752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,16384,128,0.02072177827358246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,65536,1.407073762681749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,65536,2.344445334540473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,16384,0.615464899275038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,16384,0.3829529020521376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,512,0.04030488928159078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,12288,0.4187475575341119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,12288,0.24448890156216094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,65536,0.7244977951049805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,256,0.031890667147106595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,10240,0.3609031041463216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,10240,0.21045511298709443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,16384,0.1736666626400418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,16384,128,0.028303109937243994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,12288,0.13153155644734701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,8192,0.2883226606580946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,8192,0.17062044143676758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,10240,0.11145867241753472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,65536,1.2765369415283203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,7168,0.2499919997321235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,7168,0.1753386656443278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,8192,0.09142044517729019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,6144,0.21360445022583008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,6144,0.13050666120317247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,16384,0.3125671015845405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,7168,0.08112888866000705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,12288,0.23264267709520128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,5120,0.17797688643137613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,5120,0.11122043927510579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,6144,0.07173511054780748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,4096,0.14354666074117026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,4096,0.09176711241404216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,10240,0.1946408881081475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,8192,0.1596791081958347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,3584,0.12616533703274196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,3584,0.08148355616463555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,5120,0.06205066707399157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,4096,0.05182222194141812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,3072,0.10902311404546101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,7168,0.14090666506025526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,3072,0.07165600193871392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,3584,0.04684977730115255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,2560,0.09181777636210124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,6144,0.12439466847313775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,2560,0.06161066558625963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,5120,0.10560533073213364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,2048,0.0741751127772861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,2048,0.05188799897829691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,3072,0.042522665527131825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,2560,0.03751466671625773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,1536,0.05744889047410753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,1536,0.042591998974482216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,4096,0.08848355876074897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,2048,0.03252089023590088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,1024,0.038839999172422625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,1024,0.03382133444150289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,3584,0.08045066727532281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,1536,0.026591110560629103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,768,0.03305866652064853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,768,0.03170933326085409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,3072,0.07305777735180326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,512,0.03421422176890903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,512,0.023561777340041265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,512,0.028592887851927016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,1024,0.02278577784697215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,256,0.028613332245084975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,256,0.017136888371573556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,256,0.024737776981459722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,2560,0.06524799929724799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,2048,0.05766222212049696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,128,0.014001778430408902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,128,0.022256000174416438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,768,0.02106488909986284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,64,0.01254755589697096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,64,0.02137511140770382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,12288,32,0.012777778009573618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,12288,32,0.02200977835390303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,512,0.019520888725916546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,256,0.017940445078743827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,65536,1.9800879160563152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,65536,1.1849946975708008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,1536,0.04987555411126879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,12288,128,0.01768977774514092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,16384,0.4528115590413411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,1024,0.04247200157907274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,16384,0.2792897754245334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,12288,0.20100177658928764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,12288,0.3357040087381999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,12288,0.21071555879380968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,768,0.03575999869240655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,10240,0.29987910058763295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,65536,0.6150168842739528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,10240,0.1767653359307183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,16384,0.14744355943467882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,8192,0.2406888802846273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,8192,0.14479288789961073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,12288,0.11336178249782985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,10240,0.0956453349855211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,7168,0.20941332976023355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,7168,0.1275520059797499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,12288,128,0.025389333566029865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,8192,0.07888799905776978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,6144,0.18944888644748262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,6144,0.11124000284406875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,65536,1.1387137307061088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,7168,0.06992799705929227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,5120,0.14878933959537083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,5120,0.09445777866575454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,6144,0.06207288636101616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,16384,0.2690960036383735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,4096,0.11951111422644721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,4096,0.07730133003658719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,10240,0.16821332772572836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,3584,0.07163644499248929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,3584,0.10534844133589004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,5120,0.05372533202171326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,3584,0.06872444682651095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,8192,0.13748178217146131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,3072,0.09039822551939224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,3072,0.06009599897596571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,4096,0.04512888855404324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,2560,0.058479110399881996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,7168,0.12285689512888591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,2560,0.0764568911658393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,2560,0.05239555570814344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,3584,0.04092977775467767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,2048,0.062092443307240806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,2048,0.04405155446794298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,3072,0.03715288970205519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,2560,0.033246222469541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,1536,0.047284444173177086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,1536,0.036905778778923884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,6144,0.10743200116687351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,2048,0.02868266569243537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,1024,0.03220977716975742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,1024,0.029838220940695867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,5120,0.09244710869259304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,4096,0.07957866456773546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,768,0.02474400069978502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,1536,0.02366666661368476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,768,0.027950220637851294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,1024,0.02033333314789666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,512,0.018443556295500863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,512,0.025006221400366888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,768,0.018984888990720112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,3072,0.0649777783287896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,256,0.0140328887436125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,256,0.0214382227924135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,512,0.017656889226701524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,2048,0.05163288778728909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,128,0.012042666474978128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,128,0.019779556327395968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,1536,0.04463466670778063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,64,0.01033333357837465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,256,0.016418667303191293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,64,0.018925334016482036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,10240,32,0.010514666636784872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,10240,32,0.019330667124854196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,10240,128,0.015776000089115567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,1024,0.038034667571385704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,65536,0.9277013142903646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,65536,1.6222427156236436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,16384,0.44100001123216415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,65536,0.9387413660685221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,16384,0.24339466624789766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,768,0.032872888776991106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,512,0.031194667021433514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,12288,0.282371547487047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,12288,0.1723404460483127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,65536,0.5127502017550999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,10240,0.23910755581325957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,10240,0.15770222081078425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,16384,0.12200889322492812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,12288,0.09444533454047309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,8192,0.19502666261461046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,8192,0.1178071101506551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,256,0.026949332820044622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,10240,0.0800879995028178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,7168,0.168175114525689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,7168,0.10499378045399983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,10240,128,0.024325332707828943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,8192,0.06600266695022583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,6144,0.1449857817755805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,6144,0.09190489186180963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,7168,0.05883555279837715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,5120,0.08008444309234619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,5120,0.12028443813323975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,5120,0.07788533634609647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,6144,0.052390221092436046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,4096,0.06863200002246432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,4096,0.09757422076331244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,4096,0.0641911096043057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,5120,0.04592000113593208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,3584,0.06276355849372016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,16384,0.22569688161214194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,3584,0.08542933066685994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,3584,0.056468447049458824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,3072,0.0571848882569207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,12288,0.1698382165696886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,3072,0.07333333624733819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,3072,0.04946400059594048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,10240,0.14291821585761175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,2560,0.0614853302637736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,4096,0.03843466771973504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,2560,0.04379466507169935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,3584,0.03502488798565335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,2048,0.05012977785534329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,2048,0.037447998921076454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,8192,0.11653778288099502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,3072,0.03175466590457492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,1536,0.03780444463094076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,1536,0.031537777847713895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,2560,0.02868710954984029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,2048,0.02497244377930959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,1024,0.026400888959566753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,1024,0.025840888420740765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,7168,0.1036764449543423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,6144,0.09242222044203018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,1536,0.020638222495714825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,768,0.020610666937298246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,768,0.02362577782736884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,1024,0.0181057784292433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,512,0.015445333388116626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,512,0.021176889538764954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,768,0.016573333077960543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,256,0.025262221693992615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,256,0.01258933295806249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,256,0.018222222725550335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,512,0.015451555450757345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,256,0.014655111564530266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,128,0.010363555616802638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,128,0.016920000314712524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,2560,0.05142133434613546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,64,0.010065777434243096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,64,0.01651022169325087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,8192,32,0.010268444816271463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,8192,32,0.016587555408477783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,2048,0.04609244399600559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,8192,128,0.014183110660976835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,65536,1.424091551038954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,1536,0.03960444529851278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,65536,0.860706647237142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,1024,0.033992889854643084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,16384,0.20383111635843912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,16384,0.453837341732449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,12288,0.23750045564439562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,12288,0.28531021542019314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,65536,0.45786311891343856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,12288,0.15191022555033365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,768,0.030877331892649334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,512,0.027838223510318335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,16384,0.11136533154381646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,10240,0.1305555502573649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,10240,0.23017777336968315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,8192,0.18130844169192842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,8192,0.10502755641937256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,12288,0.08538933595021565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,10240,0.07300711340374418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,7168,0.15915111700693765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,7168,0.09389155440860325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,8192,0.06033511294258965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,8192,128,0.022516444325447083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,6144,0.13632710774739584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,7168,0.0539555549621582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,6144,0.08126133018069796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,65536,0.8840746879577637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,5120,0.11117511325412327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,5120,0.0744782222641839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,6144,0.0481644438372718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,16384,0.21385155783759224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,4096,0.09177333116531372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,4096,0.056278222137027316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,5120,0.042578667402267456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,3584,0.061463998423682324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,3584,0.07584089040756226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,3584,0.04994844396909078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,4096,0.03604088889227973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,3584,0.033358222908443876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,10240,0.13724266158209905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,3072,0.06526488727993436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,3072,0.044396443499459155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,8192,0.11342933442857529
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,7168,0.10154400269190471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,2560,0.05483377642101712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,2560,0.03887733485963609
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,3072,0.030945777893066406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,2048,0.04391288757324219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,2048,0.0336773329310947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,6144,0.0906791090965271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,2560,0.027207111318906147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,1536,0.03330577744377984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,1536,0.028743998871909246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,2048,0.02373955481582218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,1536,0.019360888335439894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,1024,0.023016888234350417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,5120,0.07827377981609769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,1024,0.023583110835817125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,768,0.03029777937465244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,768,0.018464000688658822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,768,0.02181511123975118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,4096,0.06680622365739611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,512,0.026667555173238117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,512,0.01460355520248413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,512,0.019142222073343065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,1024,0.016735111673672993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,768,0.015987555185953777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,256,0.011053333679835001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,256,0.016911110944218107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,512,0.01477599971824222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,256,0.013893332746293811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,128,0.009311999711725447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,128,0.015697777271270752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,3072,0.05595911211437649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,64,0.008613333106040955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,64,0.015562666787041558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,7168,32,0.008905777500735389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,7168,32,0.015456888410780164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,2048,0.044511109590530396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,65536,0.7965297698974609
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,2560,0.05031733380423652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,7168,128,0.01370666672786077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,65536,1.1821929083930123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,65536,0.7223386764526367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,16384,0.282042662302653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,16384,0.20770043796963164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,1536,0.03909777932696872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,12288,0.195835550626119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,12288,0.21460621886783174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,12288,0.15492444568210179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,65536,0.40406044324239093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,10240,0.178476439581977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,10240,0.12319466802808975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,16384,0.09898844692442153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,1024,0.033244444264305964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,12288,0.07544889052708943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,8192,0.14434044890933567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,8192,0.09163377682367961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,10240,0.06497333447138469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,7168,0.1267768939336141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,7168,0.08116088973151313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,8192,0.05452266666624281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,6144,0.08441866768731011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,6144,0.10762133200963338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,6144,0.07030933433108859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,7168,0.04895200000868904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,256,0.024988444315062627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,5120,0.09073511097166274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,5120,0.06093777550591362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,6144,0.04402133491304186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,7168,128,0.022279111875428095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,4096,0.07341688871383667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,5120,0.03916177815861172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,4096,0.049908445941077344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,3584,0.05822044610977173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,3584,0.06449155675040351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,3584,0.0444106658299764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,4096,0.03348177671432495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,3584,0.030375109778510198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,16384,0.19700000021192762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,3072,0.039600888888041176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,3072,0.057670222388373486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,2560,0.047876444127824574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,2560,0.04681333237224155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,2560,0.0348693331082662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,10240,0.1267440054151747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,8192,0.10591200325224136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,2048,0.03750933210055033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,2048,0.030571556753582422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,3072,0.0284222231970893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,2560,0.025063999825053748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,1536,0.029016888803905908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,7168,0.09561422136094834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,1536,0.026591999663247004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,2048,0.021959111094474792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,1024,0.02022133270899455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,1024,0.021571555071406897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,1536,0.01851733360025618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,768,0.02940533227390713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,768,0.017076444294717576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,768,0.02004888819323646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,1024,0.015929778416951496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,5120,0.07380533218383789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,512,0.013065777719020844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,768,0.015088889333936902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,512,0.016758221719000075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,4096,0.0629955530166626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,256,0.01204177737236023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,256,0.015229332778188916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,512,0.013995556367768182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,256,0.013100444442696042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,128,0.010682666467295753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,128,0.014456000592973499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,3072,0.05270399981074863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,64,0.010038221875826517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,64,0.014138667119873894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,6144,32,0.00998933364947637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,6144,32,0.014135110709402295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,6144,128,0.012881778180599213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,2048,0.042767110798094005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,65536,1.0518986384073894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,65536,0.6318106651306152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,1536,0.037408000893063016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,16384,0.2518746587965223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,1024,0.03131377696990967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,16384,0.15067733658684626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,65536,0.36058934529622394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,12288,0.19116089079115126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,12288,0.11426044834984674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,16384,0.08581955565346612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,512,0.025616889198621113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,12288,0.0665608909395006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,10240,0.15221244759029812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,10240,0.09770044353273179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,256,0.023642667465739783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,10240,0.05762400229771932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,8192,0.12041244241926406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,8192,0.07872711287604438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,6144,128,0.020824000239372253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,7168,0.1045608917872111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,7168,0.06958578030268352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,8192,0.048560887575149536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,7168,0.04421866602367825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,6144,0.0897795557975769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,6144,0.06094844473732842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,65536,0.6485022438897027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,6144,0.04011733333269755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,5120,0.0753199987941318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,5120,0.052448001172807485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,16384,0.1660604476928711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,5120,0.03600710961553786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,4096,0.06042044692569309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,4096,0.0437102218468984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,12288,0.127637333340115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,3584,0.05149066779348585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,3584,0.053095112244288124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,3584,0.03866133424970839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,4096,0.03052266769938999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,3072,0.0468355549706353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,3072,0.045796444018681846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,3072,0.03586488962173462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,10240,0.10813333590825398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,3584,0.02771111163828108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,8192,0.09000533156924778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,2560,0.03833955526351929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,2560,0.03124089042345683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,3072,0.025843555728594463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,2048,0.03108000093036228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,2048,0.027879112296634253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,7168,0.08165778054131402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,2560,0.022912888063324824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,1536,0.02380888826317257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,1536,0.02419288953145345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,2048,0.019903999235894945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,1024,0.02869866622818841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,1024,0.017010667257838778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,1024,0.019497777024904888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,6144,0.07263911432690091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,1536,0.016682666209008958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,768,0.013835555149449242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,768,0.017754667335086398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,1024,0.01461955573823717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,5120,0.06388977501127455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,512,0.01072622256146537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,512,0.015005333556069268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,768,0.013780444032616086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,512,0.012727110750145383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,256,0.007997333175606197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,256,0.013744889034165276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,4096,0.05527289046181572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,128,0.020463110672103036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,128,0.007188444336255391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,128,0.013411555853154926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,256,0.01202311118443807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,64,0.006600889066855113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,64,0.012817777693271637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,5120,32,0.006568000134494569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,5120,32,0.013064000341627332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,5120,128,0.011931555966536203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,65536,0.5741857952541775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,65536,0.7283893161349826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,65536,0.5114240116543239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,2560,0.04255022274123298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,2048,0.03770400087038676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,16384,0.18458578321668836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,16384,0.12572267320421007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,65536,0.3078426784939236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,12288,0.13776000340779623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,12288,0.09423110882441203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,16384,0.0741457806693183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,1536,0.03387377659479777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,12288,0.05727022224002414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,10240,0.12168355782826741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,10240,0.08019200298521253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,10240,0.050480888949500195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,8192,0.09771111276414658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,8192,0.06498310963312785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,768,0.02755022214518653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,512,0.023346667488416035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,7168,0.08565600050820245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,7168,0.05826399723688761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,5120,256,0.021391110287772283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,8192,0.04267022344801161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,6144,0.07303910785251193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,6144,0.050921777884165444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,7168,0.03862933317820231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,6144,0.035038222869237266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,5120,0.06112533145480686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,5120,0.04376444551679823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,5120,0.031898667414983116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,16384,0.14772710535261366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,4096,0.04884888728459676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,4096,0.0361422234111362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,12288,0.11350754896799724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,4096,0.027097778187857732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,3584,0.0429422226217058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,3584,0.03312089045842489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,10240,0.09707288609610663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,3584,0.024757333927684363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,3072,0.03707199957635667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,3072,0.030055999755859375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,8192,0.08162311050626966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,3072,0.023170666562186346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,2560,0.031481779283947416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,2560,0.02697244452105628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,7168,0.07357688744862874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,6144,0.06634400288263957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,2048,0.026242666774325903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,2048,0.024350222614076402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,2560,0.020784000555674236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,5120,0.05809422334035238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,1536,0.020562666985723708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,1536,0.020647111866209242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,2048,0.018041777941915724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,1024,0.026421333352724712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,1024,0.015776000089115567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,1024,0.01707466608948178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,1536,0.015449777245521545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,1024,0.013406222065289816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,768,0.013094222380055321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,768,0.015166223049163818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,4096,0.050479110744264394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,768,0.012706666356987424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,512,0.009916444619496664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,512,0.013103110922707452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,512,0.011742221812407175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,3584,0.04639910989337497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,256,0.0076142218377855085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,256,0.01238222254647149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,3072,0.04253777861595154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,2560,0.03889777925279405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,128,0.019891555110613506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,128,0.007135999699433644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,128,0.011742221812407175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,64,0.007320889168315464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,64,0.011738667057620155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,4096,32,0.007606222397751278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,4096,32,0.011708444191349877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,256,0.011004444625642566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,4096,128,0.010952889091438718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,65536,0.6952817704942492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,1536,0.030815998713175457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,65536,0.49951913621690536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,16384,0.18080800109439424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,2048,0.035070220629374184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,16384,0.12672089205847845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,65536,0.30143290095859104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,12288,0.12817688783009848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,12288,0.09005866448084514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,16384,0.07136533657709758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,768,0.02569244470861223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,10240,0.11597600248124863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,10240,0.07577244440714519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,12288,0.05561244487762451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,10240,0.049048887358771436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,8192,0.09302755859163071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,8192,0.06153421931796604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,512,0.02198666665289137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,8192,0.04119999872313605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,7168,0.08065778017044067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,7168,0.0552151103814443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,4096,256,0.020850666695170935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,6144,0.06316622098286946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,6144,0.0686195558971829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,6144,0.048876444498697914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,7168,0.037520888778898455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,6144,0.03417244553565979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,5120,0.05735911263359917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,5120,0.04306133257018196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,5120,0.03105866577890184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,4096,0.04860888918240865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,4096,0.04633511106173197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,4096,0.035924444595972695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,4096,0.02646133303642273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,65536,0.5203030904134115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,3584,0.040765331851111516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,3584,0.03257066673702664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,16384,0.13749066988627115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,3584,0.02405333353413476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,3072,0.03554844525125291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,3072,0.029650668303171795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,12288,0.1066888901922438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,3072,0.022109333011839125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,2560,0.03028800090154012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,2560,0.026437333888477747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,10240,0.09197600020302667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,2048,0.0333315564526452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,2048,0.02583377725548214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,2048,0.023582221733199224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,2560,0.019668444991111755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,8192,0.07714133130179511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,1536,0.020390222469965618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,1536,0.020224000016848247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,2048,0.017135999268955655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,1536,0.01459200017982059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,1024,0.015634665886561077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,1024,0.016136889656384785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,7168,0.06983555687798394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,5120,0.055481778250800244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,768,0.012810666528013019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,1024,0.012763555679056378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,768,0.014840889308187695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,768,0.012079111403889127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,512,0.010047111246320937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,512,0.012936000194814471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,512,0.011390222443474663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,3584,0.04493066668510437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,256,0.007708444363541073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,256,0.012196444802814059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,3072,0.040652443965276085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,256,0.010728889041476779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,128,0.006592888798978593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,128,0.011728000309732227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,2560,0.03662044472164578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,64,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,64,0.011397333608733283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3584,32,0.007661333514584436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3584,32,0.011733333269755045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3584,128,0.010659555594126383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,65536,0.5504675441318089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,65536,0.45680798424614805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,1536,0.029166221618652344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,65536,0.24283554818895128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,16384,0.14190755950080022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,16384,0.09823111030790542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,1024,0.02533333334657881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,16384,0.06531733274459839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,12288,0.10718844334284465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,12288,0.07556000020768908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,512,0.021574222379260596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,768,0.02478222217824724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,12288,0.05023555623160469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,10240,0.10076266527175903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,10240,0.06430933210584852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,10240,0.04375999834802416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,8192,0.07992533180448744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,8192,0.05241511265436808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,256,0.020267556111017864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,8192,0.03789600067668491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,7168,0.0694755580690172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,7168,0.04668088754018148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,7168,0.03334133492575751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3584,128,0.018931556079122756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,6144,0.05966933568318685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,6144,0.058280891842312284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,6144,0.04126666651831733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,6144,0.031199110878838435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,5120,0.048595554298824735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,5120,0.036313778824276395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,65536,0.4833200242784288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,5120,0.028149333265092637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,4096,0.03932355509863959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,4096,0.03015822172164917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,4096,0.024173332585228816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,16384,0.12988177935282388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,3584,0.034832000732421875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,3584,0.028014222780863445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,3584,0.021899556120236714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,12288,0.1012951135635376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,3072,0.030079109801186457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,3072,0.025507556067572698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,3072,0.02074044446150462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,10240,0.08700088659922282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,2560,0.02570933269129859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,2560,0.02321333355373806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,8192,0.0737431115574307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,2560,0.018610666195551556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,2048,0.021408889028761122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,2048,0.020615999897321064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,2048,0.016161777906947665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,7168,0.06633155875735812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,1536,0.015885333220163982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,1536,0.017816000514560275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,5120,0.0525857773092058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,1024,0.024520888924598694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,1024,0.011752000285519494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,1536,0.013821333646774292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,1024,0.014134221606784396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,1024,0.011892444557613797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,768,0.010061333576838175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,768,0.013103110922707452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,4096,0.04556711183653938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,768,0.011184000306659274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,512,0.00833422193924586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,512,0.011752888560295105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,3584,0.04239555531077915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,512,0.010441777606805166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,256,0.0069715554515520734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,256,0.011045333411958484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,256,0.009993777506881291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,3072,0.03866666555404663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,128,0.0060159998635451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,128,0.010679999987284342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,3072,128,0.009694221946928237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,64,0.00556355549229516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,64,0.010647111468844943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,3072,32,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,3072,32,0.010456889039940305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,2560,0.03552177879545424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,2048,0.032059556908077665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,65536,0.4852088822258844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,65536,0.3845626778072781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,16384,0.12343733840518528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,16384,0.12365421983930801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,16384,0.09064177672068278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,65536,0.22848354445563424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,12288,0.09440355830722386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,12288,0.09375111262003581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,12288,0.07126844591564603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,16384,0.06037066380182902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,1536,0.028548445966508653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,10240,0.07960533433490329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,12288,0.047447111871507436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,10240,0.061208003097110324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,768,0.024352888266245525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,10240,0.04155022237035964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,8192,0.0635075569152832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,8192,0.050272888607449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,8192,0.03535644544495477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,7168,0.05557866891225179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,7168,0.04513422316975064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,7168,0.032279110617107816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,512,0.019892444213231403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,6144,0.04821155468622843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,6144,0.04016800059212579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,256,0.01936444474591149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,6144,0.029402666621738013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,5120,0.04884888728459676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,5120,0.040384888648986816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,5120,0.03595911131964789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,5120,0.02712000078625149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,3072,128,0.01809599995613098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,4096,0.03235822253757053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,4096,0.02980088856485155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,3584,0.03941866755485535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,3584,0.028635554843478735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,3584,0.027435556054115295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,4096,0.022846221923828125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,3584,0.0210524449745814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,3072,0.024895111719767254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,3072,0.02482844392458598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,65536,0.44087200694613987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,3072,0.01973511113060845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,2560,0.021009777983029682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,2560,0.02256177696916792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,2560,0.01755644381046295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,2048,0.02983377708329095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,2048,0.01719466679626041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,2048,0.019880889190567862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,10240,0.08031111293368869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,2048,0.015443555182880826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,1536,0.013853333062595792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,1536,0.017112889223628573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,1536,0.01297155519326528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,8192,0.06780622402826945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,1024,0.010633777413103314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,1024,0.013888888888888888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,7168,0.06150044335259331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,1024,0.01146666705608368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,768,0.008967111508051554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,768,0.01276266657643848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,6144,0.05530666642718845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,512,0.01924000018172794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,768,0.011007111105653973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,512,0.007585778004593319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,512,0.011424889167149862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,512,0.00999466660949919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,256,0.006244444598754247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,256,0.011038222246699862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,256,0.009407111340098912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,4096,0.04288888971010844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,128,0.005533333453867171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,128,0.010663111176755695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2560,128,0.009376000199053023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,64,0.0052168890833854675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,64,0.010371555884679159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2560,32,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2560,32,0.010685332947307162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,65536,0.418887111875746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,65536,0.376235564549764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,3072,0.036172442966037326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,65536,0.30277689297993976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,2560,0.033024887243906655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,65536,0.1813653310139974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,16384,0.09775288899739583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,16384,0.07337688737445407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,12288,0.0891653365559048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,12288,0.07420888874265882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,12288,0.05669333537419637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,16384,0.04931111137072245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,10240,0.07630578014585707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,10240,0.06195200151867337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,10240,0.048103110657797925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,12288,0.0396124455663893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,1536,0.02682755556371477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,8192,0.049908445941077344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,8192,0.04115733173158433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,10240,0.03500444359249539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,8192,0.03006311257680257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,7168,0.04369599951638115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,7168,0.0372995568646325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,1024,0.022602667411168415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,7168,0.027829332484139338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,6144,0.05238666799333361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,6144,0.037864890363481306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,6144,0.03251911203066508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,6144,0.02567200031545427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,5120,0.03180444571706984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,5120,0.02947910957866245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,5120,0.023711999257405598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,768,0.02237333357334137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,4096,0.026251556144820318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,4096,0.02442933287885454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,4096,0.01999022232161628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,256,0.018233777748213876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,3584,0.023285332653257582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,3584,0.022658665974934895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2560,128,0.01750666730933719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,3584,0.018579555882347953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,3072,0.020601777566803824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,3072,0.0209840006298489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,3072,0.017330666383107502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,16384,0.11379022068447536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,2560,0.01794399983353085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,2560,0.018792000081804063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,2560,0.015580443872345818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,2048,0.029141333368089464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,2048,0.015258666541841296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,2048,0.01684177749686771
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,8192,0.0646631121635437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,7168,0.05844089057710436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,1536,0.012669333153300814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,1536,0.014178666803571912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,2048,0.01349866638580958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,1536,0.011669333610269757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,1024,0.009749333063761393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,1024,0.011808888779746162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,5120,0.046557333734300405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,1024,0.010246222217877706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,768,0.00832266691658232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,768,0.011078221930397881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,4096,0.040876444843080305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,768,0.009651555783218807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,512,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,512,0.010018666585286459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,512,0.009022222624884712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,256,0.017247110605239868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,256,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,256,0.009648888475365108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,3584,0.038078222009870745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,256,0.00867022242810991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,128,0.005201777650250329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,3072,0.03497066762712266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,128,0.009696000152164036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,64,0.004576000074545543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,64,0.009300444689061906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,2048,32,0.004879111217127906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,2048,32,0.009359999663299983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,2048,128,0.00867555538813273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,65536,0.30829421679178876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,65536,0.30647023518880206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,2560,0.03167377909024557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,65536,0.18107910950978598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,16384,0.07796088854471843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,16384,0.06923200024498834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,16384,0.04825511243608263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,12288,0.07820977767308553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,12288,0.059756444560156934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,12288,0.05461333195368449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,12288,0.03872355487611558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,10240,0.06731288962894015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,10240,0.050013333559036255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,10240,0.04751555456055535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,10240,0.034318221939934626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,1536,0.025814221964942083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,8192,0.040072888135910034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,8192,0.039239999320771955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,1024,0.021324444148275588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,8192,0.0292231109407213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,7168,0.05216000146336026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,7168,0.03531466590033637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,7168,0.03537777728504605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,768,0.020302222834693063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,6144,0.03115288747681512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,6144,0.03206844462288751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,7168,0.026721777187453374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,5120,0.042103111743927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,5120,0.02567200031545427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,5120,0.028493334849675495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,6144,0.024661334024535284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,512,0.017584888471497428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,4096,0.02179733415444692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,4096,0.023920888702074688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,5120,0.02288888891537984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,4096,0.019632889164818656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,3584,0.01956533392270406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,3584,0.022275555464956496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,3584,0.01815733313560486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,3072,0.031650665733549334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,3072,0.017714665995703805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,3072,0.020424889193640817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,3072,0.016979555288950603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,2560,0.02942488922013177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,2560,0.015432889262835184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,2560,0.0184515549076928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,2048,128,0.016534222496880423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,2048,0.0270506673389011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,65536,0.3492017851935492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,2048,0.012845333251688214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,2048,0.016194666425387066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,1536,0.022964444425370958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,2560,0.015203555425008139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,1536,0.010382222632567087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,1536,0.014118222726715935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,2048,0.01312622262371911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,1024,0.008001777860853408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,1024,0.011759999725553723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,1536,0.011072888970375061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,1024,0.009540444446934594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,768,0.0069351109365622205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,768,0.011077333655622272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,768,0.009335999687512716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,512,0.016701334052615695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,16384,0.0988302230834961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,512,0.005929777605666055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,512,0.009979555176364051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,512,0.008997333546479544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,256,0.004929777648713854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,256,0.009661333428488838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,8192,0.05747555361853706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,256,0.008618666893906063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,128,0.004551110996140374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,128,0.009333333207501305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1536,128,0.008641777767075432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,64,0.004204444587230682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,64,0.009304888546466827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1536,32,0.004226666771703296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1536,32,0.009362666971153682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,6144,0.04703377683957418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,65536,0.21130132675170898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,4096,0.036904888020621404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,65536,0.19957510630289713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,65536,0.12656444973415798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,3584,0.03442400031619602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,16384,0.05452000101407369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,16384,0.048319998714658946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,16384,0.037490665912628174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,12288,0.04141955574353536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,12288,0.039707554711235896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,12288,0.030400888787375555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,10240,0.06131911277770996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,10240,0.03490399983194139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,10240,0.034620443979899086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,10240,0.027328888575236004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,1024,0.01959733333852556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,8192,0.027678221464157104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,8192,0.02931200133429633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,8192,0.0236435549126731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,768,0.02126755482620663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,7168,0.02421155571937561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,7168,0.026536888546413843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,7168,0.021719111336602107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,6144,0.043587555487950645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,6144,0.020954666866196528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,6144,0.02438844409253862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,6144,0.02030933399995168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,256,0.016885333591037326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,5120,0.018239999810854595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,5120,0.02197155521975623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,5120,0.01925244430700938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1536,128,0.016183111402723525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,4096,0.01533066729704539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,4096,0.018565333551830716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,65536,0.3191288842095269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,4096,0.016171556380059984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,3584,0.013732444081041547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,3584,0.01754844519827101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,3584,0.01497333414024777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,3072,0.0299635562631819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,3072,0.012250666817029318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,3072,0.016531555189026725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,3072,0.01423822177780999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,16384,0.09083822038438584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,12288,0.0715342230266995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,2560,0.011148444480366178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,2560,0.014898666077189975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,2560,0.012638222012254926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,2048,0.009471110999584198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,2048,0.012822222378518848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,2048,0.01072533345884747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,8192,0.05215200119548374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,1536,0.008014221986134848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,1536,0.011752888560295105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,1536,0.00942488925324546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,7168,0.0481644438372718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,1024,0.006550222221348021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,1024,0.009734222458468543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,1024,0.008621333373917473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,768,0.018903111418088276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,768,0.00554577757914861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,768,0.009340444372759925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,768,0.008301333420806462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,512,0.015490666031837463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,512,0.004899555610285865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,512,0.008618666893906063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,5120,0.039121776819229126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,512,0.007999999655617608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,256,0.0041857775714662345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,256,0.00831288927131229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,256,0.007947555846638149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,4096,0.03404711021317376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,128,0.003918222255176968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,128,0.00832711077398724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,1024,128,0.007705777883529663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,64,0.0035342222286595237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,64,0.00831822223133511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,1024,32,0.0035413333939181436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,1024,32,0.00833511104186376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,3584,0.03233866559134589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,65536,0.17256800333658853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,65536,0.18806221750047472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,65536,0.1316737731297811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,2560,0.027612444427278306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,16384,0.043931557072533496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,16384,0.047057777643203735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,16384,0.03740266627735562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,12288,0.06734844711091784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,12288,0.033364444971084595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,12288,0.038718223571777344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,2048,0.023932443724738225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,12288,0.029926220575968426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,10240,0.027828445037206013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,10240,0.03392533461252848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,10240,0.027110222313139174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,1536,0.020555555820465088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,8192,0.023060444328520033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,8192,0.028526223368114893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,8192,0.023295111126369897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,1024,0.017904000149832833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,7168,0.045588443676630654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,7168,0.02032711108525594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,7168,0.026092444856961567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,7168,0.021648888786633808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,6144,0.01790844400723775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,6144,0.023652444283167522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,6144,0.020231111182106864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,5120,0.036943111154768206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,5120,0.01555288831392924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,5120,0.021536888347731695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,256,0.016480889585283067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,5120,0.019151111443837483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,4096,0.013134222063753339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,4096,0.018259555101394653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,1024,128,0.015105777316623263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,4096,0.01607022186120351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,3584,0.012025777664449481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,3584,0.017190222938855488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,3584,0.01497511068979899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,65536,0.30085333188374835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,3072,0.011110222174061669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,3072,0.015844444433848064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,3072,0.014127110441525778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,16384,0.08500978019502427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,2560,0.009658666948477427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,2560,0.014436443646748861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,2560,0.012410666379663678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,2048,0.023392889234754775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,2048,0.008214222060309516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,2048,0.012859555582205454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,2048,0.01072177787621816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,10240,0.05850578016704983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,1536,0.0069226668112807805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,1536,0.011011555790901184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,1536,0.00941244430012173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,1024,0.017871111631393433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,1024,0.005924444645643234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,1024,0.009400889277458191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,1024,0.008348444269763099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,8192,0.05020799901750353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,768,0.0052737775776121355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,768,0.009345778160625035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,768,0.008310221963458592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,512,0.015835555063353646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,512,0.004552000098758274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,512,0.008539555801285638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,6144,0.041414221127827965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,512,0.007998222278224098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,256,0.0042239998777707415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,256,0.008336889247099558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,4096,0.032960888412263654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,256,0.007654222349325816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,128,0.0039013334446483185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,128,0.008314666648705801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,3584,0.030704001585642498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,64,0.003823111040724648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,64,0.008351999852392409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,768,128,0.007648000286685095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,768,32,0.0038515557017591265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,768,32,0.008350222474998897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,65536,0.11706399917602539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,65536,0.29536623424953884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,65536,0.18458933300442168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,3072,0.02808177802297804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,16384,0.036601778533723615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,65536,0.12394577927059597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,16384,0.04641511042912801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,16384,0.037201778756247625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,12288,0.027759111589855615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,12288,0.03817244370778402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,12288,0.02976800004641215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,2560,0.025648888614442613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,10240,0.02346399923165639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,10240,0.033377776543299355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,1536,0.02014400064945221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,10240,0.026517333255873785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,8192,0.019285332825448778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,8192,0.028212444649802312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,8192,0.0229751103454166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,768,0.018921777606010437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,7168,0.017525333497259352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,7168,0.02581600017017788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,7168,0.02092355489730835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,128,0.015463999576038785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,6144,0.015495111544926962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,6144,0.023345778385798138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,768,256,0.01586666703224182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,6144,0.019648000597953796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,5120,0.013515555196338229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,5120,0.021264889174037512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,12288,0.06492355797025892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,16384,0.08334400256474812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,5120,0.018236445056067575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,10240,0.05692355500327217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,4096,0.011965333587593503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,4096,0.017817777064111497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,4096,0.015805333852767944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,3584,0.010801778071456485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,3584,0.016805332567956712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,3584,0.014430221584108142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,3072,0.009630222287442949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,3072,0.015757333901193406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,3072,0.013449778159459433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,8192,0.04889422323968676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,2560,0.008605333666006723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,2560,0.014395554860432943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,2560,0.011680000358157687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,7168,0.0442826681666904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,2048,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,2048,0.012706666356987424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,2048,0.01035288886891471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,6144,0.04019999835226271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,1536,0.006312888943486744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,1536,0.010796444283591377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,1536,0.008976889153321585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,5120,0.03587822119394938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,1024,0.0054551110499435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,1024,0.009457777771684859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,1024,0.00831200016869439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,3584,0.029488888051774766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,768,0.004565333326657613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,768,0.009332444104883406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,768,0.007991111112965478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,512,0.014771555860837301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,512,0.0041697778635554844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,3072,0.0273644444015291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,512,0.008314666648705801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,4096,0.03196177879969279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,256,0.0038942222793896994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,512,0.007636444436179266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,256,0.008336889247099558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,256,0.007620444728268518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,128,0.003499555504984326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,128,0.00830844458606508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,2560,0.023638221952650283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,512,128,0.007480888730949826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,64,0.0032106666929192017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,64,0.008290666672918532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,512,32,0.0031591111587153543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,512,32,0.007954667011896769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,65536,0.06600444184409247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,65536,0.18245600329505074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,65536,0.12388622760772705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,2048,0.0214711113108529
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,16384,0.02237422267595927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,16384,0.04587022132343716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,16384,0.036531554328070745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,1536,0.018968888454967074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,12288,0.0206666671567493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,12288,0.037392887804243304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,12288,0.029567112525304157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,10240,0.05470488799942864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,10240,0.017208000024159748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,10240,0.03282311227586534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,10240,0.026484444737434387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,1024,0.016833777228991192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,8192,0.01456622282663981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,8192,0.02777066661251916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,8192,0.022662222385406494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,768,0.017847110827763874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,7168,0.013385777672131857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,7168,0.025220443805058796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,256,0.014840000205569796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,7168,0.020901333954599168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,6144,0.012090666426552666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,6144,0.02312711046801673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,6144,0.019512888458040025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,512,128,0.014150222142537435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,5120,0.010739555789364709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,5120,0.020798222886191476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,5120,0.018396443790859647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,65536,0.2801262272728814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,4096,0.009357333183288574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,4096,0.017441777719391715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,16384,0.07982399728563097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,4096,0.01567911108334859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,3584,0.008678221868144141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,3584,0.016523554921150208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,3584,0.014474666780895658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,3072,0.025871111287011042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,3072,0.007816000117195977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,3072,0.015383111106024848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,3072,0.013246222502655454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,12288,0.06345333655675252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,2560,0.006908444480763541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,2560,0.014119999276267158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,2560,0.01145155562294854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,8192,0.04679022232691447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,2048,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,2048,0.012090666426552666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,7168,0.04271022147602505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,2048,0.010674667027261523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,1536,0.00526311124364535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,1536,0.01072266697883606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,1536,0.009031111167536842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,6144,0.03875822159979079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,1024,0.004522666749027041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,1024,0.009345778160625035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,1024,0.008292444050312042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,5120,0.03424177898301019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,768,0.00424177779091729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,768,0.00902133352226681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,768,0.00794488853878445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,3584,0.028311110205120508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,512,0.003847111016511917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,512,0.008290666672918532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,4096,0.030645334058337744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,512,0.007654222349325816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,256,0.0031679999083280563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,256,0.008295999632941352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,256,0.007622222105662028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,2560,0.02204711072974735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,128,0.003168888803985384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,128,0.007958222594526079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,256,128,0.007315555380450354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,64,0.0031519999934567344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,64,0.00794222205877304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,256,32,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,256,32,0.008008889026112026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,2048,0.02052177819940779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,65536,0.05394577648904589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,65536,0.18158755037519667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,65536,0.12353689140743679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,1536,0.01846933364868164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,16384,0.018523555662896898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,16384,0.04557333389918009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,16384,0.03616266780429416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,768,0.01679999960793389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,12288,0.016179554992251925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,12288,0.037272890408833824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,12288,0.029366221692827012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,1024,0.015811555915408664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,10240,0.014651555154058667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,10240,0.03291022115283542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,10240,0.026220444175932143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,512,0.013784888717863293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,8192,0.014127110441525778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,8192,0.02758666707409753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,8192,0.022436444958051045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,128,0.01311466677321328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,7168,0.012708444562223224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,7168,0.02509333358870612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,7168,0.020968889196713764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,256,256,0.013776889277829064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,6144,0.011607999602953592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,6144,0.0229120006163915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,6144,0.01941155559486813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,65536,0.2769920031229655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,5120,0.010386666489972008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,5120,0.02085777786042955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,5120,0.018230222993426852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,16384,0.07862044705284967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,4096,0.008636444807052612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,4096,0.017535999417304993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,4096,0.015837333268589444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,3584,0.02754488918516371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,3584,0.007970666719807519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,3584,0.01650311052799225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,3584,0.014129777749379476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,12288,0.06227111154132419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,3072,0.007282666862010956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,3072,0.015237333046065437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,3072,0.01314311143424776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,10240,0.05342844459745619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,2560,0.006270222365856171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,2560,0.013438222308953604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,2560,0.011366222467687396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,8192,0.04633333285649618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,2048,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,2048,0.011952000359694162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,2048,0.01036177741156684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,6144,0.03799110982153151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,1536,0.004942222187916438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,7168,0.04201511210865445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,1536,0.010397333237859938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,1536,0.009039999710188972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,1024,0.0042444442709287005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,1024,0.009353777600659264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,1024,0.008316444853941599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,5120,0.03371822171741062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,768,0.00388355553150177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,768,0.008987555901209513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,768,0.007983999947706858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,4096,0.029337777031792536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,512,0.0035048888789282907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,512,0.008357333640257517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,512,0.007598222129874759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,3072,0.024300444457266066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,256,0.003206222214632564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,256,0.008023111356629265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,256,0.00758133331934611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,2048,0.019537778364287484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,128,0.00313688897424274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,128,0.008194666769769456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,2560,0.02199999988079071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,64,0.0028320000403457214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2048,128,128,0.007610666255156199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,64,0.008009778128729926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,128,32,0.0031439999325407874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,128,32,0.007956444389290279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,65536,0.049103998475604586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,16384,0.018602665927675035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,65536,0.18130844169192842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,16384,0.045511109961403735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,12288,0.015119110544522604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,12288,0.0372924440436893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,10240,0.01384622189733717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,10240,0.03270311156908671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,8192,0.01310755560795466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,8192,0.027441778116756018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,7168,0.011709333293967776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,7168,0.025231111380789015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,6144,0.010600888894663917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,6144,0.02291999922858344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,5120,0.010063110954231685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,5120,0.020593777298927307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,4096,0.008618666893906063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,4096,0.017530666457282174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,3584,0.007704888780911763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,3584,0.01645511057641771
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,3072,0.0068764446510208984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,3072,0.015232889188660515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,2560,0.005923555543025334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,2560,0.013711999687883588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,2048,0.005615111026499007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,2048,0.011681777735551199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,1536,0.0047031111187405055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,1536,0.010328888893127441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,1024,0.004181333300140169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,1024,0.009317333499590555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,768,0.0038408889538711975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,768,0.009005332986513773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,512,0.00351200004418691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,512,0.007981333467695449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,256,0.003136000078585413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,256,0.008325333396593729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,128,0.0031795555518733132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,128,0.007927999728255803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,64,0.0028204443968004654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,64,0.007992888490358988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,64,32,0.0028239999794297745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,64,32,0.007921777665615082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,65536,0.04808888832728068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,16384,0.016371554798550077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,65536,0.18033599853515625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,12288,0.013125333521101209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,16384,0.04522488845719231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,1536,0.01750577820671929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,12288,0.03699200020896064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,10240,0.012028444144460889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,10240,0.03256266646915012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,8192,0.013784888717863293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,8192,0.02740977704524994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,7168,0.01278222186697854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,7168,0.025030222203996446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,6144,0.011614222493436603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,5120,0.010999110837777456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,6144,0.02271466619438595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,1024,0.01566755606068505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,5120,0.020619556307792664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,4096,0.010004444254769219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,4096,0.01742044422361586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,3584,0.00867022242810991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,3584,0.016340444485346477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,3072,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,3072,0.01514488955338796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,2560,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,2560,0.01330311099688212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,2048,0.0051893335249688895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,2048,0.011718221836619906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,1536,0.00453955555955569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,1536,0.010273777776294285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,1024,0.0038951109680864546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,1024,0.009172444542249044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,768,0.0034959999223550162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,512,0.008363555702898238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,768,0.009032888544930352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,512,0.0033004445334275565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,128,0.0028213332924577924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,256,0.003177777760558658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,256,0.008208889100286696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,64,0.00793688909875022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,768,0.016188444362746347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,128,0.007978666987684038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,64,0.002518222149875429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2048,32,32,0.0028320000403457214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2048,32,32,0.008055110772450766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,256,0.012031111452314587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,512,0.013861333330472311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,16384,0.7687484423319498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,16384,0.8571466869778104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,16384,1.617985725402832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,12288,0.6462186707390679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,12288,1.162992901272244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,10240,0.5203359921773275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,10240,0.887754652235243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,10240,0.6451653374565972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2048,128,128,0.012702222499582501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,8192,0.43372533056471085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,8192,0.7531840006510416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,8192,0.4362400107913547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,7168,0.3819066683451335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,7168,0.7593537966410319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,12288,0.5735866758558485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,6144,0.5236808988783095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,6144,0.32421689563327366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,7168,0.3468204339345296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,6144,0.3054364522298177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,5120,0.4496453073289659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,16384,0.42569155163235134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,5120,0.29919200473361546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,4096,0.24955643547905815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,12288,0.3249146673414442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,4096,0.3573119905259874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,4096,0.2324746714697944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,10240,0.2747440073225233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,8192,0.22528089417351616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,3584,0.32124267684088814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,3584,0.20701777935028076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,3072,0.17234310838911268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,3072,0.2751893202463786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,3072,0.17716888586680093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,7168,0.20127377245161268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,5120,0.2570071220397949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,2560,0.23312534226311576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,2560,0.1514728864034017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,3584,0.1918977763917711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,6144,0.17407822608947754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,2048,0.18986311223771837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,2048,0.12730133533477783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,2560,0.15215822060902914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,5120,0.1497759951485528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,1536,0.14700622028774685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,1536,0.10365600056118435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,4096,0.12569422192043728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,1024,0.10044266780217488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,1024,0.10337866677178277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,1024,0.07771466837988959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,3584,0.11077155669530232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,768,0.06230666902330187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,768,0.0793057746357388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,768,0.06922222508324517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,3072,0.09791111283832127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,2048,0.13424621687995061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,512,0.05822755230797661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,512,0.06258666515350342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,2560,0.08624177508884007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,1536,0.11657599608103435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,256,0.039782222774293684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,256,0.05790755483839247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,512,0.08193600177764893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,2048,0.07213244173261854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,128,0.031930665175120033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,128,0.055608888467152916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,1536,0.05740444527732002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,64,0.027385777897304956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,64,0.055016001065572105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,65536,32,0.02749866743882497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,65536,32,0.05502133236991035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,65536,0.8330160246955024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,1024,0.045647998650868736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,65536,1.6869458092583551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,256,0.04595999916394552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,65536,0.8868471251593696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,16384,0.49662044313218856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,16384,0.22067911095089385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,768,0.04137511054674784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,512,0.037245333194732666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,12288,0.2835235595703125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,12288,0.16679822074042427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,65536,0.4806399875217014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,16384,0.12065333790249294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,65536,128,0.042345778809653394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,10240,0.14416444301605225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,10240,0.24666044447157118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,256,0.03449777762095133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,8192,0.1968017816543579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,8192,0.11677955256568061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,12288,0.09188177850511338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,10240,0.07678844531377156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,7168,0.17206399970584443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,7168,0.10298222303390503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,16384,0.212774223751492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,6144,0.08834666675991482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,12288,0.16280799441867405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,6144,0.14720000161064997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,6144,0.08960533142089844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,65536,128,0.03389066788885329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,5120,0.1218328873316447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,8192,0.06451466348436144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,5120,0.07569333579805163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,4096,0.06764266888300578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,4096,0.0993173321088155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,4096,0.06293511390686035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,7168,0.055997331937154136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,10240,0.1366879940032959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,3584,0.08677599827448527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,3584,0.055995557043287486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,6144,0.04924089047643873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,8192,0.1134622229470147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,3072,0.07601688967810737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,3072,0.04970044559902615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,5120,0.042596442831887134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,7168,0.09984266757965088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,2560,0.06292444467544556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,2560,0.041839109526740186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,4096,0.03701511025428772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,3584,0.033737778663635254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,2048,0.05128355489836799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,2048,0.035467555125554405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,3072,0.029318223396937054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,5120,0.07848888635635376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,1536,0.039247999588648476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,1536,0.030528889762030706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,2560,0.027240888939963445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,1024,0.03661511010593838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,1024,0.026829333768950567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,1024,0.024040000306235418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,2048,0.02408711115519206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,768,0.029721776644388836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,768,0.021122665868865118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,768,0.02257066633966234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,3584,0.062463998794555664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,1536,0.01980977753798167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,512,0.015819556183285184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,512,0.02049777739577823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,3072,0.05745244688457913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,2560,0.052264889081319175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,256,0.013063111239009433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,256,0.016753777861595154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,1024,0.017097777790493436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,128,0.02290488945113288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,128,0.010771555205186209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,128,0.015442666080262927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,2048,0.047629333204693265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,64,0.008975110948085785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,64,0.015096000499195524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,16384,32,0.009284444153308868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,768,0.01567555632856157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,16384,32,0.015472888946533203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,65536,0.6377333535088433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,65536,1.2159022225273979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,512,0.01458666721979777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,16384,0.1662800047132704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,1536,0.04213066564665901
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,16384,0.28758133782280815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,16384,0.17380711767408583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,65536,0.680116441514757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,256,0.013758222262064615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,12288,0.23519823286268446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,12288,0.13003467188941106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,10240,0.10728622145122951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,16384,128,0.01349955548842748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,10240,0.18245066536797416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,10240,0.10988710986243354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,8192,0.08893599775102402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,8192,0.14668444792429605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,8192,0.08902133173412746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,65536,0.37857688797844785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,16384,0.09484799702962239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,7168,0.12898843818240696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,12288,0.07304977708392672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,7168,0.07900355259577434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,10240,0.0613937775293986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,6144,0.11003022061453925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,6144,0.0682871143023173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,8192,0.05145333210627238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,512,0.03173866536882188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,5120,0.09275999996397231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,5120,0.059044447210099965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,16384,256,0.025462221768167283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,7168,0.04548799991607666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,4096,0.07441777653164335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,4096,0.048171556658214994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,6144,0.040700445572535195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,3584,0.0521342224544949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,3584,0.06503466765085857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,3584,0.04264000058174133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,5120,0.03574577636188931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,12288,0.1284559965133667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,3072,0.0598542226685418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,4096,0.0314026673634847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,3072,0.03814311159981622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,3584,0.02870755394299825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,2560,0.047304888566335045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,2560,0.032580445210138954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,7168,0.08050222529305352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,2048,0.040702223777770996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,2048,0.038804445001814104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,2048,0.02840533190303379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,6144,0.0733226670159234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,3072,0.0258586671617296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,1536,0.029714667134814795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,1536,0.02424888809521993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,2560,0.023592889308929443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,5120,0.06378222174114652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,1024,0.020782222350438435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,1024,0.01995377739270528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,4096,0.05629511011971367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,2048,0.020670221911536325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,768,0.01738133364253574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,768,0.01886044442653656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,1536,0.01717866626050737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,1024,0.014667555689811707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,512,0.01294311136007309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,512,0.0158933334880405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,3072,0.0482666658030616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,768,0.01405244403415256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,256,0.012427555190192329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,256,0.013810666898886362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,2560,0.04389333393838671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,512,0.013154666456911298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,128,0.01035377797153261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,128,0.013064888616402945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,256,0.012359999948077731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,64,0.009961778091059791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,64,0.012763555679056378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,12288,32,0.00794488853878445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,12288,32,0.013094222380055321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,1536,0.03644799855020311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,1024,0.03154933452606201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,12288,128,0.01202222208182017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,65536,1.0777511596679688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,65536,0.5695119963751899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,16384,0.25677598847283256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,16384,0.15147022406260172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,768,0.027099554737408955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,512,0.027437332603666518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,12288,0.19338044855329725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,65536,0.3262106577555339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,12288,0.11081244548161824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,16384,0.08160710997051664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,10240,0.17353778415256074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,256,0.023327999644809302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,10240,0.09290044175253974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,8192,0.08262222343020968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,8192,0.13900178008609348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,8192,0.07484533389409383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,12288,128,0.02125777800877889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,12288,0.06327911218007405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,7168,0.12233332792917888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,7168,0.0669777790705363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,10240,0.05322133170233833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,6144,0.06696799728605482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,6144,0.1038408875465393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,6144,0.05883555279837715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,8192,0.04522488845719231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,5120,0.059670223130120166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,5120,0.08574577834871079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,5120,0.049472000863817006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,7168,0.03974666529231601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,65536,0.5669182141621908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,4096,0.0675999985800849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,4096,0.04903466502825419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,6144,0.03569511241383023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,5120,0.031932443380355835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,3584,0.059692442417144775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,16384,0.1494551102320353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,3584,0.0362062222427792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,12288,0.11488266785939534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,4096,0.028185778194003638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,3072,0.05094044407208761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,3072,0.033193777004877724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,10240,0.09800177812576294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,2560,0.04285688863860237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,2560,0.028547555208206177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,3584,0.025648888614442613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,3072,0.023119111855824787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,2048,0.034619556532965765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,2048,0.025171554750866357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,7168,0.07413866784837511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,1536,0.034120000070995755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,2560,0.021336888273557026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,1536,0.02658933401107788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,1536,0.021684444612926904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,2048,0.018912000788582694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,1024,0.0184551113181644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,1024,0.017831999394628737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,1536,0.015461333923869662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,768,0.0262755552927653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,768,0.01516177753607432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,4096,0.05285244517856174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,768,0.016248888439602323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,3584,0.04900177650981479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,512,0.012004444168673621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,1024,0.01344622257683012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,512,0.014118222726715935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,256,0.022544888986481562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,3072,0.04524622360865275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,256,0.008811555802822113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,256,0.01239377756913503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,768,0.01290844463639789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,128,0.007627555893527136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,2560,0.04212355613708496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,128,0.011753777662913004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,64,0.008649778034951951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,64,0.011752888560295105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,10240,32,0.008656889200210571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,10240,32,0.011737777955002256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,512,0.011737777955002256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,256,0.011047999891969891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,65536,0.7594826486375598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,2048,0.03831288880772061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,65536,0.4780764579772949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,16384,0.12108443842993842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,16384,0.19848977194892037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,16384,0.11872977680630153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,12288,0.09576355748706394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,10240,128,0.011053333679835001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,12288,0.14199911223517522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,12288,0.09028266535864936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,10240,0.08164266745249431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,1024,0.03013066781891717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,10240,0.12271200286017524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,10240,0.0769066678153144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,65536,0.2654942141638862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,8192,0.09878400299284194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,16384,0.06788355774349637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,8192,0.061962664127349854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,7168,0.06271911329693265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,7168,0.08702311250898574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,7168,0.05480533175998264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,12288,0.052842666705449425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,10240,0.0451822214656406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,6144,0.07447910971111722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,6144,0.04790311058362325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,512,0.025071110990312364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,8192,0.038584887981414795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,5120,0.061971558464898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,5120,0.040757334894604154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,7168,0.03403644429312812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,6144,0.03128533230887519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,4096,0.05015200045373705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,4096,0.03461422191725837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,10240,128,0.020571556356218126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,5120,0.027801778581407335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,3584,0.04400355617205302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,3584,0.030276444223192003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,4096,0.02466222147146861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,65536,0.46490224202473956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,3584,0.022629333866967097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,3072,0.03844177722930908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,3072,0.027444443768925134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,2560,0.03615466753641764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,2560,0.03195555673705207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,2560,0.024749333659807842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,8192,0.06953155332141452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,3072,0.020216888851589628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,2048,0.02664533257484436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,2048,0.022149332695537146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,2560,0.018882667024930317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,6144,0.0569288863076104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,1536,0.020920889245139226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,1536,0.01903377804491255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,2048,0.01649688846535153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,5120,0.05071555574735006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,1024,0.015450666348139444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,1024,0.015605333778593274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,4096,0.04536711176236471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,768,0.024547555380397375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,768,0.012961777547995249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,768,0.013797333670987023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,1536,0.014142221874660917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,512,0.02335022224320306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,512,0.009983110758993361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,512,0.012106666962305704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,1024,0.012097777591811286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,3584,0.0422364440229204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,256,0.0075502221783002215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,256,0.011031111081441244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,768,0.01164177805185318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,3072,0.03972266780005561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,128,0.00794488853878445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,128,0.010543111297819348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,512,0.010688888529936472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,64,0.007544889218277401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,64,0.010379555324713389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,8192,32,0.007344000041484833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,8192,32,0.010728889041476779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,256,0.009884444375832876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,65536,0.6727324591742622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,8192,128,0.010041777458455827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,2048,0.03340355555216471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,65536,0.4759955406188965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,16384,0.19964267147911918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,16384,0.10947999689314102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,1536,0.029862221744325426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,12288,0.09132533603244358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,12288,0.1289457745022244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,12288,0.08369333214230007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,65536,0.24893334176805285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,16384,0.06405511167314318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,10240,0.07899110847049289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,10240,0.11677955256568061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,10240,0.07068533367580838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,1024,0.026428444517983332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,8192,0.09469955497317845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,8192,0.05747199720806546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,12288,0.050642665889528066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,7168,0.06183466646406385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,7168,0.08259644773271349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,7168,0.05109955535994636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,10240,0.04246844516860115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,8192,0.0365626679526435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,256,0.021937777598698933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,6144,0.07087822092903985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,6144,0.04488089018397861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,5120,0.04932889011171129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,5120,0.05957688887914022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,7168,0.03278755479388767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,5120,0.03853333327505324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,4096,0.04414044486151802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,4096,0.047891553905275136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,4096,0.032403555181291364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,8192,128,0.020107555720541213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,6144,0.029368887344996136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,3584,0.04203822215398153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,3584,0.02973955538537767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,5120,0.026712889472643535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,4096,0.0232595553000768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,3072,0.036512000693215266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,3072,0.026872888207435608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,65536,0.43049245410495335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,16384,0.11643732918633355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,3584,0.02158044444190131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,2560,0.03089511063363817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,2560,0.02405066622628106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,3072,0.019558222757445443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,2048,0.02572800053490533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,2048,0.021240888370407954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,2560,0.0181013329161538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,2048,0.015495111544926962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,1536,0.020384889509942796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,1536,0.018351111147138808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,8192,0.06752000252405803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,1024,0.02569955587387085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,1024,0.015847111741701763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,1024,0.014859555496109856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,1536,0.013417777915795645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,6144,0.055454221036699086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,768,0.013062222136391534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,768,0.013456000222100152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,1024,0.011805333197116852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,768,0.010987555815113915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,512,0.010245333115259806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,512,0.011800000237094032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,3072,0.03826755616399977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,3584,0.04109777675734626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,256,0.007657777931955125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,256,0.011061333119869232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,512,0.01034400032626258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,2560,0.03522133496072557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,128,0.006424888968467712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,128,0.010594666832023196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,256,0.009653333160612319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,64,0.007328000333574083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,64,0.010112000008424124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,2048,0.03220977716975742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,7168,32,0.010392000277837118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,7168,32,0.01036444471942054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,65536,0.5739102363586426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,7168,128,0.009645333720578088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,1536,0.029099557134840224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,65536,0.3668524424235026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,16384,0.15581156147850886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,16384,0.10602400037977432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,65536,0.21572889222039116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,16384,0.055439998706181846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,12288,0.10944977733823989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,12288,0.07066844569312201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,768,0.024319999747806128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,10240,0.07321244478225708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,10240,0.10232000218497382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,10240,0.06007377968894111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,512,0.022655111220147874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,12288,0.04376711116896736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,8192,0.08149333132637872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,8192,0.0488319993019104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,10240,0.03660444418589274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,256,0.021249777740902368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,8192,0.03288977675967746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,7168,0.07127555873658922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,7168,0.04337866769896614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,7168,128,0.019490665859646268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,6144,0.06044266621271769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,6144,0.037575112448798284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,7168,0.02880711025661892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,6144,0.02605155607064565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,5120,0.05052444338798523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,5120,0.032497776879204646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,16384,0.10945244630177815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,65536,0.3925360043843587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,5120,0.023716444770495098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,4096,0.041174223025639854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,4096,0.02734577821360694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,12288,0.08553866545359294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,3584,0.035939554373423256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,3584,0.02533688810136583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,4096,0.020984888076782227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,8192,0.06283199787139893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,3072,0.030642665094799463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,3072,0.02293955617480808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,3584,0.01958044370015462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,7168,0.05749422311782837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,2560,0.02659822172588772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,3072,0.01774044500456916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,2560,0.020941333638297185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,6144,0.052044444613986544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,2048,0.021873777111371357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,2048,0.018410666121376883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,2560,0.01646311084429423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,1536,0.027781334188249376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,1536,0.015881778465376962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,1536,0.016013332539134555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,5120,0.04708266589376661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,2048,0.014688000082969666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,1024,0.013423110875818463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,1024,0.012704888979593912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,1536,0.012434666355450949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,1024,0.010979555547237396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,4096,0.04160799913936191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,768,0.011384888655609555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,768,0.011700444751315646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,3584,0.039048890272776283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,512,0.00906133320596483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,512,0.01033866653839747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,768,0.010038221875826517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,256,0.020621332857343886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,256,0.007314666277832455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,256,0.009665778113736046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,512,0.009689778089523315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,3072,0.036122666464911565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,128,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,128,0.009824000298976898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,256,0.009003555609120263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,64,0.006577777779764599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,64,0.010048888623714447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,6144,32,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,6144,32,0.010291555689440833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,2560,0.03320088982582092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,65536,0.34890932506985134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,6144,128,0.008624888956546783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,65536,0.4964710871378581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,65536,0.33846222029791934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,16384,0.12722044520907932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,2048,0.030766222212049697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,16384,0.13077510727776423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,16384,0.0837777786784702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,12288,0.09597155782911514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,12288,0.10506133238474528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,12288,0.06399199697706434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,10240,0.06769955820507474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,10240,0.08003733555475871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,10240,0.0543431109852261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,65536,0.19611822234259713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,16384,0.05288088983959622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,8192,0.06470844480726454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,8192,0.044734223021401294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,1024,0.0244159996509552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,12288,0.042174223396513194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,7168,0.05597066879272461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,7168,0.040624890062544085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,10240,0.035624000761244036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,6144,0.04786133435037401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,6144,0.048635555638207324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,6144,0.035649776458740234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,8192,0.03098755412631565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,5120,0.042704890171686806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,768,0.02363555630048116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,5120,0.04082489013671875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,5120,0.03176888823509216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,7168,0.027689778142505225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,4096,0.03240088952912225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,4096,0.027013333307372198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,6144,0.02516888909869724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,5120,0.022688888841205176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,3584,0.028980443874994915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,3584,0.024657777614063684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,4096,0.020384889509942796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,512,0.021073778470357258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,3072,0.024877332978778418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,3072,0.02255022194650438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,3584,0.018739556272824604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,3072,0.017066667477289837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,2560,0.02108977735042572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,2560,0.02032977839310964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,6144,128,0.018224888377719455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,2048,0.028613332245084975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,2048,0.017442666822009616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,2048,0.018234666850831773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,2560,0.015571556157535978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,2048,0.013784888717863293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,1536,0.013982221484184265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,1536,0.015787555111779105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,1536,0.011740444435013665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,8192,0.05714755588107639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,1024,0.01033866653839747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,7168,0.05276266733805338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,1024,0.012452444268597497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,1024,0.010670222342014313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,768,0.008926221893893348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,768,0.011334222224023608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,4096,0.03828800055715773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,512,0.019380443625979953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,512,0.0075502221783002215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,512,0.010355555348926121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,768,0.009860444400045607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,3584,0.03570400012864007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,256,0.01811733345190684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,256,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,256,0.009740444521109263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,512,0.008956444760163626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,128,0.005426666802830166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,128,0.009249777429633671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,3072,0.03316088848643833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,64,0.005416000054942236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,64,0.009030222064918941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,5120,32,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,5120,32,0.009012444151772393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,256,0.008385777473449707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,2560,0.03051555487844679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,5120,128,0.008332444561852349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,65536,0.38611022631327313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,65536,0.2653955618540446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,16384,0.09911555714077419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,16384,0.06935644149780273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,65536,0.15266756216684976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,16384,0.04218044545915392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,12288,0.07506222195095487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,12288,0.050556444459491305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,1536,0.025731555289692346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,12288,0.033598221010631986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,10240,0.06310666932000054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,10240,0.043064001533720225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,10240,0.028740445772806805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,8192,0.05372533202171326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,8192,0.05116088853942024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,8192,0.03537333342764113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,768,0.02228533393806881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,1024,0.02307555576165517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,7168,0.04424444503254361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,7168,0.03126577867401971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,8192,0.025578666064474318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,7168,0.02279377811484867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,6144,0.038196444511413574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,6144,0.028237332900365193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,5120,128,0.01683999929163191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,5120,0.040595554643207125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,6144,0.021009777983029682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,5120,0.03208622336387634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,5120,0.025102222959200542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,5120,0.019522666931152344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,4096,0.026519111461109583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,4096,0.022151110900772944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,65536,0.3256293402777778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,16384,0.09011911021338569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,4096,0.01734844512409634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,3584,0.023710222707854375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,3584,0.020561777883105807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,12288,0.07187733385297987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,3072,0.020522667302025687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,3072,0.01819644371668498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,3584,0.015896888242827523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,3072,0.014767999450365702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,2560,0.01812088820669386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,10240,0.06261955367194282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,2560,0.016574222180578444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,2560,0.013736888766288757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,2048,0.015488889482286243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,2048,0.014860444598727755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,2048,0.01239466667175293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,1536,0.024655110306209985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,1536,0.013087111214796701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,1536,0.012375999655988483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,7168,0.0492888887723287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,6144,0.044532444741990834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,1024,0.00999022192425198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,1024,0.010671999719407825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,1536,0.010375111467308467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,768,0.021290666527218286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,768,0.007991111112965478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,768,0.009581333233250512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,1024,0.008648888932334052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,768,0.008348444269763099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,512,0.006907555378145642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,512,0.008650666309727563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,4096,0.03626222080654568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,512,0.007983999947706858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,256,0.0058373332851462895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,256,0.008271999657154083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,3072,0.03148622314135233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,3584,0.033872889147864446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,128,0.004931555440028508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,128,0.008041777544551426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,256,0.007628444168302748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,64,0.004883555488453971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,64,0.007986666427718269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,4096,32,0.004863111095296012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,4096,32,0.008003555238246918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,65536,0.26942398813035756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,2560,0.029250668154822454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,4096,128,0.007631111476156447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,65536,0.3592231008741591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,65536,0.24845332569546172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,2048,0.027067555321587458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,16384,0.09288266632292007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,16384,0.06115200122197469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,65536,0.1466995610131158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,12288,0.07020799981223212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,12288,0.04726399978001913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,16384,0.043014221721225314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,10240,0.057526224189334445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,10240,0.05836533175574409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,10240,0.04157955447832743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,12288,0.03346044487423367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,10240,0.028416888581381902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,8192,0.04745599958631727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,8192,0.034161776304244995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,1024,0.021440888444582622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,8192,0.025447110335032146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,7168,0.04191644324196709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,7168,0.030960000223583643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,512,0.017101332545280457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,7168,0.022328888376553852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,6144,0.03628977802064683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,6144,0.027841776609420776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,6144,0.020614221692085266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,256,0.01684888866212633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,5120,0.030512889226277668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,5120,0.025011556016074285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,4096,128,0.016470222009552848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,4096,0.03330755564901564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,4096,0.025436444414986506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,5120,0.01902666687965393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,4096,0.021419554948806763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,4096,0.017210667332013447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,3584,0.022795556320084467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,3584,0.019925332731670804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,16384,0.07875377602047391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,12288,0.06351911359363131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,3584,0.0158897770775689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,3072,0.020358221398459542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,3072,0.018235555953449674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,3072,0.014465777410401238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,2560,0.01812622282240126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,2560,0.016620443926917184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,2560,0.013337777720557319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,8192,0.04832444588343302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,2048,0.015495111544926962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,2048,0.014295111099878946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,7168,0.04446399874157376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,1536,0.023191110955344305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,1536,0.012816000315878125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,1536,0.011981333295504252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,2048,0.012102222277058495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,1536,0.010095111197895473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,1024,0.010089777410030365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,1024,0.010398222340477837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,6144,0.04032977753215366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,768,0.02055999967787001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,768,0.008294222255547842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,768,0.009330666727489894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,1024,0.008966222405433655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,768,0.008095111283991072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,512,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,512,0.008678221868144141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,5120,0.03700088792377048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,256,0.016541333662139047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,256,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,256,0.00832711077398724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,512,0.00793333351612091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,3584,0.031410667631361223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,128,0.005223999834722943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,128,0.00794933322403166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,256,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,64,0.004902222090297275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,64,0.008268444074524773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3584,32,0.005573333137565189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3584,32,0.007974222302436829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3584,128,0.0075777777367168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,65536,0.3091919951968723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,3072,0.029151999288135107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,65536,0.2317582236395942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,65536,0.1445280048582289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,16384,0.07965599828296237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,2560,0.027435556054115295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,16384,0.06003733476003011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,12288,0.06042222181955973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,12288,0.05973066886266073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,12288,0.046711110406451754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,16384,0.041788445578681103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,2048,0.025791999366548326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,12288,0.033306668202082314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,10240,0.05010933346218533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,10240,0.040476444694730974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,8192,0.04602666695912679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,8192,0.040694223509894475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,8192,0.034324444002575345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,10240,0.02796444296836853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,7168,0.04262666569815742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,7168,0.03548622131347656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,7168,0.030921777089436848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,8192,0.025200888514518738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,1024,0.020647111866209242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,6144,0.030974222554100886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,6144,0.02757866680622101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,7168,0.021959111094474792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,6144,0.020299555526839364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,5120,0.026224000586403742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,512,0.01721600029203627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,5120,0.02458933326933119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,4096,0.031776888502968684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,4096,0.021480000681347314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,4096,0.021243555678261652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,5120,0.018422222799725003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,3584,0.03008177876472473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,3584,0.018963555494944256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,3584,0.01961599952644772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,4096,0.016872889465755887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3584,128,0.016203555795881484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,3072,0.017108443710539076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,3584,0.015578667322794596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,3072,0.0177831103404363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,2560,0.026055999928050574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,2560,0.014950222439236112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,2560,0.016192888220151264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,3072,0.01418133411142561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,2048,0.024320888850424025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,2048,0.012896888785892062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,2560,0.0129457778400845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,2048,0.014081777797804939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,65536,0.2537155681186252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,1536,0.010279110736317104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,1536,0.011976000335481433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,16384,0.07551378011703491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,1024,0.018544000056054857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,1024,0.007975111405054728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,2048,0.011339555184046427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,1024,0.010025777750545079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,768,0.019673777951134574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,768,0.00684799998998642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,768,0.009343999955389235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,1536,0.009648000200589498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,512,0.016862221890025668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,512,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,512,0.008452444440788692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,1024,0.008296888735559251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,768,0.008295999632941352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,256,0.004881777697139316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,256,0.007994666695594788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,512,0.00776533368561003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,128,0.015816888875431485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,128,0.004600000050332812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,128,0.007983999947706858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,256,0.007612444460391998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,64,0.004168888760937585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,64,0.007983110845088959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,3072,32,0.004209777961174647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,3072,32,0.007975999679830339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,3072,128,0.007455999652544658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,10240,0.05289333396487766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,65536,0.23339554998609754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,65536,0.25381244553460014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,65536,0.22351733843485513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,16384,0.06376800272199842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,16384,0.059064891603257924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,65536,0.14202222559187147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,16384,0.04006400042110019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,12288,0.048262221945656665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,6144,0.03883644607332017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,12288,0.04565866788228353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,10240,0.049363556835386485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,10240,0.04058666692839728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,10240,0.03963022099600898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,12288,0.03237422307332357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,10240,0.02753244505988227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,8192,0.03273155623012119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,8192,0.03336266676584879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,5120,0.03542044427659776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,7168,0.03985066546334161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,7168,0.028767999675538804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,7168,0.030268443955315485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,8192,0.024449777272012498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,7168,0.021370665894614324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,6144,0.025007999605602686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,6144,0.02720622221628825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,3072,0.027810666296217177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,6144,0.019831111033757527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,5120,0.020969778299331665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,5120,0.024350222614076402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,1536,0.02164622147878011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,5120,0.01797777745458815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,4096,0.017222222354676988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,4096,0.02073777715365092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,3072,256,0.016187555260128446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,3584,0.028079999817742243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,3584,0.015601777368121676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,4096,0.016301333904266357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,3584,0.01886488828394148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,3584,0.015296889675988091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,3072,0.013801777528391944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,3072,0.017621333400408428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,3072,0.013682666752073499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,2560,0.024672889047198828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,2560,0.012073777616024017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,2560,0.016012445092201233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,12288,0.056573331356048584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,16384,0.06939999924765693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,2048,0.02344622214635213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,2048,0.010695999695195092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,2048,0.013740444348918067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,2560,0.012331555287043253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,1536,0.008678221868144141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,1536,0.011704000333944956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,2048,0.01036711119943195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,1536,0.00960177762640847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,1024,0.007253333098358578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,1024,0.009712888962692684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,8192,0.04324711031383938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,1024,0.008284444610277811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,768,0.006259555617968242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,768,0.009196444518036312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,6144,0.03667466508017646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,768,0.007947555846638149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,512,0.005254222287072076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,512,0.008352888955010308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,512,0.007598222129874759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,5120,0.03296711047490438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,256,0.004576888763242298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,256,0.008361777497662438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,256,0.007333333293596904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,4096,0.03012266755104065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,128,0.0041751112374994485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,128,0.007977777885066139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,3072,0.026339555780092876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,64,0.0038968887594011095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,64,0.007932444413503012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2560,32,0.0041706665522522396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2560,32,0.007659555309348636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2560,128,0.007342221836249034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,65536,0.20926755004458955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,65536,0.15424088637034097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,65536,0.09655021958880955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,1536,0.020794666475719877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,16384,0.05407199925846524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,16384,0.039701332648595176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,16384,0.029872000217437744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,768,0.01868888901339637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,12288,0.04088088870048523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,12288,0.03261422117551168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,1024,0.018239999810854595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,12288,0.02457422183619605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,10240,0.034254223108291626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,10240,0.028488000233968098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,10240,0.021020443903075323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,512,0.015791111522250705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,8192,0.027199111051029626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,8192,0.02351555559370253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,8192,0.018914666440751817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,256,0.016532444291644625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,7168,0.02418577836619483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,7168,0.021431111627154883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,7168,0.01676266723208957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2560,128,0.01550222271018558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,6144,0.02123911182085673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,6144,0.019893333315849304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,6144,0.0158924443854226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,65536,0.21954133775499132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,5120,0.018251554833518136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,5120,0.018219555417696636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,5120,0.014706666270891825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,16384,0.06651644574271308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,4096,0.01516266663869222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,4096,0.015607999430762397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,12288,0.05407999952634176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,4096,0.013253333667914072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,3584,0.013759111364682516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,3584,0.014496000276671516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,3584,0.012262221839692859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,10240,0.04684977730115255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,3072,0.012517333030700684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,3072,0.013798221945762634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,8192,0.04143288731575012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,3072,0.011076444553004371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,2560,0.011119999819331698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,2560,0.012083555261294046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,7168,0.03819555706448025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,2560,0.00998133338159985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,2048,0.020482665962643094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,2048,0.00979377743270662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,2048,0.010684444672531553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,2048,0.008976000050703684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,6144,0.034965333011415266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,1536,0.00833333366447025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,1536,0.009311999711725447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,1024,0.01681422193845113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,1024,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,1024,0.00832266691658232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,1536,0.007954667011896769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,768,0.018814222680197824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,768,0.005496888938877318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,768,0.007985778152942657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,1024,0.007318221860461765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,512,0.014840889308187695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,512,0.004902222090297275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,512,0.007185777856243982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,768,0.006968888971540663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,512,0.0069617778062820435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,256,0.004200889004601372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,256,0.006996444529957241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,256,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,5120,0.03155644403563605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,128,0.003891555385457145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,128,0.006959999601046245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,2048,128,0.006656888872385025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,64,0.003654222107595868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,64,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,2048,32,0.003612444632583194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,2048,32,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,4096,0.02826489011446635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,65536,0.16916799545288086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,65536,0.13977332909901938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,65536,0.09473244349161784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,16384,0.059466666645473905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,16384,0.04362666606903076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,3584,0.027388445205158655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,16384,0.038133333126703896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,16384,0.029206222958034937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,12288,0.03299644589424133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,12288,0.030498666895760432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,12288,0.023903111616770428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,10240,0.043488889932632446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,10240,0.02755733331044515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,10240,0.027087110612127516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,3072,0.025384000606007043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,2560,0.02348622183005015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,8192,0.022375999225510493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,8192,0.02325422234005398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,10240,0.020601777566803824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,8192,0.018568888306617737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,7168,0.019848888119061787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,7168,0.021278222401936848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,7168,0.01647111111217075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,1536,0.018932445181740653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,6144,0.01706755492422316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,6144,0.019142222073343065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,6144,0.015489778584904142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,256,0.016186666157510545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,5120,0.014878221684032016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,5120,0.017278222574128043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,5120,0.014303111367755465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,2048,128,0.0155102229780621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,4096,0.013144888811641268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,4096,0.015043555034531487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,4096,0.012724444270133972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,65536,0.19870132870144316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,3584,0.01181422256761127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,3584,0.014214222629865011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,3584,0.012040889097584618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,12288,0.048454221751954824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,3072,0.010863111250930362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,3072,0.012671111358536614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,3072,0.010951999988820819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,7168,0.03542044427659776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,2560,0.009450666606426239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,2560,0.011372444530328115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,8192,0.038126223617129855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,2048,0.020015110572179157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,2560,0.00980711148844825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,2048,0.008361777497662438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,2048,0.010021333065297868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,6144,0.03234222200181749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,2048,0.00870311094654931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,1536,0.007224889265166388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,1536,0.008995555341243744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,1536,0.008007110820876228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,1024,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,1024,0.008272888759771982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,1024,0.007304000357786815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,5120,0.029456890291637842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,768,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,768,0.007600000335110559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,768,0.007159111400445302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,4096,0.02682133350107405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,512,0.004538666870858935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,512,0.0070266665683852295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,512,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,3584,0.025432000557581585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,256,0.004232888834344016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,256,0.006648889018429651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,3072,0.02366666661368476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,256,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,128,0.003900444342030419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,128,0.006622222148709827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,2560,0.02051822178893619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,64,0.0038542221817705366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,64,0.006959111326270633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1536,32,0.0036186666952239144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1536,128,0.006603555546866522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1536,32,0.0069013333155049225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,65536,0.11511645052168105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,65536,0.1337911155488756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,65536,0.09331111113230388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,1536,0.017673777209387887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,16384,0.03506044546763102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,16384,0.03735733363363478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,16384,0.02868000004026625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,1024,0.016535111599498324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,12288,0.026341333985328674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,12288,0.03042222062746684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,768,0.01716977854569753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,12288,0.023524444964196947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,10240,0.022184888521830242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,10240,0.0266977780395084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,512,0.014814222852389017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,10240,0.02027466727627648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,8192,0.018589332699775696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,8192,0.022635555929607813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,8192,0.018280888597170513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,7168,0.03423377871513367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,7168,0.01687733332316081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,7168,0.020625778370433383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,7168,0.016196444630622864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,256,0.015805333852767944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,6144,0.014818666709793938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,6144,0.018925334016482036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,6144,0.01475288967291514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1536,128,0.01482311056719886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,5120,0.01349422252840466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,5120,0.01719199948840671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,5120,0.013808888693650564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,4096,0.025553777813911438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,4096,0.011397333608733283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,4096,0.0148044443792767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,65536,0.19200089242723253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,4096,0.01198933356338077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,16384,0.05786577860514323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,3584,0.010669333239396414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,3584,0.013934221532609729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,3584,0.011060444845093621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,3072,0.009369778136412302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,3072,0.012434666355450949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,12288,0.048116445541381836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,2560,0.020055999358495075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,2560,0.00830666638082928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,2560,0.010673777924643623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,3072,0.01015466699997584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,2048,0.01825066738658481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,2048,0.007121777368916406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,2048,0.009697777529557547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,2560,0.00962577760219574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,10240,0.041944887903001576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,1536,0.006292444550328785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,1536,0.008799110849698385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,2048,0.00869777798652649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,1536,0.007891555627187094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,1024,0.005192000004980299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,1024,0.007992000215583378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,1024,0.007251555720965068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,768,0.01683200067943997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,768,0.0046817776229646476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,8192,0.03719733489884271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,768,0.007567110988828871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,512,0.014124444789356656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,512,0.004010666575696734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,512,0.007245333658324347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,768,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,256,0.015815999772813585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,256,0.003559999995761447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,256,0.0069102222720781965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,512,0.006622222148709827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,128,0.014550222290886773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,128,0.003488000068399641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,128,0.006610666712125142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,256,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,64,0.0031679999083280563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,64,0.006697777658700943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,1024,32,0.0031804444475306403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,1024,32,0.006636444479227066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,1024,128,0.006346666564544042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,65536,0.0929608874850803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,65536,0.1318720049328274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,6144,0.03181511163711548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,65536,0.09228533506393433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,16384,0.02834133307139079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,16384,0.03689777851104736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,16384,0.028229332632488672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,5120,0.028833776712417603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,12288,0.020668443706300523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,12288,0.029920889271630183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,12288,0.023334221707450017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,10240,0.04201689031389025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,10240,0.017672888106769986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,10240,0.025843555728594463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,10240,0.019949333535300363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,3072,0.021710221966107685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,3584,0.0236453331179089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,8192,0.015189333094490899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,8192,0.022239999638663396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,7168,0.03381777803103129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,7168,0.013649777405791812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,8192,0.01793688866827223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,7168,0.020615999897321064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,6144,0.031249778138266668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,6144,0.01258933295806249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,7168,0.01584533353646596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,6144,0.018564444449212816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,1536,0.01719377769364251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,5120,0.01129955550034841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,5120,0.016922665966881644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,6144,0.014940443966123792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,5120,0.013407111167907715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,4096,0.009679111341635386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,4096,0.014584889014561972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,4096,0.011751111182901593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,1024,1024,0.015465777781274585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,3584,0.008970666262838576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,3584,0.01311377767059538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,3584,0.011022222538789114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,65536,0.19051110744476318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,3072,0.008291555775536431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,3072,0.012059555285506778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,3072,0.010112889111042023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,2560,0.019575110740131803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,2560,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,2560,0.010910222099887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,12288,0.047741333643595375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,2560,0.009369778136412302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,2048,0.006261333409282897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,2048,0.009654222263230218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,16384,0.05814933114581638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,2048,0.008632000121805403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,1536,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,1536,0.00868622213602066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,1536,0.007647111184067196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,8192,0.03687999977005853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,1024,0.004864000197913912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,1024,0.00796088907453749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,1024,0.007048889166778988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,768,0.017161778277821012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,768,0.0042364444169733255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,768,0.007606222397751278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,768,0.006822222222884496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,4096,0.023976888921525743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,512,0.0042035554846127825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,512,0.007144889069928064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,5120,0.028597331709331934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,256,0.015496888094478183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,256,0.003710222327046924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,3584,0.023318222827381555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,256,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,512,0.006584888945023219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,128,0.0035128887328836652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,128,0.006691555596060223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,256,0.0064311110311084324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,64,0.0034968890249729156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,64,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,768,128,0.006309333360857434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,768,32,0.0032097777972618737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,768,32,0.006628444625271692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,65536,0.0633173320028517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,65536,0.13095466295878092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,65536,0.09183200200398763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,3072,0.020959110723601446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,16384,0.02011999984582265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,16384,0.03698488738801744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,16384,0.02807733416557312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,2048,0.017826666434605915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,12288,0.020205333828926086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,12288,0.029327998558680218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,12288,0.02278577784697215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,10240,0.03924977779388428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,10240,0.017064000169436138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,10240,0.02606933315594991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,10240,0.01977511081430647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,1536,0.017812444104088675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,1024,0.015803555647532146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,8192,0.014290667242474027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,8192,0.022279111875428095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,8192,0.01796711153454251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,7168,0.013225778109497495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,7168,0.020219556159443326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,7168,0.015599111715952555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,6144,0.030105776256985132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,6144,0.011748444702890186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,6144,0.018447111050287884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,6144,0.014326221413082547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,5120,0.026770666241645813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,5120,0.010581333604123859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,5120,0.01680799987581041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,5120,0.012755555411179861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,4096,0.023001778456899855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,4096,0.009380444056457942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,4096,0.013808888693650564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,4096,0.011713777979214987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,512,0.01442399952146742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,3584,0.008057778080304464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,3584,0.012729778057999082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,3584,0.010807111031479306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,65536,0.17915022373199463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,3072,0.007250666618347168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,768,128,0.014817777607176038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,3072,0.011695110963450538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,2560,0.019039111004935372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,2560,0.006558222075303395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,3072,0.010029333333174387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,2560,0.010644444160991244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,2048,0.017533333765135873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,2048,0.005859555469618903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,2560,0.009339555270142024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,2048,0.009361777868535783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,2048,0.00832622249921163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,16384,0.054026666614744395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,1536,0.005223999834722943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,1536,0.008807111117574904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,1536,0.007611555357774098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,1024,0.004262222184075249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,1024,0.007983999947706858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,1024,0.006950221955776215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,768,0.016221332881185744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,768,0.004231111043029361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,768,0.007631111476156447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,768,0.006619555668698416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,12288,0.044871111710866295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,512,0.003858666867017746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,512,0.006959111326270633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,512,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,8192,0.03459999958674113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,256,0.0035342222286595237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,256,0.006909333169460297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,256,0.006272888845867581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,7168,0.03210577699873183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,128,0.0127697777416971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,128,0.0031315556002987754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,128,0.006608888920810487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,64,0.002983999955985281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,64,0.006917333437336816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,512,32,0.0031395554542541504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,512,128,0.006296000132958095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,512,32,0.006598222172922558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,65536,0.03868710994720459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,65536,0.13121599621242946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,65536,0.09101421965493096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,3584,0.021331555313534204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,16384,0.01647111111217075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,16384,0.0369777778784434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,16384,0.028038223584493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,3072,0.02023555503951179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,12288,0.013841778039932251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,12288,0.02942133280966017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,12288,0.022688888841205176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,1536,0.016812443733215332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,10240,0.013774221969975365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,10240,0.025626666016048853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,10240,0.01943466729587979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,1024,0.01444355481200748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,8192,0.01388533330625958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,8192,0.022112889422310725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,8192,0.01752088963985443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,512,0.013460444079505073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,7168,0.01254666679435306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,7168,0.020253333780500624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,7168,0.015371556083361307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,512,256,0.014486221803559197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,6144,0.011340444286664328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,6144,0.018176888426144917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,6144,0.013748444616794586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,65536,0.1740657753414578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,5120,0.010012444522645738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,5120,0.01648977730009291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,5120,0.012430222498046027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,16384,0.05311288767390781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,4096,0.008151999778217739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,4096,0.013741333451535968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,4096,0.011578666667143503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,12288,0.04369422131114536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,3584,0.007607999775144789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,3584,0.012411555482281579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,10240,0.03818666603830125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,3584,0.010728889041476779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,3072,0.006953777538405524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,3072,0.011408888631396823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,3072,0.010054222411579555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,8192,0.033929778469933405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,2560,0.006215999937719769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,2560,0.010659555594126383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,2560,0.009348444640636444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,2048,0.01704355577627818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,2048,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,2048,0.009358222285906473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,7168,0.030869334936141968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,2048,0.008300444318188561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,1536,0.004906666775544484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,1536,0.008627555436558193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,1536,0.007605333295133378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,6144,0.028520001305474177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,1024,0.014448000325096978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,1024,0.004193777839342753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,1024,0.00792711145348019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,1024,0.0069724445541699724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,768,0.0038720000949170855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,768,0.007633777956167857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,4096,0.02188266648186578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,768,0.006634666687912411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,512,0.003506666670242945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,3584,0.020462221569485135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,512,0.0069546666410234236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,5120,0.024316444993019104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,512,0.006559999866618051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,256,0.0032346666687064697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,256,0.006782222125265334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,256,0.0063386667105886675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,128,0.00286311118139161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,128,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,256,128,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,64,0.00286666676402092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,64,0.006729777902364731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,256,32,0.002867555452717675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,256,32,0.006647111227114995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,65536,0.17269955741034615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,65536,0.030488888422648113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,65536,0.13116444481743708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,65536,0.09081244468688965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,16384,0.05239199929767185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,16384,0.013516444298956128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,16384,0.03730310996373495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,16384,0.02740266587999132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,3072,0.01957155598534478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,12288,0.0141004439857271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,12288,0.029116445117526587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,12288,0.02238222294383579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,2560,0.01796355512407091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,10240,0.01276533305644989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,10240,0.025721778472264607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,10240,0.01921066641807556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,1536,0.015456888410780164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,8192,0.011343999869293638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,8192,0.021959111094474792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,8192,0.01720266706413693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,768,0.016175111134847004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,7168,0.010654222634103564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,7168,0.01993600030740102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,7168,0.015271999769740634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,512,0.013178666432698568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,6144,0.009319110876984065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,6144,0.0181377778450648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,6144,0.013413333230548434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,128,0.0120666664507654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,5120,0.0086666668454806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,5120,0.016163556112183463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,5120,0.012422222230169507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,256,256,0.013111111190583972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,4096,0.008002666963471307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,4096,0.013447999954223633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,4096,0.011391999820868174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,12288,0.042817778057522245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,3584,0.006992888947327931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,3584,0.012431110772821637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,3584,0.0107360002067354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,3072,0.01831733352608151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,3072,0.006803555621041193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,3072,0.011417778001891242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,3072,0.010037333601050908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,10240,0.03769422239727444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,2560,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,2560,0.010399999717871347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,2560,0.009306666751702627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,2048,0.01650044487582313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,2048,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,2048,0.009325332939624786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,2048,0.008290666672918532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,8192,0.03304977880583869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,1536,0.004910222358173794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,1536,0.008611555728647444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,1536,0.007616889145639207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,7168,0.030019554826948378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,1024,0.004222222086456087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,1024,0.00793599999613232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,1024,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,6144,0.026895999908447266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,768,0.0035342222286595237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,768,0.007639110916190677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,768,0.006692444284756978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,5120,0.023891554938422307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,512,0.0035075553589397003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,512,0.007002666592597961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,4096,0.02086666723092397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,512,0.006591111007663939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,256,0.003175111073586676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,256,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,3584,0.019655999210145738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,256,0.006324444380071428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,128,0.0028782222006056043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,128,0.006629333313968446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1024,128,128,0.006245333287451003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,64,0.0028577778074476454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,64,0.0069288888739215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,128,32,0.0028417776856157514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,128,32,0.006618666566080517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,65536,0.03159911102718777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,65536,0.13132711251576742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,16384,0.01275466630856196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,16384,0.03676177726851569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,12288,0.010618666807810465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,12288,0.02910400099224514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,10240,0.010478222535716163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,10240,0.025410667061805725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,8192,0.00958933350112703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,8192,0.021930666433440313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,7168,0.008747555315494537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,7168,0.019609777463806998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,6144,0.008171555896600088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,6144,0.017954667409261067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,5120,0.007753777835104201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,2560,0.01716800034046173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,5120,0.015824000040690105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,4096,0.007326222128338284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,4096,0.01312622262371911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,3584,0.006976000136799282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,3584,0.012033777932325998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,3072,0.006618666566080517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,3072,0.011372444530328115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,2560,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,2560,0.010370666782061258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,2048,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,2048,0.009333333207501305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,1536,0.004574222283230888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,1536,0.008613333106040955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,1024,0.003903111235962974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,1024,0.00795199970404307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,768,0.003551111039188173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,768,0.007346666521496243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,512,0.0032239999208185407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,512,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,256,0.0032017777363459268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,256,0.006584888945023219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,128,0.002874666617976295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,128,0.006569777925809224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,64,0.002532444480392668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,64,0.006628444625271692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,64,32,0.0028693332440323303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,64,32,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,65536,0.02678311036692725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,16384,0.013697778185208639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,65536,0.129786663585239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,16384,0.036509331729676985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,12288,0.011471110913488599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,12288,0.028786665863460962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,10240,0.010666666759385003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,10240,0.025444444682863023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,8192,0.00998133338159985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,8192,0.021738666627142165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,7168,0.009273777405420939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,7168,0.01960533360640208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,6144,0.008660444782839881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,6144,0.01789066692193349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,5120,0.0075866662793689305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,5120,0.015519110692871941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,4096,0.0075911109646161394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,4096,0.013083555632167392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,3584,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,3584,0.012442666623327466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,3072,0.006335111127959357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,2560,0.005881777654091517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,3072,0.011419555379284753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,2560,0.01000533335738712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,2048,0.005920889063013925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,2048,0.009404444032245213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,1536,0.0052177777720822235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,1536,0.008519111408127679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,1024,0.004368888835112254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,1024,0.007654222349325816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,768,0.003903111235962974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,768,0.007322666545708974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,512,0.003536888708670934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,512,0.006969778074158563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,256,0.0031928889867332247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,256,0.0062871111763848206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,128,0.0028773333049482773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,128,0.006637333167923822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,64,0.0028373334142896864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,64,0.006574222197135289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1024,32,32,0.0029057777590221832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1024,32,32,0.006588444527652528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,1536,0.015468445089128284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,16384,1.13573424021403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,16384,0.6648222075568305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,1024,0.013800000150998434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,768,0.0147733340660731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,12288,0.8912035624186198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,12288,0.5073359807332357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,12288,0.49816354115804035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,512,0.01238666640387641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,256,0.01201866649919086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,10240,0.7042222023010254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,10240,0.42061511675516766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,8192,0.5526169141133627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,8192,0.3288195663028293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1024,128,128,0.011756444142924415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,16384,0.6054071320427789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,7168,0.47704802619086367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,7168,0.28978578249613446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,10240,0.3836391237046983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,6144,0.2548302279578315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,6144,0.42741955651177305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,8192,0.3117973274654812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,6144,0.26277245415581596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,7168,0.274216890335083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,5120,0.3527324464586046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,5120,0.2230008973015679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,16384,0.32866401142544216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,12288,0.25190843476189506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,4096,0.2895982265472412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,4096,0.17320444848802355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,10240,0.2130124436484443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,8192,0.17479822370741102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,3584,0.2540408770243327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,3584,0.17545866966247559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,5120,0.2070337798860338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,4096,0.17225866847568086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,3072,0.21743999587164986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,3072,0.13755733437008327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,7168,0.15484178066253662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,3584,0.15583022435506186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,2560,0.18475466304355195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,2560,0.11859288480546738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,6144,0.13702932993570963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,5120,0.1163511143790351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,2048,0.1493848827150133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,2048,0.09966133038202922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,4096,0.09722755352656047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,1536,0.10144178072611491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,1536,0.11479643980662028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,1536,0.07969155576494005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,3072,0.14203288820054796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,1024,0.08808444605933295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,1024,0.08072621954811944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,1024,0.06013155645794339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,3584,0.08584711286756729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,768,0.05202755663130018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,768,0.06242933538224962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,768,0.05445955528153313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,2560,0.12836266888512507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,512,0.07438044415579902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,512,0.047162665261162647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,512,0.049878223074807056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,2048,0.11517155170440674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,3072,0.07555466890335083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,256,0.03106044398413764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,256,0.047261334127850003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,2560,0.06701511144638062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,128,0.03588355580965678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,128,0.025349333882331848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,128,0.045893334680133395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,2048,0.056406219800313316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,64,0.021728888154029846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,64,0.04426844583617317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,65536,32,0.021818665994538203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,65536,32,0.04419555597835117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,1536,0.04444266690148247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,65536,256,0.03938488827811347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,65536,1.181433783637153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,65536,0.6644248962402344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,16384,0.17825510766771105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,16384,0.3188764519161648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,1024,0.03577955563863119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,16384,0.16910489400227866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,768,0.03254310952292548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,12288,0.2239760027991401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,12288,0.13063377804226345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,512,0.029496888319651287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,10240,0.10832711060841878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,10240,0.19788266552819145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,10240,0.11107733514573838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,65536,0.3712115552690294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,16384,0.0958097775777181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,8192,0.1613022221459283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,8192,0.09079911311467488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,12288,0.0728000005086263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,128,0.027225777506828308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,7168,0.13994667265150282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,65536,256,0.027432888746261597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,10240,0.06161866585413615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,7168,0.07929688692092896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,6144,0.12056177192264134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,6144,0.06902755631340875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,65536,0.6435795360141331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,5120,0.08786666393280029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,5120,0.10300711128446792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,5120,0.07172177897559272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,8192,0.051353779104020864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,12288,0.12899822658962673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,4096,0.05781066417694092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,4096,0.08054844538370769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,4096,0.04892355534765455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,3584,0.05368533399369982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,3584,0.07048710849550036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,3584,0.043622222211625844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,7168,0.04508799976772732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,6144,0.04031199879116482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,3072,0.06030222442415026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,3072,0.037913779417673744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,8192,0.09000533156924778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,5120,0.03496711121665107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,2560,0.046069333950678505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,2560,0.05134933193524679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,2560,0.03274666600757175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,4096,0.030887110365761652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,2048,0.04153600003984239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,2048,0.02830044428507487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,3584,0.027855111493004694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,1536,0.038141333394580416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,1536,0.031974222924974226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,1536,0.02441155579355028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,3072,0.025664000047577754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,1024,0.03393866618474325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,1024,0.02238666680124071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,7168,0.08164711130989923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,1024,0.019827556278970506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,2560,0.023473777704768713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,768,0.017998221847746108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,768,0.018803555104467604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,6144,0.07299644417232938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,2048,0.020609777834680345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,512,0.013745778136783175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,512,0.016385777129067313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,1536,0.017024000485738117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,256,0.023744889431529578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,256,0.012436444560686747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,256,0.013921777407328287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,1024,0.014579556054539151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,3072,0.04933333396911621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,128,0.010703999963071613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,128,0.01298577752378252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,768,0.013502221968438892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,64,0.00998044427898195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,64,0.012597333225939008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,16384,32,0.010032888915803697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,16384,32,0.012822222378518848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,65536,0.5393626425001357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,2048,0.04195822278658549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,512,0.013086222112178802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,256,0.012077333198653327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,65536,0.8186311191982694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,65536,0.544523557027181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,16384,0.25787022378709584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,16384,0.15216267108917236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,12288,0.13987378279368082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,12288,0.15552443928188747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,16384,128,0.011963555382357704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,12288,0.10252266459994847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,768,0.027169777287377253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,10240,0.13624710506863064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,10240,0.08778578042984009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,65536,0.27471645673116046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,8192,0.08044533597098456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,8192,0.10890133513344659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,8192,0.07140888770421346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,16384,0.07133955425686307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,7168,0.07255911164813571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,7168,0.097397334045834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,512,0.029442667961120605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,7168,0.06380266613430448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,6144,0.0659608907169766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,6144,0.08409333229064941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,6144,0.05556178092956543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,12288,0.05433422327041626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,16384,128,0.021239999267790053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,10240,0.04615733358595106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,5120,0.06970489025115967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,5120,0.04738666613896688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,4096,0.05237333310974968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,4096,0.05600800116856893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,4096,0.038916445440716214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,8192,0.03916177815861172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,7168,0.03457511133617825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,3584,0.04906222224235535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,3584,0.035022222333484225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,6144,0.030917333232031927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,5120,0.028014222780863445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,3072,0.04251911242802938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,3072,0.03151999910672506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,4096,0.024246222443050806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,16384,0.141595549053616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,3584,0.02232444451914893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,2560,0.03596266773011949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,2560,0.028174221515655518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,10240,0.09433688720067342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,2048,0.028646220763524372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,2048,0.024484443995687697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,3072,0.020603555772039626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,2560,0.01872177753183577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,1536,0.022309333086013794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,1536,0.02089777754412757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,5120,0.05882044633229574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,1024,0.030823998981051978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,1024,0.01593244406912062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,1024,0.017500445246696472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,2048,0.016903110676341586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,3584,0.048713776800367564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,768,0.013054221868515015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,768,0.016041777200169034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,1536,0.014009777042600842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,3072,0.04478844337993198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,512,0.010445333189434476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,1024,0.012416000167528788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,512,0.013437333206335703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,768,0.01167555567291048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,256,0.007308444215191736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,256,0.012267555627557965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,2560,0.042146666182412036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,512,0.010397333237859938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,128,0.0070426662762959795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,2048,0.03811733259095086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,128,0.011691555380821228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,64,0.006248888870080312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,64,0.011361777782440186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,12288,32,0.006303999986913469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,12288,32,0.011543110840850405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,256,0.010030222435792288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,1536,0.034507556094063654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,65536,0.7075760099622938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,12288,128,0.010032888915803697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,65536,0.4454951286315918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,16384,0.11577510833740234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,16384,0.1770026683807373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,16384,0.11459999614291722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,768,0.026364444030655756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,12288,0.13800888591342503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,12288,0.08722844388749863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,65536,0.2527857886420356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,10240,0.07851466867658827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,512,0.026290666725900438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,16384,0.06493866443634033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,10240,0.11578222115834554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,10240,0.07388177845213148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,8192,0.06719555457433064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,8192,0.09349599811765884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,8192,0.060305780834621854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,256,0.022612444228596155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,7168,0.06115200122197469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,7168,0.08338844113879734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,7168,0.05350222190221151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,12288,0.050458666351106435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,6144,0.07058666812049018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,6144,0.04642933275964525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,10240,0.04272977842224968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,8192,0.036632001399993896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,5120,0.05964710977342394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,12288,128,0.020625778370433383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,5120,0.04011199871699015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,4096,0.044575111733542554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,7168,0.03265155686272515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,4096,0.04872533347871569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,4096,0.033160001039505005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,6144,0.030096888542175293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,65536,0.4294862217373318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,3584,0.04522666666242811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,3584,0.030001779397328694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,3072,0.03906755646069845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,3072,0.03877244393030802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,3072,0.02753511071205139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,5120,0.02642311155796051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,2560,0.036538667149013944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,4096,0.02328355610370636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,2560,0.0329733325375451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,2560,0.024129778146743774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,2048,0.03392533461252848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,2048,0.02700622214211358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,2048,0.021574222379260596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,3584,0.02161866592036353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,3072,0.019911110401153564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,1536,0.021026665965716045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,1536,0.018453333112928603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,12288,0.09088266558117336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,2560,0.018171555466122098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,1024,0.016235555211702984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,1024,0.015491555134455362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,2048,0.015929778416951496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,6144,0.05598666932847765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,1536,0.013939556148317126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,768,0.013302221894264221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,768,0.013793778088357715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,1024,0.01204177737236023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,512,0.01034755590889189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,512,0.011682666838169098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,5120,0.050049778487947255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,256,0.021779555413458083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,256,0.007870222131411234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,256,0.01072799993885888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,768,0.011234666738245221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,3584,0.04131733377774557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,512,0.010227555202113258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,128,0.006683555742104848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,128,0.010393777655230628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,64,0.006608888920810487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,64,0.01038044442733129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,10240,32,0.006625777731339137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,10240,32,0.010485333700974783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,256,0.009666666388511658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,65536,0.5725822448730469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,1536,0.03012088934580485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,10240,128,0.009563555320103964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,65536,0.38498844040764707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,16384,0.14277511172824436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,16384,0.10009333160188462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,1024,0.027336888843112524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,12288,0.08364266819424099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,12288,0.11071377992630005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,12288,0.06982755661010742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,768,0.024393777052561443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,65536,0.22932889726426867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,10240,0.09690044323603313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,10240,0.059008002281188965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,16384,0.054217778974109225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,8192,0.06241244739956326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,512,0.02436711059676276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,8192,0.07817600170771281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,8192,0.04763111140992907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,12288,0.04393511017163595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,7168,0.06810755862130059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,7168,0.04296088880962796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,10240,0.03636711173587375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,8192,0.03165510959095425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,6144,0.058887998263041176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,6144,0.037738667594061956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,10240,128,0.01976355579164293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,5120,0.04633511106173197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,7168,0.0277511113219791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,5120,0.04974933465321859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,5120,0.03228622343805101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,6144,0.025316443708207872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,4096,0.03971111112170749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,4096,0.027215111586782668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,5120,0.0225262227985594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,65536,0.3773662249247233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,3584,0.03467822074890137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,3584,0.0247662216424942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,4096,0.02032355633046892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,16384,0.1059911118613349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,3072,0.0299653344684177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,3072,0.022799111074871484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,3584,0.018473777506086562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,2560,0.03424888849258423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,2560,0.02546311087078518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,2560,0.02023555503951179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,10240,0.071615113152398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,3072,0.017307556337780423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,7168,0.05657511287265354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,2048,0.021054221524132624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,2048,0.018193778064515855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,1536,0.028590222199757893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,1536,0.016182222300105624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,1536,0.015772443678643968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,2560,0.01585244470172458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,1024,0.025464889076020982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,1024,0.01330311099688212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,1024,0.01272266689274046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,6144,0.05167466733190748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,768,0.02351288828584883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,768,0.01167377746767468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,768,0.011511999699804517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,2048,0.014290667242474027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,1536,0.012317332956526013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,512,0.009369778136412302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,512,0.010328888893127441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,1024,0.010710222025712332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,768,0.010062221851613786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,256,0.007592889169851939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,256,0.009431111315886179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,4096,0.04156444470087687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,128,0.01833777791923947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,128,0.006788444601827198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,128,0.009181333084901174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,512,0.009352888498041365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,64,0.006311999840868845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,64,0.009002666506502364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,8192,32,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,8192,32,0.009318222602208456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,3584,0.038634667793909706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,65536,0.3398631148868137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,256,0.008961777720186446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,65536,0.5294888814290365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,65536,0.3366853396097819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,8192,128,0.008984888593355814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,16384,0.1389955547120836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,16384,0.0851057767868042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,3072,0.0366542206870185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,12288,0.1009075575404697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,12288,0.0654639999071757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,65536,0.1911546654171414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,16384,0.05139466789033678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,10240,0.10247377554575603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,10240,0.05556977457470364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,2048,0.031694223483403526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,12288,0.04132444328731961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,8192,0.08304888672298856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,8192,0.045082665152019925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,10240,0.034790221187803484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,7168,0.05168088939454821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,7168,0.07291822301016913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,7168,0.040510220660103693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,512,0.022629333866967097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,8192,0.03051466743151347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,6144,0.06213066975275675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,6144,0.03641155693266127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,8192,256,0.020907556017239887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,7168,0.026975999275843304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,5120,0.050979554653167725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,5120,0.031515555249320135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,6144,0.024832000335057575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,5120,0.022073777185546026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,4096,0.04076000054677328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,4096,0.026621333426899377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,16384,0.09434222512774998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,12288,0.07483910852008395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,3584,0.03589244352446662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,3584,0.02443555494149526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,4096,0.0199937770764033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,3072,0.0336773329310947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,3072,0.031116445859273274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,10240,0.06509777572419909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,3072,0.022447110878096685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,2560,0.03164533442921109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,2560,0.026231999198595684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,2560,0.020106666617923312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,3584,0.018204443984561496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,3072,0.01700177788734436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,8192,0.05620266993840536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,2048,0.02148088812828064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,2048,0.017831999394628737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,1536,0.026704889204767015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,2560,0.015658666690190632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,1536,0.017491555876202054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,1536,0.015632000234391954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,2048,0.013946667313575745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,1024,0.011358222199810876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,1024,0.012552888856993781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,1536,0.011897777517636618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,6144,0.04732444551255968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,768,0.009689778089523315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,768,0.011176889141400656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,5120,0.04270222120814853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,1024,0.010563555690977307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,512,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,4096,0.038359112209743924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,512,0.010002666877375709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,256,0.01993777851263682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,256,0.006219555520349079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,256,0.009356444080670675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,768,0.009384888741705153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,512,0.00907466643386417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,128,0.005587555468082428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,3584,0.03559199968973795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,128,0.009315555294354757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,64,0.005224888937340842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,256,0.00828888929552502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,64,0.008979555633332994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,7168,32,0.004919111314747068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,7168,32,0.00927466650803884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,65536,0.44350488980611164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,65536,0.33239022890726727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,7168,128,0.008014221986134848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,2048,0.02885155545340644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,65536,0.1522844500011868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,16384,0.11329332987467448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,16384,0.07917510800891452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,1024,0.023935111032591924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,12288,0.08614133463965522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,12288,0.06131911277770996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,16384,0.04076177875200907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,12288,0.03216266632080078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,10240,0.06973777876959907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,10240,0.05225066675080193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,768,0.022688888841205176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,512,0.02118844456142849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,8192,0.055735111236572266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,8192,0.043218665652804904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,10240,0.028166221247778997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,8192,0.024382222029897902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,7168,0.048904889159732394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,7168,0.0391884446144104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,7168,128,0.01773599949147966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,65536,0.312171565161811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,7168,0.02197955548763275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,6144,0.04180444280306498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,6144,0.03495733274353875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,16384,0.08822400040096706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,5120,0.03575911124547323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,6144,0.020443555381562974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,5120,0.030619555049472388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,12288,0.07056711117426555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,5120,0.018607111440764535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,4096,0.029024001624849107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,4096,0.02604622145493825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,10240,0.06118577718734741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,3584,0.025400888588693406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,4096,0.01657511128319634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,3584,0.023991111252042983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,8192,0.053025778796937734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,3072,0.022123555342356365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,3072,0.021912889348136053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,3584,0.015278221832381355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,7168,0.04883289006021288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,2560,0.029866665601730347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,2560,0.018915555543369718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,2560,0.019574221637513902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,2048,0.02756711178355747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,2048,0.01588000026014116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,2048,0.017439110411538016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,3072,0.014460444450378418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,2560,0.013608000344700284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,1536,0.013032000097963544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,1536,0.015294222368134392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,2048,0.012057777908113269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,6144,0.04478933413823446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,1024,0.010384000009960597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,1024,0.012111110819710625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,5120,0.040354665782716535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,1536,0.010623111493057676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,768,0.008649778034951951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,768,0.011000889043013254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,1024,0.009132444030708736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,512,0.01924711134698656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,512,0.006974221931563483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,512,0.00999644481473499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,768,0.008543111383914948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,4096,0.03638755612903171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,256,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,256,0.00928177767329746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,512,0.007867555651399825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,3584,0.034154666794670954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,128,0.004875555634498596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,128,0.00902844468752543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,256,0.007712889048788283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,64,0.0043333334227403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,64,0.008964444200197855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,6144,32,0.004564444637960858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,6144,32,0.00868088917599784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,6144,128,0.007426666716734569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,65536,0.3727280033959283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,3072,0.031934221585591636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,65536,0.23832265535990396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,65536,0.14204355080922446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,16384,0.09710311227374607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,16384,0.06237155861324734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,1536,0.0255724440018336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,1024,0.022856000396940444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,12288,0.07430933581458198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,16384,0.03961688942379422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,12288,0.048080000612470836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,10240,0.05250666538874308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,10240,0.06231377522150675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,10240,0.041013333532545306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,768,0.02093422247303857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,12288,0.03157866663402981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,8192,0.04649600055482653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,8192,0.04944444364971585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,8192,0.03340533375740051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,7168,0.042945779032177396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,7168,0.04343733191490173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,10240,0.0273991111252043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,7168,0.03056710958480835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,256,0.018608000543382432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,8192,0.02407199972205692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,6144,0.03812444541189406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,6144,0.027461333407296076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,7168,0.021686222818162706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,5120,0.032368001010682844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,5120,0.024505777491463557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,6144,128,0.016515556308958266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,6144,0.019869332512219746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,4096,0.026174222429593403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,4096,0.020988444487253826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,5120,0.017966222431924608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,65536,0.25251732932196724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,3584,0.02316444449954563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,3584,0.01955111159218682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,4096,0.016540444559521146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,16384,0.07401600148942736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,3072,0.02021511064635383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,3584,0.0151591118839052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,3072,0.017668444249365065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,2560,0.027747554911507502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,2560,0.01753777762254079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,2560,0.016413332687483895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,3072,0.014095111025704278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,2560,0.013199999928474426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,2048,0.01531555586391025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,2048,0.01440088947614034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,12288,0.06013688776228163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,2048,0.011735999749766456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,1536,0.012875555290116204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,1536,0.012563555604881711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,6144,0.039340443081325956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,1536,0.009999111294746399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,1024,0.010015111002657149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,1024,0.010063110954231685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,5120,0.03625511129697164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,1024,0.008609777523411645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,768,0.008672888908121321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,768,0.009362666971153682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,4096,0.032784889141718544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,3584,0.030600888861550227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,512,0.0069288888739215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,512,0.008622222476535374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,3072,0.029378665818108454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,768,0.008000888758235509
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,256,0.005919999960396025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,256,0.008087111016114553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,512,0.007630222373538547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,256,0.007292444507280986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,128,0.004903111192915174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,128,0.008001777860853408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,2048,0.025795555777019922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,64,0.004869333157936732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,64,0.00795022232664956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,5120,32,0.005577777822812398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,5120,32,0.007670222057236566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,5120,128,0.007301333049933116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,65536,0.31937066713968915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,65536,0.2650800016191271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,1536,0.02362666692998674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,16384,0.07008533345328437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,16384,0.08051111300786336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,16384,0.05776266919242012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,65536,0.135425779554579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,12288,0.0569173362520006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,12288,0.0619751082526313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,12288,0.044828444719314575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,16384,0.03819555706448025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,1024,0.02070577773782942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,10240,0.05151289039187961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,10240,0.039289779133266874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,12288,0.030613332986831665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,10240,0.026496888862715826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,8192,0.04166400101449754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,8192,0.032837334606382586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,768,0.02068088948726654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,8192,0.023456000619464453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,512,0.01809599995613098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,7168,0.03677422139379714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,7168,0.0302817788388994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,256,0.01828888886504703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,6144,0.03187111020088196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,6144,0.02720444401105245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,7168,0.020799999435742695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,5120,128,0.016265778077973258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,5120,0.02715377840730879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,5120,0.02383111086156633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,6144,0.019419555862744648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,5120,0.017328000730938382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,4096,0.022222222553359136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,4096,0.020819556381967332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,65536,0.23296801249186197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,4096,0.015799999237060547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,3584,0.019732443822754752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,3584,0.019312888383865356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,3584,0.014704000618722705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,10240,0.04976800084114075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,3072,0.017032000753614638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,3072,0.017863111363516915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,3072,0.01400888959566752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,2560,0.026457778281635706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,2560,0.014899555179807874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,2560,0.01590044465329912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,8192,0.04359199934535556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,2560,0.012998222476906247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,2048,0.01311644415060679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,2048,0.014136888914638095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,7168,0.040455109543270536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,2048,0.011058666639857821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,1536,0.01055022246307797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,1536,0.011797332929240333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,6144,0.037301331758499146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,1536,0.009301332963837517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,5120,0.03427199853791131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,4096,0.03127288818359375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,1024,0.008237332933478886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,1024,0.009876444107956356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,768,0.007024889190991719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,768,0.009356444080670675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,1024,0.00830844458606508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,768,0.007768889268239339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,512,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,512,0.008279999925030602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,3584,0.02903555499182807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,256,0.016193777322769165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,256,0.004912888838185204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,256,0.007939555578761632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,512,0.007599111232492659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,256,0.007271111011505127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,128,0.004221333397759331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,128,0.007956444389290279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,3072,0.027452444036801655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,64,0.004192000048028098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,64,0.00794311116139094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,4096,32,0.0041991112132867174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,4096,32,0.00795377790927887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,4096,128,0.007275555696752336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,65536,0.27776622772216797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,65536,0.21477599938710532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,65536,0.13322577211591932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,2048,0.024031110935741003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,16384,0.07202133205201891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,16384,0.05634044276343452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,16384,0.03727644350793626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,12288,0.052694221337636314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,12288,0.05446933375464546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,12288,0.0443120002746582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,1536,0.021908443835046556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,10240,0.04648800028695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,10240,0.0454053348965115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,10240,0.038592888249291316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,1024,0.0192248887485928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,12288,0.0299653344684177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,8192,0.03722488880157471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,8192,0.04106400079197354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,768,0.019555555449591745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,8192,0.0322693321439955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,7168,0.038033776813083224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,7168,0.03277422322167291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,7168,0.029658665259679157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,10240,0.026157332791222468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,6144,0.028061333629820082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,6144,0.026775999201668635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,8192,0.02297244469324748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,512,0.016843555702103507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,5120,0.023893333143658105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,7168,0.020605333977275424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,5120,0.02392622166209751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,4096,0.029382222228580054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,4096,0.020625778370433383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,6144,0.019141332970725167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,4096,0.02056533263789283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,3584,0.02755555510520935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,3584,0.01686044368478987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,3584,0.019031110737058852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,5120,0.01719999975628323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,3072,0.026143999563323125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,3072,0.015215999550289579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,3072,0.01721155477894677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,4096,0.015856888559129503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,2560,0.024746666351954143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,2560,0.013264888690577613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,2560,0.015736889508035447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,3584,0.014733332726690503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,3072,0.013454222016864352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,2048,0.011680000358157687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,4096,128,0.01579466627703773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,2048,0.013506666653686099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,2560,0.012262221839692859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,1536,0.010249777800507015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,1536,0.011516444385051727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,2048,0.01070755554570092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,1024,0.017835555805100333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,1024,0.007271111011505127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,1024,0.009749333063761393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,65536,0.21187288231319854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,768,0.018883556127548218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,768,0.006569777925809224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,1536,0.00907911111911138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,768,0.009008888569143083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,1024,0.00828888929552502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,512,0.00554933316177792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,512,0.008270222279760573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,768,0.007793777518802219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,512,0.007584888901975419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,256,0.004606222112973531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,256,0.007970666719807519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,16384,0.06466577450434367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,256,0.007258666886223688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,128,0.003945777813593547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,128,0.00793511089351442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3584,128,0.007256000406212277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,64,0.0038177776667806837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,64,0.007699555820888943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3584,32,0.0038542221817705366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3584,32,0.00792711145348019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,6144,0.035014222065607704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,65536,0.23532266087002227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,65536,0.21204800075954863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,5120,0.0321742229991489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,16384,0.06206133630540636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,65536,0.0987280011177063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,16384,0.06076177623536852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,16384,0.05650399790869819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,12288,0.05027111040221321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,12288,0.045801778634389244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,12288,0.044453332821528115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,2048,0.023064000738991633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,16384,0.02680711117055681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,12288,0.022256000174416438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,10240,0.03903822104136149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,10240,0.038513776328828596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,1536,0.020613332589467365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,8192,0.03193333413865831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,8192,0.031918221049838595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,10240,0.01982399986849891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,8192,0.017633777525689866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,512,0.016152888536453247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,7168,0.028295109669367473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,7168,0.029038220643997192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,6144,0.03350222110748291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,6144,0.024365334047211543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,6144,0.02613688839806451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,7168,0.015969778100649517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,5120,0.03080000148879157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,5120,0.020605333977275424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,5120,0.02349600030316247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,6144,0.014814222852389017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,4096,0.02751911183198293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,4096,0.017136888371573556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,256,0.016215110818545025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,4096,0.02001422146956126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,5120,0.013801777528391944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,3584,0.015564444992277356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,3584,0.018560000591807894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,4096,0.012876444392734103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3584,128,0.01612977849112617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,3072,0.02489600082238515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,3072,0.013784888717863293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,3072,0.01719377769364251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,3584,0.011791110866599612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,2560,0.01240799989965227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,2560,0.015208888385030957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,3072,0.011013333168294696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,2048,0.021715554926130507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,2048,0.01073333372672399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,2048,0.013101333545313941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,2560,0.010384000009960597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,65536,0.20159378316667345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,2048,0.008997333546479544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,1536,0.011367110742463006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,1536,0.008734222087595198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,1536,0.00796088907453749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,1024,0.006908444480763541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,1024,0.009664889011118147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,1024,0.007151111132568783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,10240,0.04443200098143684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,768,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,768,0.009008000294367472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,8192,0.03933155536651611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,768,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,512,0.005010666532648934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,512,0.00832711077398724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,7168,0.03644088904062907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,512,0.006685333533419504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,256,0.00423911131090588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,256,0.007983999947706858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,256,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,3584,0.026224000586403742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,128,0.003914666672547658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,128,0.0079297779334916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,3072,128,0.006615110983451207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,64,0.003530666646030214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,64,0.00793688909875022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,3072,32,0.003532444437344869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,3072,32,0.007644444704055786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,2560,0.023619555764728125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,65536,0.19565422005123564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,65536,0.20403644773695204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,65536,0.14248977767096624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,1024,0.01680799987581041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,65536,0.08856444226370917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,16384,0.05243022243181864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,16384,0.03705599904060364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,1536,0.018936889039145578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,12288,0.040192888842688665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,12288,0.029870222012201946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,16384,0.027239110734727647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,10240,0.04354666670163473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,10240,0.033399111694759794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,10240,0.026693332526418898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,12288,0.02214488883813222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,8192,0.038458665211995445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,8192,0.027241778042581346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,8192,0.02275644408331977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,10240,0.019681778219011094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,768,0.018228444788191054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,8192,0.01735466718673706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,7168,0.023629332582155865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,7168,0.020950223008791607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,7168,0.015651555524932016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,6144,0.02055111030737559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,6144,0.019156444403860304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,6144,0.014588443769348992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,5120,0.030188444587919448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,512,0.015524443652894763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,5120,0.01760088900725047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,5120,0.017239999439981248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,256,0.015834665960735746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,5120,0.013688888814714221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,4096,0.027659555276234944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,4096,0.014879110786649915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,4096,0.014861333701345654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,4096,0.01238666640387641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,3584,0.013561777770519257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,3584,0.013901333014170328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,3072,0.02442666722668542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,3072,0.012332444389661154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,3072,0.01274577776590983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,3584,0.011403555671374003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,2560,0.022974222898483276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,2560,0.011111111276679568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,2560,0.011759111450778114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,3072,0.010750222537252637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,3072,128,0.015856888559129503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,2048,0.009694221946928237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,2048,0.010099555883142684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,2560,0.009683555199040307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,2048,0.008338666624493068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,1536,0.007328888608349695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,1536,0.00904444439543618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,16384,0.06031644344329834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,1536,0.007634667058785756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,1024,0.005959110955397288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,1024,0.01643200053109063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,1024,0.008004444340864817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,768,0.018253333038753934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,768,0.005535999933878581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,768,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,1024,0.0069191112286514705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,768,0.006690666493442323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,512,0.004577777865860197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,512,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,12288,0.04959200157059563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,256,0.01553066737122006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,256,0.004217777815130022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,256,0.006993778049945831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,512,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,128,0.015805333852767944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,128,0.0038853333228164246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,128,0.0069644442862934535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,7168,0.03562044435077243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,256,0.006434666613737742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,64,0.0036275556517971885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,64,0.0069671107663048645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2560,32,0.0038559999730851916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2560,32,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,65536,0.1729360024134318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2560,128,0.0063075555695427795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,65536,0.1279182169172499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,65536,0.08521599902047051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,6144,0.03288088904486762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,16384,0.04391466577847799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,16384,0.03603911068704393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,12288,0.04519822200139364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,12288,0.03386755453215705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,12288,0.02921688887808058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,16384,0.02589688864019182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,10240,0.04013599952061971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,10240,0.0280675556924608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,10240,0.025742222865422566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,12288,0.02124622298611535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,8192,0.03534577621353997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,8192,0.02310488952530755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,8192,0.02195555633968777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,10240,0.018876444962289598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,8192,0.016846223009957206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,3584,0.025722665919197932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,7168,0.020181333025296528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,7168,0.020336889558368258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,7168,0.015524443652894763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,6144,0.017863111363516915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,6144,0.018581334087583754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,6144,0.014327999618318347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,2048,0.020223110914230347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,1536,0.017865777015686035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,5120,0.014949333336618213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,5120,0.01680888897842831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,5120,0.013064000341627332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,4096,0.012793777717484368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,4096,0.014526221487257214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,4096,0.01181688904762268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2560,512,0.015127999915017022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,3584,0.011968889170222811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,3584,0.013165333204799227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,3584,0.010791111323568555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,65536,0.17434133423699272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,3072,0.011056000159846412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,3072,0.012538666526476542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,3072,0.01000088867213991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,2560,0.020999999509917364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,2560,0.009679111341635386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,2560,0.01106844428512785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,16384,0.05444622370931837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,2560,0.009348444640636444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,2048,0.008605333666006723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,2048,0.009678222239017487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,6144,0.030504888958401147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,7168,0.0329039990901947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,2048,0.008624888956546783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,1536,0.007088888850477006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,1536,0.008646222452322641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,1024,0.01555022266176012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,1024,0.005919999960396025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,1024,0.00797155582242542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,1536,0.00793511089351442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,768,0.017823111679818895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,768,0.0052595556610160405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,768,0.007642666498819987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,1024,0.007243555453088548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,512,0.0144213338692983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,512,0.004564444637960858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,512,0.0069617778062820435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,768,0.0069395556218094295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,5120,0.028100444210900202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,256,0.004234666625658671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,256,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,512,0.006585777633719974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,4096,0.02570755614174737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,128,0.00387999994887246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,128,0.0069155556460221606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,256,0.006278222219811545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,64,0.0038497779104444715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,64,0.0069431112044387394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,2048,32,0.0038888889054457345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,2048,32,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,65536,0.16874578264024523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,2048,128,0.006311111152172089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,65536,0.13650843832227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,65536,0.12252177132500543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,16384,0.05340177814165751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,16384,0.03393066591686673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,16384,0.0354728897412618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,65536,0.08504977491166855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,3584,0.023018666439586218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,12288,0.02569244470861223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,16384,0.025765332910749648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,12288,0.02864444586965773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,12288,0.02090488870938619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,10240,0.021526222427686054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,10240,0.02510844502184126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,10240,0.018573333819707234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,3072,0.022006221943431433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,8192,0.018061333232455783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,8192,0.02163999941613939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,8192,0.016647110382715862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,2048,0.018558222386572096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,7168,0.01625866691271464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,7168,0.01998044384850396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,7168,0.015224888920783997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,1536,0.017228444417317707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,6144,0.013966222604115805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,6144,0.018432888719770644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,6144,0.013953778478834363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,256,0.016128889388508268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,5120,0.0123511114054256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,5120,0.016763556334707472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,2048,128,0.014949333336618213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,5120,0.012715555727481842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,4096,0.01128000020980835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,4096,0.01445777714252472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,4096,0.011183111204041375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,12288,0.04399466514587402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,10240,0.03917333483695984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,3584,0.010556444525718689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,3584,0.012730666332774691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,3584,0.010211555494202508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,3072,0.009298666483826106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,3072,0.01183733344078064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,8192,0.03445955448680454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,3072,0.00979644474056032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,2560,0.008290666672918532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,2560,0.01072355525361167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,2560,0.009004444711738164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,7168,0.03193155593342251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,2048,0.006863111009200414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,2048,0.009299555586444007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,2048,0.008267555799749162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,6144,0.029691557089487713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,1536,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,1536,0.008767999708652496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,1536,0.007647111184067196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,1024,0.0151555554734336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,1024,0.004913777940803104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,1024,0.00798044436507755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,1024,0.007144889069928064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,768,0.016823111308945548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,768,0.004598222259018156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,768,0.007649777664078607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,4096,0.024708444873491924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,5120,0.027464000715149775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,512,0.003920000046491623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,768,0.006896889044178857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,512,0.0069431112044387394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,512,0.006586666736337874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,256,0.003907555507289039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,256,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,3584,0.02271288964483473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,256,0.006262222097979651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,128,0.015542222393883599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,128,0.0032133333798911837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,128,0.006731555693679386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1536,128,0.006235555642180973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,64,0.003129777808984121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,64,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1536,32,0.0031591111587153543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1536,32,0.006903999795516332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,65536,0.16686044798956978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,65536,0.0968080030547248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,65536,0.12115732828776042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,3072,0.020682666036817763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,16384,0.026857778429985046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,16384,0.035340444909201726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,65536,0.0830471118291219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,2560,0.01959111127588484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,12288,0.04336355461014641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,12288,0.022356443934970435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,12288,0.02843733297453986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,16384,0.02454400062561035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,10240,0.03825599948565165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,10240,0.01905955539809333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,10240,0.025040000677108765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,12288,0.020003555549515616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,8192,0.01737511157989502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,10240,0.01789777808719211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,8192,0.021290666527218286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,8192,0.016117332710160148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,7168,0.014491554763582019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,7168,0.019902222686343722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,7168,0.014607111612955729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,6144,0.029280887709723577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,2048,0.017883555756674874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,6144,0.013343111508422427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,6144,0.017902221944597032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,6144,0.013594666288958656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,5120,0.011852444046073489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,1536,0.016188444362746347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,5120,0.016467556357383728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,5120,0.011818666425016193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,4096,0.009633777870072259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,4096,0.0135057775510682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,4096,0.010674667027261523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,3584,0.021253334151373968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,3584,0.008562666674455007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,3584,0.012424888710180918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,3584,0.00997688869635264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,3072,0.019256000717480976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,3072,0.007988444632954067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,3072,0.011345778074529437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,3072,0.009330666727489894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,2560,0.018526222970750596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,2560,0.007088888850477006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,2560,0.01036622209681405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,2560,0.008936000367005667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,512,0.014083556003040738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,2048,0.006115555349323485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,2048,0.009337777892748514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,2048,0.008150222400824228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1536,256,0.015863110621770222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,1536,0.0052533331844541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,1536,0.008837333156002892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,1536,0.0073520003093613525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,16384,0.052128887838787503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,8192,0.033923556407292686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,1024,0.0046302220887608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,1024,0.007941332956155142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,768,0.01680888897842831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,768,0.004253333227501975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,7168,0.0317466656366984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,768,0.0074177781740824384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,1024,0.0069777775141927935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,512,0.0038728887836138406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,512,0.006933333145247565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,768,0.006568000134494569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,256,0.0159582214223014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,256,0.003391111062632667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,512,0.00629066675901413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,256,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,256,0.006288888967699475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,128,0.0032044444233179092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,128,0.006585777633719974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,1024,128,0.006238222122192383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,64,0.0033440000067154565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,64,0.006646222124497096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,1024,32,0.003220444545149803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,1024,32,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,5120,0.026375111606385972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,4096,0.022514666120211285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,65536,0.07202933231989543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,65536,0.15659911102718777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,65536,0.11963555547926162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,16384,0.04978311061859131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,16384,0.02162844439347585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,16384,0.034992890225516424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,65536,0.08144088586171468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,12288,0.02001333402262794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,12288,0.02817244331041972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,16384,0.02438844409253862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,12288,0.01992266707950168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,10240,0.0367253323396047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,10240,0.017255999975734286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,10240,0.024667556087176006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,10240,0.01775911119249132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,8192,0.03256355391608344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,8192,0.014863999353514777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,8192,0.020970667401949566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,8192,0.01607822212908003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,2048,0.01716888944307963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,7168,0.013438222308953604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,7168,0.019260444574885897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,6144,0.026890666948424444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,6144,0.011759999725553723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,7168,0.014455111490355598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,6144,0.017608889275126986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,5120,0.023984000086784363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,5120,0.010680889089902243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,5120,0.01612977849112617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,6144,0.013049778011110095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,4096,0.021543110410372417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,4096,0.009705777797434065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,5120,0.011709333293967776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,4096,0.013680000272062091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,4096,0.010688888529936472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,3584,0.008292444050312042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,3584,0.012379555238617791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,3584,0.009977777798970541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,3072,0.01920622256067064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,3072,0.007617777420414819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,3072,0.011024889018800525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,1536,0.01584533353646596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,2560,0.018242667118708294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,2560,0.006575111299753189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,2560,0.010207110808955299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,3072,0.009311111436949836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,2560,0.008948444492287105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,2048,0.0058871110280354815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,2048,0.009294222626421187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,2048,0.007975111405054728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,1536,0.016132444143295288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,1536,0.005097777893145879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,1536,0.008667555948098501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,1536,0.007303111255168915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,1024,0.015160888433456421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,1024,0.004534222185611725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,1024,0.007642666498819987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,1024,0.006628444625271692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,512,0.014127110441525778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,768,0.0041724443435668945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,768,0.007316444483068254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,768,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,1024,128,0.01515111161602868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,512,0.0038008888562520346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,512,0.006976000136799282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,512,0.006266666783226862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,256,0.015967999895413715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,256,0.0032186667538351486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,256,0.006635555376609166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,256,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,128,0.014816888504558139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,128,0.003216888962520493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,128,0.006599111275540457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,768,128,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,64,0.0028799999919202593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,64,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,768,32,0.0028773333049482773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,768,32,0.006934222247865465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,65536,0.15330399407280818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,65536,0.04965600040223864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,12288,0.04127644499142965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,65536,0.12053778436448838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,65536,0.08175022072262235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,16384,0.01978755493958791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,16384,0.03463911016782125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,16384,0.024251555403073628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,12288,0.03999733262591892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,12288,0.01720533271630605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,12288,0.0277839998404185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,12288,0.019893333315849304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,7168,0.030643555853101943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,10240,0.01568622224860721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,10240,0.02459999918937683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,10240,0.017530666457282174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,3584,0.020046222541067336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,8192,0.013789333403110504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,8192,0.02113155523935954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,8192,0.015767110718621146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,7168,0.029186666011810303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,7168,0.012738666600651212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,7168,0.019166222876972623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,2048,0.016917333006858826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,7168,0.013596444494194455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,6144,0.01163200040658315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,6144,0.017674666312005784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,6144,0.01238933371173011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,5120,0.02344088918632931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,5120,0.010012444522645738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,5120,0.015750222735934787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,1024,0.014811555544535318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,5120,0.011359111302428775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,4096,0.02048177851570977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,4096,0.008767111433876885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,4096,0.013059555656380124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,4096,0.010669333239396414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,3584,0.01920355525281694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,3584,0.007879999776681265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,3584,0.011957333319716983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,3584,0.01000355515215132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,3072,0.007214222517278459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,3072,0.011002666420406766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,3072,0.009359111388524374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,768,0.016869333055284288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,2560,0.00629333323902554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,2560,0.010039110978444418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,2560,0.008784000244405534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,768,512,0.013778666655222574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,2048,0.005727111051479976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,2048,0.009296889106432596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,2048,0.00815555536084705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,1536,0.015844444433848064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,1536,0.004862222406599256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,1536,0.008609777523411645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,1536,0.007311111523045435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,16384,0.04913155568970574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,1024,0.004294222427739037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,1024,0.007647111184067196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,1024,0.006694222076071634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,768,0.01643111142847273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,768,0.004192000048028098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,768,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,768,0.00655377780397733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,512,0.013418667018413544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,512,0.0036657779581016968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,512,0.0069848886794514125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,512,0.006255110932721033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,10240,0.035368889570236206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,8192,0.031839112440745033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,256,0.003544888976547453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,256,0.006626666833957036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,256,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,128,0.0032017777363459268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,128,0.006636444479227066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,512,128,0.005929777605666055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,64,0.0031715554909573663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,64,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,512,32,0.0032177778581778207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,512,32,0.006616000086069107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,65536,0.14770844247606066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,65536,0.03174044357405768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,65536,0.12067911359998916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,65536,0.08084444204966228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,6144,0.025241777300834656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,16384,0.014373333917723762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,16384,0.034584889809290566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,16384,0.02379200028048621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,12288,0.03836177786191305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,12288,0.012776888906955719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,12288,0.027756444282001917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,12288,0.019543111324310303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,2560,0.01724799970785777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,10240,0.011674666570292579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,10240,0.02426044477356805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,3072,0.01843644513024224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,10240,0.017473777135213215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,8192,0.010888000329335531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,8192,0.02067377832200792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,8192,0.015139554937680563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,7168,0.02659911082850562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,7168,0.010327999790509542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,7168,0.01923733287387424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,7168,0.013408000270525614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,6144,0.024151111642519634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,2048,0.016873778568373788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,6144,0.009309333231714036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,6144,0.017376888129446242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,6144,0.012378666963842181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,5120,0.008400888906584846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,5120,0.014755555325084262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,5120,0.011397333608733283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,4096,0.019592000378502738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,4096,0.00757155567407608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,4096,0.013156444662147097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,4096,0.010662222074137794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,1024,0.014129777749379476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,3584,0.018623999423450895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,3584,0.007279111279381647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,3584,0.011707555916574268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,3584,0.00997511131895913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,3072,0.006609777609507243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,3072,0.011002666420406766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,3072,0.009354666703277165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,256,0.015440000428093804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,2560,0.0058702222175068325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,2560,0.00998222248421775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,2560,0.008842666943868002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,512,128,0.013824889229403602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,2048,0.005212444398138259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,2048,0.009282666775915358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,16384,0.04651822315322029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,2048,0.008034666379292807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,1536,0.004871110949251386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,1536,0.008635555704434713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,1536,0.007263110743628607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,10240,0.03388977713055081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,1024,0.004179555508825514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,1024,0.007653333246707916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,1024,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,8192,0.030226667722066242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,768,0.003887999802827835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,768,0.007371555599901412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,768,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,512,0.012481778032249875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,512,0.0034906665484110513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,512,0.006950221955776215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,512,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,256,0.013125333521101209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,256,0.0032097777972618737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,256,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,5120,0.022239999638663396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,256,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,128,0.002847111059559716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,128,0.006918222126033571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,3072,0.01772711177666982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,256,128,0.005986666513813867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,64,0.0028284444577164124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,64,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,256,32,0.0028391112056043413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,256,32,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,65536,0.026702221896913316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,2560,0.01754666699303521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,65536,0.12102311187320286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,65536,0.0803262193997701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,16384,0.013607111242082385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,16384,0.034317334493001304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,16384,0.02352977792421977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,2048,0.01606755620903439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,12288,0.011754666765530905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,12288,0.027662222584088642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,12288,0.019243554936514962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,1024,0.014088888963063559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,10240,0.01055022246307797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,1536,0.014825777875052558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,10240,0.02439733346303304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,10240,0.017110221915774874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,8192,0.028532445430755615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,8192,0.009712888962692684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,8192,0.02092711130777995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,768,0.015351111690203348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,8192,0.014836443795098199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,7168,0.00868711123863856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,7168,0.01924711134698656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,7168,0.013145777914259167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,6144,0.007988444632954067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,6144,0.016529776983790927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,6144,0.012056888805495368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,65536,0.14608355363210043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,5120,0.007306666837798224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,5120,0.01482399966981676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,5120,0.011362666885058085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,256,128,0.01278133359220293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,4096,0.01919555498494042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,4096,0.007886221839321984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,4096,0.012705778082211813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,4096,0.010713777608341642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,3584,0.007300444775157505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,3584,0.012039110892348819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,12288,0.03832799858517117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,3072,0.01757777730623881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,3584,0.0099973330895106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,3072,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,3072,0.011056000159846412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,3072,0.009367999931176504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,16384,0.045781334241231285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,2560,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,2560,0.010054222411579555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,2560,0.008950221869680617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,2048,0.005294222384691238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,2048,0.009024000002278222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,2048,0.007974222302436829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,10240,0.033935109774271645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,1536,0.004588444613748127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,1536,0.008344000412358178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,1536,0.007272889216740926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,1024,0.013464888764752282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,1024,0.0041964443193541635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,1024,0.007605333295133378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,1024,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,7168,0.025694222913848028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,768,0.0035626664757728577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,768,0.007272889216740926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,768,0.006315555423498154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,512,0.012473777764373355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,5120,0.02107911143038008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,512,0.0031840000301599503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,512,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,512,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,256,0.011754666765530905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,256,0.0032044444233179092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,6144,0.02332177758216858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,256,0.006583111153708563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,256,0.005954666684071223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,128,0.002854222224818336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,128,0.00664533343580034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,768,128,128,0.00599911105301645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,64,0.002528888897763358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,64,0.006616000086069107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,128,32,0.0025475554996066624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,128,32,0.006699555450015598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,65536,0.025776000486479864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,65536,0.12070843908521865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,16384,0.011363555159833698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,16384,0.03439555565516154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,12288,0.009920888476901585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,10240,0.00998222248421775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,12288,0.027779555983013574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,10240,0.02436888880199856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,8192,0.00903911143541336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,8192,0.020576000213623047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,7168,0.008648888932334052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,7168,0.018795554836591084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,6144,0.007992888490358988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,6144,0.016140444411171805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,5120,0.007990222010347579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,5120,0.014507555299335055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,4096,0.007336888876226213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,4096,0.012455999851226807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,3584,0.006980444822046492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,3584,0.011705777711338468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,3072,0.006626666833957036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,3072,0.010947555303573608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,2560,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,2560,0.009992889232105678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,2048,0.00498933345079422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,2048,0.008966222405433655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,1536,0.0046453335218959385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,1536,0.008622222476535374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,1024,0.003858666867017746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,1024,0.007611555357774098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,768,0.0035386664999855887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,768,0.007256000406212277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,512,0.0032106666929192017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,512,0.006588444527652528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,256,0.0032186667538351486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,256,0.006583999842405319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,128,0.0026586666289303037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,128,0.006640888750553131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,3584,0.0181697772608863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,64,0.0028417776856157514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,64,0.006587555425034628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,64,32,0.0024844445288181305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,64,32,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,65536,0.022122666239738464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,16384,0.014481777946154276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,65536,0.11954843997955322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,16384,0.033891555335786604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,12288,0.010617777705192566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,12288,0.02737955583466424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,10240,0.009536888864305284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,10240,0.024030221833123103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,8192,0.008949333594905006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,8192,0.020234665936893888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,7168,0.008341333104504479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,7168,0.01850044396188524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,6144,0.0076142218377855085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,6144,0.01588888963063558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,5120,0.0069724445541699724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,5120,0.014478221535682678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,4096,0.007622222105662028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,4096,0.012366222010718452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,3584,0.007232888705200619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,3072,0.006664000037643645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,3584,0.01145066652033064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,3072,0.010844444235165914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,1536,0.01443288889196184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,2560,0.006628444625271692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,2560,0.010048000348938836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,2048,0.005197333378924264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,1536,0.008344000412358178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,2048,0.009080000221729279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,1536,0.004507555729813046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,2560,0.016763556334707472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,1024,0.003910222401221593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,1024,0.007637333538797166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,2048,0.015184000134468079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,768,0.003569777641031477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,768,0.007251555720965068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,512,0.0034737777378824023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,512,0.006635555376609166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,64,0.0030088888274298776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,256,0.0028844444702068963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,256,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,128,0.0028133332315418455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,128,0.006656888872385025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,64,0.006291555447710886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,768,32,32,0.0026231110095977783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,768,32,32,0.006648889018429651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,12288,0.5850213368733724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,16384,0.7969688839382596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,16384,0.4249573283725315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,16384,0.4604035483466254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,768,0.014101333088344999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,10240,0.2748062345716688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,12288,0.39295289251539445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,10240,0.4863662189907498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,10240,0.287155548731486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,8192,0.22738756073845756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,8192,0.38857600424024796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,8192,0.23057511117723253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,7168,0.3482213285234239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,7168,0.200525336795383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,768,128,128,0.011351111034552256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,6144,0.17978221840328643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,6144,0.3012648953331841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,6144,0.18092711766560873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,12288,0.3256524403889974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,5120,0.15522666772206625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,5120,0.2518640094333225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,5120,0.1504568921195136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,7168,0.20144089063008627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,4096,0.13695911566416422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,4096,0.19881867037879097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,4096,0.12352532810635036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,16384,0.23536178800794813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,3584,0.1255235539542304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,3584,0.17328266302744547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,12288,0.1803893380694919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,3584,0.10929422246085273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,10240,0.1518097718556722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,3072,0.15404089291890463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,3072,0.09751111268997192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,8192,0.12449244658152263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,7168,0.11118133862813313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,2560,0.13006222248077393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,2560,0.08355110883712769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,6144,0.09740977817111546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,3072,0.11580888430277507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,2048,0.1040595571200053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,2048,0.06942933135562472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,5120,0.0820924441019694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,2560,0.10628177722295125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,1536,0.07975822024875216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,1536,0.055686222182379834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,4096,0.06838399834103055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,3584,0.06103377872043186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,1024,0.05531733234723409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,1024,0.04200355543030632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,3072,0.053688890404171415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,2048,0.09695733255810207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,768,0.043587555487950645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,768,0.03892888956599765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,2560,0.047825776868396334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,512,0.06723555591371325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,512,0.03182755576239692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,512,0.03585066729121738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,1536,0.08715822299321492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,256,0.03330311179161072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,256,0.02103999919361538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,256,0.032863110303878784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,1024,0.07722577783796522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,768,0.043549332353803843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,128,0.0195448895295461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,128,0.03213511241806878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,2048,0.04041599896219041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,64,0.01668177710639106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,64,0.03087110983000861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,65536,32,0.01720266706413693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,65536,32,0.02989333205752903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,1536,0.03361599975162082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,65536,0.7912382019890679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,65536,0.5610266791449653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,1024,0.02682311170630985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,768,0.024023110667864483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,16384,0.20019377602471244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,16384,0.12233155303531223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,65536,128,0.030164443784289893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,65536,0.26132623354593915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,12288,0.1496995555029975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,12288,0.09232533640331692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,512,0.022108443909221228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,16384,0.06906488868925306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,10240,0.12961778375837538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,10240,0.07852800024880303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,65536,0.4454888767666287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,8192,0.07100177473492093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,8192,0.1002017789416843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,8192,0.062315556738111705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,256,0.020375111036830477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,12288,0.05326933330959744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,7168,0.0901519987318251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,7168,0.05617688761817085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,10240,0.045293334457609385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,65536,128,0.02038755516211192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,6144,0.07742310894860162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,16384,0.12030311425526936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,6144,0.049060444037119545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,5120,0.0528728895717197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,5120,0.06519911024305555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,5120,0.04182222154405382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,12288,0.09422133366266887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,8192,0.037556442949506975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,4096,0.052763554784986705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,4096,0.03454755412207709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,7168,0.03403910994529724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,3584,0.045251554912990995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,3584,0.046663112110561795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,3584,0.03083555565940009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,6144,0.03014755580160353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,3072,0.04229333334498935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,3072,0.04074222180578444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,3072,0.02770311137040456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,10240,0.08141689168082343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,2560,0.03943377733230591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,2560,0.034090667963027954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,2560,0.024461333950360615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,5120,0.026642666922675237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,4096,0.023698666029506262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,2048,0.02751733362674713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,2048,0.021707556313938562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,7168,0.06407733096016778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,1536,0.03394488824738397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,3584,0.021752888957659405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,1536,0.021357332666714985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,1536,0.018896889355447557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,3072,0.019711999429596793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,1024,0.015803555647532146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,1024,0.015815110670195687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,2560,0.01822933389080895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,768,0.025081778566042583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,768,0.013250666360060373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,768,0.014469333820872836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,6144,0.058725330564710826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,512,0.027781334188249376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,512,0.01051999959680769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,512,0.012294222083356647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,2048,0.016249777542220224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,256,0.022260444031821355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,256,0.00793333351612091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,4096,0.04814489020241631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,256,0.01070577816830741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,1536,0.01402666668097178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,128,0.006604444649484422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,128,0.010386666489972008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,1024,0.012067555553383298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,64,0.005923555543025334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,64,0.010024888647927178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,16384,32,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,16384,32,0.010407999985747868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,768,0.011352889239788055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,65536,0.3717973232269287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,512,0.010090666512648264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,65536,0.5974248780144585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,65536,0.35280089908176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,2048,0.03671911027696397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,16384,0.14895999431610107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,16384,0.09249955415725708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,12288,0.08377599716186523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,12288,0.11317511399586995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,256,0.009654222263230218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,12288,0.0704302191734314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,10240,0.07281333208084106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,10240,0.10503022538291083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,10240,0.059508442878723145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,16384,128,0.009710222482681274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,1024,0.030850665436850652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,8192,0.08511377705468072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,65536,0.22028799851735434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,8192,0.04883822136455112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,7168,0.05804977814356486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,7168,0.07487910985946655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,16384,0.055164443122016064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,7168,0.043459554513295494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,6144,0.05330488748020596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,6144,0.06433866421381633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,6144,0.038019554482565984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,12288,0.04286844531695048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,5120,0.047770665751563184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,5120,0.05338933401637607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,5120,0.03252799974547492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,10240,0.03602844476699829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,8192,0.030730668041441176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,4096,0.04379111197259691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,4096,0.027503111296229895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,7168,0.027468444572554693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,6144,0.02518400053183238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,16384,128,0.02017688916789161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,3584,0.038188444243537054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,3584,0.024969778127140466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,3072,0.0385155545340644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,3072,0.03297066688537598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,3072,0.022481777601771887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,5120,0.022105778257052105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,4096,0.02010933392577701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,16384,0.10476889212926228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,2560,0.0284186667866177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,2560,0.020284444093704224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,2048,0.03353244397375319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,2048,0.02264444364441766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,2048,0.017843556072976854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,3584,0.01852799952030182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,3072,0.016953777935769822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,1536,0.030364443858464558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,1536,0.01670222149954902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,1536,0.01611911091539595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,8192,0.06320799721611871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,1024,0.01387288918097814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,1024,0.012758221891191272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,2560,0.015451555450757345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,768,0.02405511173937056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,768,0.011702222128709158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,768,0.011695110963450538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,2048,0.014184888866212634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,1536,0.012351999680201212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,512,0.010031999813185798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,512,0.010109333528412713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,4096,0.043149332205454506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,1024,0.010816000401973724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,256,0.0076044441925154785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,256,0.009127111070685917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,768,0.009742221898502773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,128,0.018576888574494254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,128,0.006672888994216919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,128,0.009366222553782994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,3584,0.04047822289996677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,64,0.006285333385070165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,64,0.009840000006887648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,12288,32,0.006568000134494569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,512,0.009517333573765224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,12288,32,0.009872000250551436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,2560,0.035664889547559954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,65536,0.5217306878831651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,65536,0.32519910070631236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,16384,0.09201244513193767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,256,0.008632000121805403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,16384,0.12954221831427679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,16384,0.0834862192471822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,12288,0.07414133018917508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,12288,0.10093155834409927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,12288,128,0.008557333714432186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,12288,0.06384533643722534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,65536,0.18558043903774687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,10240,0.08464711242251927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,10240,0.054173333777321704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,1024,0.027792887555228338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,8192,0.05634578069051107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,16384,0.0510079993142022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,8192,0.068667557504442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,8192,0.04451466600100199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,12288,0.04032266802257962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,7168,0.06039466460545858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,7168,0.03977866636382209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,512,0.02455377744303809
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,10240,0.03445955448680454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,6144,0.05159644285837809
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,12288,256,0.021367111139827307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,6144,0.03506933318244086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,8192,0.029709332519107397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,5120,0.04337066743108961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,5120,0.03091999888420105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,7168,0.026260443859630164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,6144,0.023915555742051866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,4096,0.03439022103945414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,4096,0.026016000244352553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,5120,0.02141422198878394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,65536,0.3207102351718479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,3584,0.030448890394634668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,3584,0.024068444967269897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,4096,0.019254222512245178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,3072,0.034500443273120456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,3072,0.026214222113291424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,3072,0.021689777572949726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,10240,0.06425866815778944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,2560,0.03223377797338698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,2560,0.022229333718617756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,2560,0.019645333290100098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,3584,0.01791733337773217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,2048,0.03028977910677592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,2048,0.018224888377719455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,2048,0.01748533381356133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,7168,0.05182488759358724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,3072,0.01647822227742937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,1536,0.014383999837769402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,1536,0.015455111861228943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,6144,0.0476382209195031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,2560,0.014825777875052558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,1024,0.01107377807299296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,1024,0.012811555630630918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,2048,0.013403555585278405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,5120,0.04282311267322964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,768,0.009686222506894005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,768,0.010974222587214576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,4096,0.039001779423819646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,512,0.0229066660006841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,512,0.007711999946170383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,512,0.009704888694816166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,1536,0.011739555332395764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,1024,0.010344888601038191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,256,0.006278222219811545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,256,0.009342222577995723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,3584,0.03684355484114753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,768,0.009361777868535783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,512,0.008812444077597724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,128,0.005535111245181825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,128,0.008966222405433655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,64,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,64,0.008658666577604082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,10240,32,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,10240,32,0.00885688927438524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,256,0.008255999949243333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,65536,0.40631998909844297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,65536,0.23970667521158853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,10240,128,0.007648000286685095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,1536,0.0277502222193612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,16384,0.10339288579093085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,16384,0.06409244404898749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,1024,0.025118221839269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,65536,0.14350844754113093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,12288,0.07806399795744154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,12288,0.049695110983318753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,16384,0.03968800107638041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,768,0.02260622216595544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,10240,0.06532533301247491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,10240,0.041834665669335254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,256,0.020170667105250888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,8192,0.04656889041264852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,8192,0.051932444175084434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,8192,0.03399822115898132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,12288,0.0317813323603736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,7168,0.04290489024586148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,7168,0.04617599977387322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,7168,0.03088266650835673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,10240,0.02739288906256358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,6144,0.0395688878165351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,6144,0.039637333816952176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,6144,0.02740977704524994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,10240,128,0.01742044422361586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,5120,0.03644444545110067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,8192,0.023810666468408372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,5120,0.03363022208213806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,5120,0.024029332730505202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,7168,0.02159022291501363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,4096,0.02740266587999132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,4096,0.02068533334467146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,6144,0.01965777741538154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,65536,0.24848532676696777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,3584,0.023759111762046814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,3584,0.01923822197649214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,5120,0.01772177716096242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,4096,0.01611288885275523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,3072,0.020995555652512446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,3072,0.01753777762254079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,16384,0.07225600216123793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,3584,0.014818666709793938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,2560,0.017840888765123155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,2560,0.016004444824324716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,12288,0.06048266755210029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,2048,0.026711111267407734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,2048,0.015419556034935845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,2048,0.014514666464593677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,3072,0.01388711151149538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,1536,0.025378665990299646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,1536,0.01236533373594284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,1536,0.012800888882742988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,2560,0.012679110798570844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,10240,0.052573333183924355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,1024,0.0099982221921285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,1024,0.0103511114915212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,2048,0.011947555674446953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,768,0.02204088866710663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,768,0.008658666577604082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,768,0.009357333183288574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,1536,0.01034400032626258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,1024,0.008462222086058723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,512,0.0069226668112807805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,512,0.008278222547637092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,768,0.008278222547637092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,4096,0.03361599975162082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,256,0.005246222019195557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,256,0.00795288880666097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,512,0.007368889119890001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,3584,0.03155466583040025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,256,0.007080888582600488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,128,0.004810666872395409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,128,0.00793333351612091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,64,0.004485333131419288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,64,0.007656888829337225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,8192,32,0.004524444540341695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,8192,32,0.007624000310897827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,3072,0.030024889442655776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,65536,0.37339199913872617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,65536,0.23111377822028267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,2560,0.028559111886554297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,16384,0.07005155748791166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,16384,0.09624177879757351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,65536,0.22907378938463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,16384,0.060586664411756724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,8192,128,0.007089777953094906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,12288,0.07225955857170953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,12288,0.045900444189707436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,65536,0.1347279946009318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,16384,0.03882044553756714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,10240,0.06061155266231961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,10240,0.0392213331328498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,1024,0.023620444867346022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,12288,0.03126400046878391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,8192,0.048546665244632296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,512,0.020261334048377145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,8192,0.03273866573969523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,10240,0.02693422304259406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,7168,0.04323644439379374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,7168,0.029476443926493328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,8192,0.023507555325826008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,256,0.01771377854877048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,6144,0.037267555793126426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,6144,0.026941332552168105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,7168,0.021027555068333943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,5120,0.035065776771969266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,5120,0.03160888950030009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,5120,0.023748444186316595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,6144,0.01940533353222741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,8192,128,0.01680888897842831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,4096,0.03250222073660956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,4096,0.025732444392310247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,4096,0.020591111646758187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,5120,0.017515555024147034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,3584,0.022999111149046157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,3584,0.018926221463415358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,4096,0.015683554940753512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,3072,0.029135998752382066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,3072,0.020582222276263766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,3072,0.01742755538887448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,12288,0.05708711014853584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,3584,0.014883556299739413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,2560,0.017948443690935772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,2560,0.015979554918077257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,3072,0.013645333548386892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,10240,0.05043466554747688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,2048,0.01576266686121623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,2048,0.014354666074117025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,2560,0.012741333080662621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,7168,0.041542222102483116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,1536,0.01310400002532535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,1536,0.01237777786122428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,8192,0.045138667027155556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,2048,0.011318222516112857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,1024,0.01035644445154402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,1024,0.009750222166379293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,6144,0.03861422340075175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,1536,0.01019377758105596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,768,0.008967111508051554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,768,0.008826666408114964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,1024,0.008091555701361762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,768,0.007808888951937358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,3584,0.030674666166305542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,512,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,512,0.008271999657154083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,256,0.01755288905567593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,256,0.0058293334311909145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,256,0.007655999726719326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,512,0.007310222420427535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,128,0.016473778420024447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,128,0.004882666799757215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,128,0.007649777664078607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,256,0.007019555403126611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,64,0.004879999905824661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,64,0.0075866662793689305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,2560,0.027408000495698717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,7168,32,0.006250666661394968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,7168,32,0.007648888561460707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,65536,0.3282071219550239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,65536,0.222455104192098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,7168,128,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,2048,0.026154667139053345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,16384,0.0825608902507358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,16384,0.05639644463857015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,65536,0.1303457816441854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,12288,0.05420533484882779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,12288,0.06343377961052789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,12288,0.04379822148217095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,1536,0.024288889434602525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,10240,0.04773244592878553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,10240,0.052448001172807485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,10240,0.03818133473396301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,16384,0.03788000014093187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,1024,0.022617777188618977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,8192,0.04210844304826525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,8192,0.03253422180811564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,12288,0.030223998758527968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,7168,0.03936266567971971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,768,0.020966221888860066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,7168,0.0378479990694258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,7168,0.0291920006275177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,6144,0.036659555302725896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,6144,0.03237333231502109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,6144,0.026595556073718604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,10240,0.02643199927277035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,5120,0.033435556623670794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,5120,0.027108444107903376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,5120,0.023499555057949487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,8192,0.022988445229000513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,7168,512,0.018764444523387484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,7168,0.020378667447302077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,4096,0.02237066626548767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,4096,0.020181333025296528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,6144,0.01885511146651374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,3584,0.019582221905390423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,3584,0.018915555543369718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,5120,0.016920000314712524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,4096,0.015568888849682279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,3072,0.01735111077626546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,3072,0.017367111312018502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,3584,0.014512000812424554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,65536,0.21256711747911242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,2560,0.015452444553375244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,3072,0.01350844403107961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,2560,0.015934222274356417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,16384,0.06641422377692328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,2048,0.013381332986884646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,2048,0.014126222994592456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,2560,0.0124177775449223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,1536,0.02292177743381924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,1536,0.010404444403118556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,1536,0.012080888781282636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,2048,0.011018666956159802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,1024,0.020445333586798776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,1024,0.008980444735950893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,1024,0.009663999908500248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,1536,0.009332444104883406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,8192,0.0425893333223131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,768,0.007552000383536021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,768,0.009007111191749573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,1024,0.008060444560315874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,768,0.007423111134105259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,512,0.006250666661394968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,512,0.008277333445019191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,3584,0.028939555088678997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,4096,0.030794666873084173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,256,0.0052977779673205475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,256,0.008003555238246918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,512,0.007257777783605788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,3072,0.027668444646729365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,128,0.004882666799757215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,128,0.007674666742483775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,2560,0.026364444030655756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,64,0.004524444540341695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,256,0.0069573331210348345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,64,0.007646222081449296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,6144,32,0.004612444589535396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,6144,32,0.0075911109646161394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,65536,0.2662693394554986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,65536,0.18916711542341444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,65536,0.2038346661461724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,16384,0.06348799996905856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,6144,128,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,16384,0.06645333104663424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,16384,0.05503822366396586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,2048,0.024551999237802293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,12288,0.05097600155406528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,12288,0.0495742228296068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,12288,0.04314133193757799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,65536,0.12672711743248835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,10240,0.042790220843421094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,10240,0.037104888094796076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,16384,0.036766221125920616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,8192,0.039099554220835366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,12288,0.029492444462246362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,8192,0.03454577922821045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,8192,0.03170844581392076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,10240,0.025404444999165002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,7168,0.03030311067899068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,7168,0.028571556011835735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,8192,0.022439110610220168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,7168,0.019891555110613506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,6144,0.026176000634829204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,6144,0.025711999999152288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,768,0.01927288870016734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,5120,0.03096710973315769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,5120,0.02182933357026842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,5120,0.023031999667485554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,512,0.017507554756270517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,4096,0.028655999236636694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,4096,0.017785777648289997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,4096,0.020106666617923312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,6144,0.018223111828168232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,3584,0.027220444546805486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,3584,0.016005333926942613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,3584,0.018329777651362948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,5120,0.016724444097942777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,3072,0.02591466738118066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,3072,0.014239999983045789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,3072,0.01699022286468082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,4096,0.01521511044767168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,2560,0.024658666716681585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,2560,0.012422222230169507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,256,0.0162471118900511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,2560,0.01514844430817498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,3584,0.013983110586802164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,2048,0.010850666297806634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,2048,0.013437333206335703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,3072,0.01311288856797748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,6144,128,0.016199111938476562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,1536,0.02055022286044227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,1536,0.009296000003814697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,1536,0.011432888607184092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,2560,0.011727111207114326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,2048,0.010126222338941362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,1024,0.007632888853549957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,1024,0.009405333134863112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,1536,0.008596444295512306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,768,0.018922666708628338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,768,0.0062871111763848206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,768,0.008614222208658854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,10240,0.04410399993260702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,512,0.005227555417352253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,512,0.008280889027648503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,1024,0.007667555577225155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,256,0.016192000773217943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,256,0.004599110947714912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,256,0.007656888829337225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,768,0.007261333366235097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,512,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,128,0.0041706665522522396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,128,0.007594666547245449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,256,0.006638222270541721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,7168,0.03636444608370463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,64,0.003914666672547658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,64,0.007676444119877285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,5120,32,0.0041751112374994485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,5120,32,0.007641777396202087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,65536,0.22042489051818848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,65536,0.17729600270589194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,6144,0.03386488887998793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,65536,0.1342408921983507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,16384,0.05696800020005968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,5120,128,0.006686222222116258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,16384,0.05493333273463779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,16384,0.0365555551317003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,12288,0.04256888892915514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,12288,0.02887466549873352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,65536,0.08310488859812419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,16384,0.026351999905374315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,10240,0.035798221826553345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,10240,0.026142223013771906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,12288,0.02165511084927453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,10240,0.019151111443837483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,8192,0.029119110769695703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,8192,0.022154667311244543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,2048,0.023303111394246418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,8192,0.01684888866212633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,7168,0.025763556361198425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,7168,0.02030311193731096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,7168,0.01534400052494473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,6144,0.0317520002524058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,1024,0.01886311173439026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,6144,0.022104000051816303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,6144,0.01869422197341919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,6144,0.014347554908858405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,512,0.016559999850061204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,5120,0.018728888697094388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,5120,0.01676977839734819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,4096,0.027408000495698717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,4096,0.015676443775494892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,4096,0.014488889111412896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,5120,0.013365333278973898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,3584,0.025456888808144465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,3584,0.014409777190950183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,3584,0.013782222237851886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,4096,0.011981333295504252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,3072,0.02420177724626329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,3072,0.012712000144852532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,5120,128,0.0155404441886478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,3072,0.012776888906955719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,2560,0.023363555471102398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,2560,0.011114666859308878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,2560,0.011433777709801992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,3584,0.011409777734014722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,3072,0.010049777726332346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,2048,0.010063110954231685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,2048,0.010060444474220276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,2560,0.008947555389669206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,12288,0.04677155613899231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,2048,0.008108444511890411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,1536,0.00796888851457172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,1536,0.008791999684439765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,8192,0.03715822100639343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,1024,0.006232000059551663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,1024,0.00775822252035141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,10240,0.041209777196248375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,1536,0.007296000089910295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,768,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,768,0.007430222299363878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,1024,0.0069164443347189165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,512,0.015850666496488783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,512,0.004875555634498596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,512,0.0069644442862934535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,768,0.006551110910044775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,256,0.016185777054892648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,256,0.004250666747490565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,256,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,7168,0.03439022103945414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,512,0.006272888845867581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,128,0.01579466627703773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,128,0.003916444463862313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,128,0.006587555425034628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,256,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,64,0.003522666792074839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,64,0.00664533343580034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,4096,32,0.003843555433882607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,4096,32,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,4096,128,0.006220444622966979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,65536,0.1998951170179579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,5120,0.028791109720865887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,65536,0.12693956163194445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,16384,0.05520266625616285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,16384,0.050570666790008545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,16384,0.03462488783730401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,12288,0.045553776952955455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,12288,0.03889244463708665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,12288,0.0277804434299469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,65536,0.0809662209616767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,16384,0.025646222962273493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,10240,0.03209244542651706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,10240,0.024837333295080397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,12288,0.021087111698256597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,10240,0.01867733399073283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,8192,0.02609066665172577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,8192,0.021592888567182753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,2048,0.02124711043304867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,8192,0.016340444485346477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,7168,0.023017777336968318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,7168,0.01979733341270023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,1536,0.018886221779717337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,7168,0.015063999427689446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,6144,0.019887111253208585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,6144,0.018005333013004728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,1024,0.016846223009957206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,5120,0.02845066785812378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,5120,0.01706044375896454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,5120,0.016521778371598985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,6144,0.013992889059914483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,4096,768,0.018072000808186002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,4096,0.014569777581426831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,4096,0.014270222849316068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,5120,0.012823111481136747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,4096,0.011518222590287527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,3584,0.013096888860066732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,3584,0.013556444810496436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,3584,0.010998222563001843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,65536,0.17305866877237955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,3072,0.011749332977665795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,3072,0.01256533298227522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,10240,0.04092622134420607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,3072,0.009992000129487779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,2560,0.010536889235178629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,2560,0.011391999820868174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,8192,0.035975111855400935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,2560,0.008978666530715095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,2048,0.009343110852771336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,2048,0.009881777895821465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,7168,0.033787555164761014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,2048,0.007967111137178209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,1536,0.00796088907453749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,1536,0.0086986662613021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,1536,0.007295110987292395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,6144,0.03144977821244134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,1024,0.006280888699822956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,1024,0.007947555846638149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,1024,0.006607999818192587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,768,0.018076444665590923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,768,0.006008000009589725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,768,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,4096,0.02675466736157735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,768,0.006311111152172089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,512,0.004883555488453971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,512,0.006903111106819577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,3584,0.025059555967648823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,256,0.015847999188635085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,256,0.004221333397759331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,256,0.006622222148709827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,512,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,128,0.015792888071801927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,128,0.0038853333228164246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,128,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,256,0.00601244428091579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,64,0.0035297779573334586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,64,0.006603555546866522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3584,32,0.003673777812057071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3584,32,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,65536,0.15280977884928384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3584,128,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,65536,0.17603554990556505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,65536,0.11789955033196343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,3072,0.02351644469632043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,16384,0.04957333207130432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,16384,0.044870220952563815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,16384,0.033610665135913424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,65536,0.08436266581217448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,12288,0.03405066662364536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,16384,0.025421332981851365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,12288,0.029751999510659113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,2560,0.021698666943444148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,10240,0.03698844379848904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,10240,0.029121776421864826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,10240,0.024655999408827886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,12288,0.020830222302012973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,8192,0.03304266598489549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,8192,0.0236791107389662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,8192,0.020883555213610332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,2048,0.019903111788961623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,10240,0.018368888232443068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,7168,0.020831111404630873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,7168,0.019129777948061626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,8192,0.016149333781666227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,6144,0.018156444032986958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,6144,0.01775911119249132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,7168,0.014867555763986377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,1536,0.017893332574102614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,5120,0.026692445079485577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,5120,0.015666666958067153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,5120,0.016338666280110676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,6144,0.013808888693650564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,4096,0.013439999686347114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,4096,0.01408622165520986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,5120,0.012758221891191272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,1024,0.016487111647923786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,3584,0.012055111428101858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,3584,0.012907555533779992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,4096,0.011185777684052786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,3584,0.010292444792058732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,3072,0.011043555206722684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,3072,0.011746666497654386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,3072,0.009331555830107795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3584,512,0.01550222271018558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,2560,0.009955555200576782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,2560,0.010737777584128909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,2560,0.00831288927131229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,12288,0.04150222076310052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,2048,0.008287111090289222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,2048,0.009680888719028896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,2048,0.007631111476156447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,1536,0.01716977854569753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,1536,0.007296000089910295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,1536,0.008625778059164682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,7168,0.03071199854214986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,1024,0.015802666544914246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,1536,0.00693866651919153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,1024,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,1024,0.007616000043021307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,768,0.01752000053723653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,768,0.005214222189452913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,768,0.007439110842016008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,4096,0.025055999557177227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,1024,0.006567111031876669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,512,0.00452888881166776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,512,0.0069066666894488865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,768,0.006294222341643439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,6144,0.028747555282380845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,256,0.004232888834344016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,256,0.006605333338181178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,3584,0.023002665903833177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,512,0.006226666685607698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,128,0.0041991112132867174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,128,0.006558222075303395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,256,0.006295111030340195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,64,0.003823111040724648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,64,0.006254222244024277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,3072,32,0.0038684445122877755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,3072,32,0.006632888896597757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,3072,128,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,65536,0.15769155820210776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,3072,0.02197422252760993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,65536,0.11221067110697429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,16384,0.048271109660466514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,16384,0.04052977760632833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,16384,0.033131556378470525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,65536,0.07792888747321235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,2560,0.020861332615216572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,12288,0.030171556605233088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,12288,0.02720444401105245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,16384,0.0248026665714052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,12288,0.02048711147573259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,10240,0.025341333614455328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,10240,0.024298666252030268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,10240,0.018054222067197163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,2048,0.01925511161486308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,8192,0.02083200050724877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,8192,0.02071822186311086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,8192,0.01593155496650272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,7168,0.03053511182467143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,7168,0.018389332625601027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,7168,0.019128888845443726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,7168,0.014458666245142618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,6144,0.02851466668976678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,6144,0.016344888342751395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,6144,0.01742755538887448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,6144,0.01347555551264021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,512,0.0148053334818946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,5120,0.014142221874660917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,5120,0.016034666034910414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,128,0.015430221954981485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,3072,256,0.016151999433835346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,5120,0.01239822225438224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,4096,0.012688889271683164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,4096,0.013951111170980664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,3584,0.02296177711751726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,3584,0.01129422254032559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,4096,0.011013333168294696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,3584,0.012870222330093384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,65536,0.14892266856299505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,3072,0.010434666441546546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,3072,0.011415999796655444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,3584,0.010086221827401055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,3072,0.009024889104896123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,2560,0.009335999687512716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,2560,0.010610666539933948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,10240,0.03598311212327745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,2048,0.01792266633775499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,2048,0.00794400026400884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,12288,0.0403733319706387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,2048,0.009315555294354757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,2560,0.008351999852392409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,1536,0.006976889239417181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,1536,0.008430221842394935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,2048,0.007592889169851939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,1024,0.015496888094478183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,1024,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,1024,0.007344889144102733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,1536,0.007174222005738153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,8192,0.032807999187045626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,768,0.004541333350870344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,768,0.007270221908887227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,1024,0.006634666687912411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,768,0.006575999988449945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,512,0.004222222086456087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,512,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,512,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,5120,0.026220444175932143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,256,0.0038897775941424896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,256,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,256,0.005927111125654644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,128,0.015496888094478183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,128,0.003572444534964032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,128,0.006763555523422029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,4096,0.023641778363121882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,64,0.003499555504984326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,64,0.006304889089531369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2560,32,0.003522666792074839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2560,32,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2560,128,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,65536,0.11674133936564128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,65536,0.11008889145321316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,3072,0.021635555558734473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,16384,0.04759644468625387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,16384,0.03631022241380479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,16384,0.03297955460018582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,65536,0.07654133107927111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,2560,0.01908888916174571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,12288,0.027956446011861164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,12288,0.02696977721320258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,16384,0.024306666519906785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,12288,0.020284444093704224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,10240,0.023489778240521748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,10240,0.024013333850436743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,1536,0.01648177703221639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,10240,0.017722666263580322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,8192,0.019632889164818656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,8192,0.02053777707947625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,8192,0.015856888559129503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,7168,0.029465778006447688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,7168,0.017809778451919556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,7168,0.01882933411333296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,768,0.016534222496880423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,7168,0.014193778236707052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,512,0.014481777946154276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,6144,0.01588355501492818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,6144,0.017186666528383892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,5120,0.02502577834659153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,5120,0.013795555465751223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,5120,0.01585422290696038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,6144,0.013099555340078143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,4096,0.023296000228987798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,4096,0.011914666328165265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2560,256,0.015832000308566622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,4096,0.013515555196338229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,5120,0.011734222372372946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,3584,0.01074044406414032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,3584,0.01223288890388277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,4096,0.0099982221921285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,3584,0.009337777892748514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,3072,0.009932444327407414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,3072,0.011409777734014722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,3072,0.008744888835483128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,2560,0.017904000149832833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,2560,0.00867022242810991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,2560,0.010025777750545079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,2560,0.00832444429397583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,65536,0.14374666743808323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,2048,0.007532444265153673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,2048,0.008979555633332994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,10240,0.035192890299691096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,12288,0.03966844413015578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,1536,0.0162453336848153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,1536,0.006263999889294307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,1536,0.008052444292439355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,2048,0.007641777396202087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,1024,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,1024,0.007669332954618666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,1536,0.006984000404675801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,8192,0.03139111068513658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,768,0.004869333157936732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,768,0.007245333658324347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,1024,0.006567111031876669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,768,0.00620977787507905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,512,0.0038933331767717996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,512,0.0069208890199661255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,512,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,256,0.015438222222858004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,256,0.004152889053026835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,256,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,256,0.005700444595681296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,6144,0.027154665854242113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,128,0.003843555433882607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,128,0.006573333508438534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,3584,0.021920889616012573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,64,0.003499555504984326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,64,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,2048,32,0.0035982223020659555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,2048,32,0.006225777582989798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,2048,128,0.005966222120655908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,65536,0.0941351122326321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,3072,0.019894222418467205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,65536,0.10874133639865452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,16384,0.04628888765970866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,16384,0.02846222122510274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,16384,0.03263377812173631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,65536,0.07497777541478474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,16384,0.023997333314683702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,12288,0.02091466718249851
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,12288,0.02679288884003957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,12288,0.019816888703240287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,2048,0.016872000363137987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,10240,0.018057778477668762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,10240,0.023354666100607976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,10240,0.0173973325226042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,1024,0.015441776977645027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,8192,0.015000889698664347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,8192,0.02027466727627648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,8192,0.015463999576038785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,7168,0.029139555162853662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,7168,0.013854222165213691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,7168,0.01863466699918111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,512,0.014447111222479077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,768,0.0174026671383116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,6144,0.012552888856993781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,6144,0.0169813334941864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,7168,0.013723555538389417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,6144,0.012207110722859701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,5120,0.011052444577217102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,5120,0.015506666567590503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,5120,0.01108000013563368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,4096,0.021057777934604224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,4096,0.00998222248421775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,4096,0.012785777449607849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,2048,128,0.0147733340660731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,3584,0.020584889584117465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,3584,0.008969777988062965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,4096,0.009970666633711921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,3584,0.01147555559873581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,65536,0.14204711384243435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,3584,0.009371555513805812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,3072,0.008270222279760573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,3072,0.010652444428867765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,3072,0.008672888908121321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,2560,0.007259555160999298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,2560,0.009705777797434065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,12288,0.03908088803291321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,2560,0.008275555239783393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,2048,0.006274666637182236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,2048,0.008651555412345463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,2048,0.00759733302725686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,10240,0.03433777888615926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,1536,0.005584888988071018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,1536,0.008005333443482717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,1536,0.006992888947327931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,1024,0.014514666464593677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,1024,0.004916444420814514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,1024,0.0075902218619982404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,1024,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,768,0.016529776983790927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,8192,0.030778666337331135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,768,0.0042239998777707415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,768,0.007316444483068254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,512,0.013759999639458127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,512,0.003903111235962974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,512,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,768,0.006241777704821692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,256,0.015497777197096082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,256,0.003569777641031477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,256,0.006303999986913469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,512,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,256,0.0057555557125144536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,128,0.003552000141806073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,128,0.006604444649484422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1536,128,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,64,0.0033537778589460584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,64,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1536,32,0.0035208890007601846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1536,32,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,5120,0.024639111426141527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,6144,0.026820444398456152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,65536,0.06305511130226983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,65536,0.10798666874567668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,16384,0.04356266723738777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,16384,0.020044444335831534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,65536,0.07338844405280219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,16384,0.031983110639784075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,12288,0.03682400120629205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,12288,0.01963911122745938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,12288,0.025983111725913152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,16384,0.023401778605249193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,10240,0.032794667614830866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,10240,0.01677688956260681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,10240,0.022974222898483276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,12288,0.01937244501378801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,10240,0.01714933415253957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,8192,0.014448000325096978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,8192,0.01981511049800449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,3072,0.018242667118708294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,8192,0.01481066644191742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,7168,0.012995555169052549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,7168,0.018509333332379658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,7168,0.012537777423858643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,2560,0.017909333109855652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,6144,0.011952888634469775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,6144,0.01676977839734819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,2048,0.0169831116994222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,5120,0.022091555926534865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,5120,0.01036266651418474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,5120,0.014771555860837301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,6144,0.011409777734014722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,5120,0.010670222342014313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,4096,0.008967999782827165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,1536,0.016018667154841952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,4096,0.012137778103351593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,4096,0.009904888768990835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,3584,0.008328888979223039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,3584,0.011015999648306104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,3584,0.009307555854320526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1536,128,0.014800889624489678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,3072,0.0075902218619982404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,3072,0.01037333326207267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,3072,0.008646222452322641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,65536,0.13125599755181208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,2560,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,2560,0.009679999616410997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,2560,0.008061333662933773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,2048,0.015968888998031616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,2048,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,2048,0.008960888617568547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,2048,0.007606222397751278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,1536,0.015505777464972602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,1536,0.005239999956554837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,1536,0.007992888490358988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,8192,0.02922399838765462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,1024,0.01477688882086012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,1024,0.004531555705600315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,1536,0.006966222491529252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,1024,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,7168,0.027152889304690894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,1024,0.006585777633719974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,768,0.003863111138343811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,768,0.007304000357786815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,6144,0.024329778220918443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,768,0.006249777972698212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,512,0.003656889001528422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,512,0.0069066666894488865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,256,0.015126221709781222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,256,0.0034959999223550162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,256,0.0064604443808396654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,512,0.005919111271699269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,256,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,128,0.0031635556370019913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,128,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,1024,128,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,64,0.0028391112056043413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,64,0.006568000134494569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,3584,0.018774222996499803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,1024,32,0.0031644445326593188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,1024,32,0.006581333362393909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,65536,0.048671109808815845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,65536,0.12809244791666666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,4096,0.020058666666348774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,65536,0.10847110880745782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,16384,0.04323200053638882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,16384,0.01811733345190684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,16384,0.03190755512979295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,65536,0.07174577977922228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,12288,0.03668533431159125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,12288,0.016016888949606154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,12288,0.025936888323889837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,16384,0.023174222972657945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,3072,0.017673777209387887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,10240,0.03261955579121908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,10240,0.015328889091809591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,10240,0.022943110929595098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,12288,0.01920355525281694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,8192,0.013929777675204806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,10240,0.016921778519948322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,8192,0.01958311100800832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,8192,0.014210666219393412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,7168,0.01276266657643848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,7168,0.01828177769978841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,7168,0.012441777520709567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,6144,0.01163733336660597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,6144,0.015809777710172866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,6144,0.011424889167149862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,5120,0.020630222227838304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,5120,0.010036444498433007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,5120,0.014105778601434497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,5120,0.010504888991514841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,4096,0.019556444552209642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,4096,0.00850044439236323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,4096,0.01204177737236023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,4096,0.009769777456919352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,2560,0.01703999936580658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,3584,0.007676444119877285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,3584,0.010840888652536603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,3584,0.009352000223265754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,768,0.016542222764756944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,3072,0.0069617778062820435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,3072,0.010031110710567897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,3072,0.008616000413894653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,128,0.015452444553375244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,2560,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,1024,512,0.013131555583741931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,2560,0.009651555783218807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,2048,0.016515556308958266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,2048,0.005716444303592046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,2048,0.008839999636014303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,2560,0.008009778128729926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,2048,0.0074720001882976955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,1536,0.005183111048407025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,1536,0.007972444097201029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,1536,0.0069422221018208405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,8192,0.02882755464977688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,1024,0.0041982221106688184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,1024,0.007313778003056844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,1024,0.006575111299753189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,6144,0.02445777753988902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,768,0.004192000048028098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,7168,0.025957332717047796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,768,0.006963555183675554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,768,0.006171555568774541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,512,0.0038746665749284956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,512,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,512,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,3584,0.018233777748213876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,256,0.0035457776652442086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,256,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,256,0.00565155554148886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,3072,0.01816177699300978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,128,0.003212444484233856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,128,0.006642666541867786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,768,128,0.00563466673096021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,64,0.0031848889258172777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,64,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,768,32,0.003400000019205941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,768,32,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,2560,0.017394666870435078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,65536,0.035755554835001625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,65536,0.10873599847157796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,65536,0.07260088788138495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,1024,0.014428445034556918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,16384,0.015895111693276297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,1536,0.015174221661355762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,16384,0.031705776850382485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,16384,0.022882666852739122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,12288,0.013782222237851886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,12288,0.025855110751258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,12288,0.01866933372285631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,10240,0.030758221944173176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,10240,0.015052444405025907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,10240,0.023019555542204116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,10240,0.015862221519152325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,768,0.01626311077011956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,8192,0.013912889692518445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,8192,0.019328888919618394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,8192,0.013776889277829064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,512,0.013400000002649097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,7168,0.024805333879258897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,7168,0.012358222570684222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,7168,0.0177208897140291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,7168,0.012366222010718452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,6144,0.011238222320874533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,6144,0.015438222222858004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,6144,0.011387555963463254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,256,0.015506666567590503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,5120,0.009365333451165093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,5120,0.013454222016864352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,768,128,0.014126222994592456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,5120,0.01051999959680769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,4096,0.008011555506123437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,4096,0.012090666426552666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,4096,0.009704888694816166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,3584,0.01824977828396691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,3584,0.007328888608349695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,3584,0.011038222246699862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,3584,0.009226666556464301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,65536,0.12460177474551731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,3072,0.0069617778062820435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,3072,0.010025777750545079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,3072,0.008515555825498369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,16384,0.04159466756714715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,2560,0.006248888870080312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,2560,0.009377777576446533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,12288,0.03513866662979126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,2048,0.015556444724400839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,2048,0.00553688903649648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,2560,0.007990222010347579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,2048,0.008657777474986183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,1536,0.0147706667582194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,1536,0.004871110949251386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,2048,0.007263999846246507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,1536,0.007968000239796108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,1024,0.013801777528391944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,1024,0.004181333300140169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,1536,0.006805333412355847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,1024,0.007287999822033777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,1024,0.0063182223174307085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,768,0.0038533334930737815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,768,0.0069919998447100324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,768,0.005959110955397288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,6144,0.022483555807007685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,8192,0.026268444127506677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,512,0.003543111185232798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,512,0.00665866666369968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,256,0.014148443937301636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,256,0.0032302221904198327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,512,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,256,0.006592888798978593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,5120,0.020226667324701946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,128,0.0028088889602157804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,128,0.006618666566080517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,256,0.005728000154097875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,64,0.00282577777074443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,64,0.00656444455186526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,512,32,0.0028275555620590844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,512,128,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,512,32,0.00629688882165485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,65536,0.02574133376280467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,65536,0.10877333084742229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,65536,0.07124622000588311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,16384,0.04083555605676439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,16384,0.011973333027627734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,16384,0.031882666879230075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,16384,0.022689777943823073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,4096,0.01870133313867781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,12288,0.013770666387346057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,12288,0.026054221722814772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,12288,0.018079999420377944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,10240,0.028821332587136164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,10240,0.01239111108912362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,10240,0.022634666826989915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,10240,0.014948444234000312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,3072,0.017551110850440133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,8192,0.011053333679835001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,8192,0.019229332605997723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,8192,0.013085333009560903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,2560,0.016536888149049547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,7168,0.010339555641015371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,7168,0.016884444488419425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,7168,0.01204533295498954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,6144,0.021588444709777832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,6144,0.009418666362762451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,6144,0.014897776974572076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,6144,0.011129777464601727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,5120,0.019387554791238572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,5120,0.00872622181971868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,5120,0.013444444371594323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,5120,0.010415111151006486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,768,0.016160888804329764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,4096,0.007699555820888943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,4096,0.011720000041855706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,4096,0.009688888986905416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,512,0.013073777986897362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,3584,0.0069671107663048645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,3584,0.011039111349317761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,3584,0.00903200027015474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,3072,0.01683288812637329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,3072,0.006598222172922558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,3072,0.010339555641015371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,3072,0.008639111287064022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,65536,0.12165866957770453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,2560,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,2560,0.00941066692272822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,2560,0.008010666403505538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,512,128,0.01311288856797748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,2048,0.005564444594913059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,2048,0.00869066682126787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,12288,0.0339937773015764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,2048,0.007380444142553542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,1536,0.014144000079896716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,1536,0.004875555634498596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,1536,0.008285333712895712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,1536,0.006933333145247565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,1024,0.013573333621025085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,1024,0.0038897775941424896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,1024,0.007318221860461765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,1024,0.006480888773997624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,768,0.0039013334446483185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,768,0.006630222416586346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,768,0.00600622221827507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,8192,0.024784889486100938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,512,0.003506666670242945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,512,0.006640888750553131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,512,0.0058355554938316345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,256,0.012805333567990197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,256,0.003165333428316646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,256,0.006250666661394968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,256,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,7168,0.023729777998394434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,128,0.002867555452717675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,128,0.006579555571079254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,256,128,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,64,0.002504000026318762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,64,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,256,32,0.002862222285734283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,256,32,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,3584,0.017674666312005784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,65536,0.020573332905769348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,65536,0.10853511095046997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,4096,0.018233777748213876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,65536,0.06924444437026978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,16384,0.03985866573121812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,16384,0.011077333655622272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,16384,0.03186933199564616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,16384,0.022328888376553852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,12288,0.03278844555219015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,12288,0.009663110805882348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,12288,0.025739555557568867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,12288,0.017197334104114108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,10240,0.009671111073758867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,10240,0.02251911163330078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,10240,0.014819555812411837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,2560,0.015956444872750174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,8192,0.008399110701349046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,8192,0.018186666899257235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,8192,0.013119999733236102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,2048,0.014769777655601501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,7168,0.023026666707462732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,7168,0.007692444655630324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,7168,0.01665511065059238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,7168,0.012034667034943899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,6144,0.007589333587222629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,6144,0.014748444159825643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,6144,0.011348444554540845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,5120,0.01889155473974016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,5120,0.007941332956155142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,5120,0.013397333522637686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,5120,0.01035288886891471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,4096,0.017666666044129264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,4096,0.007613333563009898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,4096,0.011732444167137146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,4096,0.009696000152164036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,3584,0.016548444827397663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,3584,0.0069413334131240845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,3584,0.010679999987284342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,3584,0.009008000294367472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,768,0.015839111473825242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,3072,0.006568000134494569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,3072,0.010000000397364298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,3072,0.008641777767075432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,512,0.012103999654452005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,2560,0.005912888795137405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,2560,0.009309333231714036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,2560,0.008001777860853408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,256,128,0.012042666474978128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,2048,0.00526311124364535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,2048,0.008652444514963362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,2048,0.007274666594134436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,65536,0.12110755178663467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,1536,0.004594666676388847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,1536,0.008056888977686564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,1536,0.006659555352396435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,1024,0.01279022213485506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,1024,0.004015999949640698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,1024,0.007293333609898885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,1024,0.006278222219811545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,768,0.014929778046078153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,768,0.003576888806290097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,768,0.007272889216740926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,768,0.006051555689838197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,512,0.012440000143316058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,512,0.0032257777121331957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,512,0.006583999842405319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,512,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,256,0.012744888663291931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,256,0.003206222214632564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,256,0.006584888945023219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,8192,0.02472177810139126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,256,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,128,0.01181333346499337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,128,0.002809777855873108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,128,0.006584888945023219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,512,128,128,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,10240,0.028023110495673284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,64,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,64,0.006254222244024277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,128,32,0.002806222273243798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,128,32,0.006269333263238271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,65536,0.019926221834288705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,16384,0.011039999624093374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,65536,0.10836355553732978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,16384,0.03152889013290405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,12288,0.009351111120647853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,12288,0.025795555777019922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,10240,0.009669333696365356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,10240,0.022104000051816303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,8192,0.008332444561852349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,8192,0.01789777808719211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,7168,0.007632888853549957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,7168,0.016194666425387066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,6144,0.00737955586777793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,6144,0.014781332678265043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,5120,0.007649777664078607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,5120,0.013094222380055321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,4096,0.007299555672539606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,4096,0.011738667057620155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,3584,0.007286222444640265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,3072,0.006646222124497096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,3584,0.010718222293588849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,3072,0.009751111268997192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,2560,0.009353777600659264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,6144,0.02128444446457757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,2560,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,2048,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,2048,0.008656889200210571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,1536,0.004527111020353106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,1024,0.007284444239404466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,1536,0.008003555238246918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,1024,0.0038764443662431505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,768,0.0034906665484110513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,768,0.0069217777086628815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,512,0.003528000166018804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,512,0.006608888920810487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,256,0.002854222224818336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,256,0.006558222075303395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,64,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,128,0.0029084444459941653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,128,0.005998222364319696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,64,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,64,32,0.002520888836847411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,64,32,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,65536,0.02880711025661892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,3072,0.015791111522250705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,16384,0.010323555933104621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,65536,0.10633333524068196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,16384,0.03141599893569946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,12288,0.009008888569143083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,12288,0.025636444489161175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,8192,0.008400888906584846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,7168,0.00793422261873881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,10240,0.008112000094519721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,10240,0.021942221456103857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,8192,0.01792533364560869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,7168,0.016072000066439312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,6144,0.0074933336840735535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,6144,0.014774221513006421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,5120,0.007287999822033777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,3584,0.0069546666410234236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,5120,0.012785777449607849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,4096,0.007320889168315464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,4096,0.011399110986126794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,2560,0.0069431112044387394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,2560,0.015811555915408664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,3584,0.01072799993885888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,3072,0.0069351109365622205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,3072,0.010023111270533668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,2560,0.009350222018029954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,2048,0.006260444306664997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,2048,0.008960000342792934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,1536,0.004515555583768421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,1536,0.00777333312564426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,1024,0.0038568890757030914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,1024,0.0069475554757648045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,768,0.0035297779573334586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,768,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,128,0.002895111011134254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,512,0.003185777821474605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,512,0.006572444405820634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,256,0.002867555452717675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,256,0.006262222097979651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,128,0.006215999937719769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,64,0.002530666689078013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,64,0.006568888823191325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,512,32,32,0.002552888873550627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,512,32,32,0.006245333287451003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,12288,0.4703271124098036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,16384,0.621348434024387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,12288,0.30737421247694224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,2048,0.014469333820872836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,16384,0.4389173454708523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,512,128,1536,0.013799111048380533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,12288,0.3262062337663439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,10240,0.39459821912977433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,10240,0.2613280084398058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,8192,0.324107567469279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,10240,0.28612534205118817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,8192,0.2565022309621175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,8192,0.22049600548214385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,7168,0.22523199187384713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,7168,0.27283644676208496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,7168,0.2007733318540785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,6144,0.2461857795715332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,6144,0.16892888810899523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,16384,0.3925893306732178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,5120,0.15669243865542942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,5120,0.2046133279800415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,6144,0.17665333218044707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,5120,0.14770489268832737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,16384,0.227330658170912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,4096,0.16547021600935194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,12288,0.1719679964913262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,4096,0.12031288941701253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,3584,0.12552800443437365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,10240,0.14567644066280788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,3584,0.14514844947391087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,3584,0.10760977533128525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,3072,0.11459822124905056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,3072,0.12490755981869167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,3072,0.09348355399237739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,8192,0.11895555920071071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,2560,0.10510044627719456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,2560,0.1059911118613349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,2560,0.08078755272759332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,7168,0.10383111238479614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,2048,0.0951297746764289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,2048,0.08435022168689305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,2048,0.06649333238601685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,6144,0.09143910805384318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,1536,0.08522133032480876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,1536,0.06530400117238362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,1536,0.053371555275387235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,4096,0.13652533955044216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,5120,0.07788088586595324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,1024,0.04566311173968845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,1024,0.03972444600529141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,4096,0.06454310814539592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,768,0.040992889139387347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,768,0.035079999102486506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,768,0.036215109957589045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,3584,0.0571075545416938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,512,0.06475644641452365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,512,0.02576622201336755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,512,0.033069332440694175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,3072,0.05018844538264804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,2560,0.04471288786994087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,256,0.016784888174798753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,1024,0.07523822122149997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,256,0.029083556599087183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,128,0.02693333393997616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,128,0.017487110363112558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,128,0.027279110418425664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,2048,0.03807910945680406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,64,0.015264888604482016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,64,0.025034666061401367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,65536,32,0.015107555521859063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,65536,32,0.02442844377623664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,1536,0.030713776747385662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,65536,0.5950346522861057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,65536,0.4491644435458713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,65536,0.24794310993618437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,16384,0.13616711563534206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,16384,0.1556817822986179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,16384,0.1163475513458252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,65536,256,0.030459556314680312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,12288,0.10588000218073527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,12288,0.11943111154768203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,12288,0.08919822507434422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,1024,0.02495288848876953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,10240,0.09182755814658271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,10240,0.10228444470299615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,10240,0.07476266887452868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,65536,0.46372710333930117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,8192,0.08345955610275269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,8192,0.0800853305392795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,16384,0.06538310978147718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,8192,0.059752000702752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,7168,0.0728364454375373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,7168,0.07079022460513644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,7168,0.05344088872273763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,768,0.022280888424979314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,12288,0.05067733261320326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,6144,0.06266844272613525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,6144,0.04736444354057312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,10240,0.04303555687268575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,5120,0.05782933367623223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,5120,0.05208977725770739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,5120,0.040476444694730974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,512,0.020959999826219346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,256,0.01958311100800832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,65536,128,0.019351111518012155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,4096,0.04277333286073473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,4096,0.033371554480658636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,8192,0.0362737774848938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,3584,0.03725333346260919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,3584,0.029835555288526747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,7168,0.03234222200181749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,3072,0.04412622253100077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,3072,0.032144887579811945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,3072,0.026722666290071275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,6144,0.029073778125974867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,5120,0.025742222865422566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,2560,0.027429333991474573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,2560,0.023885332875781592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,4096,0.02253155575858222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,3584,0.020367999871571858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,2048,0.02217955556180742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,2048,0.021111999948819477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,3072,0.018962666392326355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,6144,0.06463555494944255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,1536,0.0173688895172543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,1536,0.01825066738658481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,2560,0.01755555636352963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,4096,0.05161422159936693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,1024,0.01367644468943278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,1024,0.01548266741964552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,2048,0.015558222929636637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,768,0.024712888730896845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,768,0.01179377817445331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,768,0.013807111316257052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,3584,0.047558221552107066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,512,0.027064000566800434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,512,0.010037333601050908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,512,0.011762667033407422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,1536,0.013400889105266996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,2560,0.041111111640930176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,256,0.007995555798212687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,256,0.011057777537239922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,1024,0.01182666669289271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,128,0.018908444378111098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,128,0.0069057775868309875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,128,0.010027555955780877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,768,0.010781333678298526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,64,0.006515555497672822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,64,0.010040889183680216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,16384,32,0.0063991112013657885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,16384,32,0.01034400032626258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,2048,0.03786844346258376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,65536,0.4778764512803819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,512,0.010386666489972008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,1536,0.034894221358829074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,16384,0.09867911206351386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,65536,0.33506043752034503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,16384,0.1190355618794759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,16384,0.09177422523498535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,12288,0.09433066844940186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,12288,0.0881955557399326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,12288,0.07349689139260186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,256,0.009401777552233802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,1024,0.03153955605294969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,10240,0.08146399921841092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,10240,0.057914667659335665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,16384,128,0.009175111022260454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,8192,0.06014221906661987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,8192,0.06587111287646823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,65536,0.19817866219414604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,8192,0.047163556019465126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,7168,0.05439822210205925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,7168,0.05775555637147692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,7168,0.0415635539425744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,16384,0.05516889029079013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,12288,0.041492445601357356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,6144,0.04964444372389051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,6144,0.03672622309790717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,10240,0.03467822074890137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,8192,0.030071109533309937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,5120,0.04217155443297493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,5120,0.031607998741997614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,16384,256,0.02128533356719547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,4096,0.040561778677834406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,4096,0.033174223370022245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,7168,0.026686223016844854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,4096,0.02625600000222524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,6144,0.02379911144574483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,3584,0.029659556017981634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,3584,0.024271999796231587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,5120,0.02141688929663764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,3072,0.035720000664393105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,3072,0.025647999511824712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,65536,0.33241245481703013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,3072,0.022499556342760723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,4096,0.019336889187494915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,2560,0.02169955604606205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,2560,0.020263999700546265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,3584,0.017671111557218764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,2048,0.03113599949412876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,2048,0.01771022213829888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,2048,0.017854221993022494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,10240,0.06866577598783705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,3072,0.01626044511795044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,1536,0.01384622189733717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,1536,0.015568888849682279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,6144,0.04991555545065138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,2560,0.015028445257080926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,1024,0.026605332891146343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,1024,0.011397333608733283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,1024,0.012803555362754397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,768,0.022513777017593384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,768,0.009349333743254343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,768,0.011367110742463006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,2048,0.013598221871587964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,512,0.022338666849666174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,512,0.00777599960565567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,512,0.010048000348938836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,5120,0.04518666532304552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,1536,0.011743110915025076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,256,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,256,0.009694221946928237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,1024,0.010662222074137794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,3584,0.03823555509249369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,128,0.004883555488453971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,128,0.008754666480753157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,768,0.009728000395827824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,64,0.004554666578769684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,64,0.008723555339707268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,12288,32,0.004858666823969947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,12288,32,0.009015999734401703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,65536,0.2743031183878581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,512,0.00868800034125646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,65536,0.3923964500427246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,65536,0.3117146756913927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,256,0.00810222244924969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,16384,0.10454133484098647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,16384,0.08074400160047743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,12288,128,0.007859555383523306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,2560,0.033470223347345986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,12288,0.06901777452892728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,12288,0.07783555322223239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,12288,0.061589333746168345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,10240,0.06047288576761881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,10240,0.06481955448786418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,10240,0.05225155419773526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,65536,0.17634844779968262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,8192,0.05318311187956068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,8192,0.052080889542897545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,8192,0.04273066586918301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,16384,0.04894489049911499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,7168,0.047938668065600924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,7168,0.04683466752370199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,12288,0.03894933395915561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,7168,0.038600001070234514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,6144,0.044234666559431285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,6144,0.04058488872316148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,1536,0.02871111035346985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,6144,0.03417866759830051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,5120,0.04016088777118259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,5120,0.034252444903055825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,10240,0.03325422273741828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,5120,0.030068443881140813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,4096,0.036834667126337685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,4096,0.028020444843504164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,4096,0.025555556019147236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,8192,0.028639998700883653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,256,0.01896711190541585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,3584,0.024451555477248296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,3584,0.023766222927305434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,7168,0.02569866677125295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,6144,0.02307555576165517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,3072,0.021199110481474135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,3072,0.021423111359278362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,12288,128,0.015806222955385845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,5120,0.02051822178893619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,2560,0.030573334958818223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,2560,0.01811644434928894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,2560,0.01924977699915568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,4096,0.018239999810854595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,2048,0.015507555670208402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,2048,0.017376888129446242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,3584,0.016908443636364408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,3072,0.015480000111791821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,1536,0.013122667041089801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,1536,0.015204444527626038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,2560,0.014144000079896716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,1024,0.02437066700723436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,1024,0.010703999963071613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,16384,0.0844328867064582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,1024,0.012096000214417776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,2048,0.01276711126168569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,768,0.008952000074916417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,768,0.011065777805116443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,1536,0.011362666885058085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,1024,0.009679999616410997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,512,0.007094221810499827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,512,0.010055999788973067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,768,0.008637333081828224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,3584,0.03472444415092468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,256,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,256,0.009356444080670675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,512,0.00797155582242542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,3072,0.03250755535231696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,128,0.005250666704442766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,128,0.008633777499198914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,2048,0.028522666957643297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,64,0.004504000147183736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,64,0.008678221868144141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,10240,32,0.005668444352017508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,10240,32,0.008657777474986183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,256,0.007680888805124495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,1536,0.026763555076387193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,65536,0.32159466213650173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,65536,0.2389777766333686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,10240,128,0.007610666255156199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,16384,0.08043466673956977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,16384,0.06093066930770874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,768,0.02165866725974613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,65536,0.13385155465867785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,12288,0.061887999375661217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,12288,0.04766755633884006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,16384,0.03797422183884515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,512,0.02033422225051456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,256,0.01789066692193349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,10240,0.051388445827696055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,12288,0.03028800090154012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,10240,0.04058311051792569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,8192,0.04227911101447212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,8192,0.03324266605907016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,10240,0.026358221968015034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,7168,0.04520888792143928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,7168,0.03650133477316962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,8192,0.0225564440091451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,7168,0.029643555482228596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,10240,128,0.015855110353893705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,6144,0.031762666172451444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,65536,0.25725865364074707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,6144,0.026591110560629103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,7168,0.020459555917316012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,5120,0.02720888952414195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,5120,0.02409155501259698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,16384,0.07965155442555745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,6144,0.018965333700180054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,4096,0.021881777379247878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,4096,0.020918221937285528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,5120,0.016951110627916124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,3584,0.03188000122706095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,3584,0.01962933275434706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,3584,0.01908888916174571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,12288,0.06490400102403429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,4096,0.015481778317027621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,10240,0.0562373333507114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,3072,0.017515555024147034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,3072,0.017253332667880587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,2560,0.028539554940329656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,2560,0.014871999621391296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,2560,0.015511110424995422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,3584,0.014087110757827759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,8192,0.04901422063509623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,2048,0.013226666384273104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,2048,0.014132445057233175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,3072,0.013410666750537025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,2560,0.01236177815331353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,1536,0.011117333339320289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,1536,0.012432888978057437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,2048,0.011155555645624796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,1024,0.021664000219768945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,1024,0.009014222357008193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,1024,0.010028444230556488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,6144,0.04147111044989692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,768,0.01925511161486308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,768,0.007359111474619971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,768,0.00902133352226681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,1536,0.009645333720578088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,5120,0.03776266508632236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,512,0.006085333310895496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,512,0.008377778033415476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,1024,0.007987555530336168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,256,0.016882666283183627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,256,0.005254222287072076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,4096,0.034225778447257146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,256,0.00832177781396442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,768,0.007294221884674496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,128,0.015445333388116626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,128,0.004943111290534337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,128,0.007990222010347579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,64,0.004553777890072929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,64,0.007627555893527136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,8192,32,0.004913777940803104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,8192,32,0.007691555553012424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,512,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,3072,0.030172444052166406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,256,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,65536,0.32394933700561523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,8192,128,0.0069724445541699724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,65536,0.220414227909512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,16384,0.08333866463767158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,16384,0.05726399686601427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,2048,0.026720888084835474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,12288,0.0645608901977539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,65536,0.12755378087361655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,12288,0.04433600107828776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,16384,0.037244445747799344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,10240,0.04719822274314033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,10240,0.053447110785378345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,10240,0.03781955440839132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,1536,0.024762666887707178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,8192,0.04195644458134969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,8192,0.044178667995664805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,8192,0.032115555471844144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,12288,0.02956533432006836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,7168,0.03830400109291077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,7168,0.03898577888806661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,7168,0.02903377678659227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,10240,0.025855110751258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,6144,0.033597333563698664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,6144,0.026039999392297532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,8192,0.022727999422285292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,8192,512,0.018408889571825664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,5120,0.028146667612923518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,5120,0.023366222778956097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,7168,0.020260444945759244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,6144,0.018447111050287884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,4096,0.023131555981106226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,4096,0.02012711101108127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,5120,0.016765332884258695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,3584,0.028580443726645574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,3584,0.020471110939979553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,3584,0.018343110879262287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,4096,0.015019555886586508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,65536,0.207458668284946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,3072,0.01853155593077342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,3072,0.017216889394654166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,3584,0.014070222775141398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,12288,0.05385155479113261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,2560,0.015623110863897534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,16384,0.06559644142786662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,2560,0.01572711103492313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,2048,0.024696888195143804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,2048,0.01366133325629764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,2048,0.014039110806253223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,3072,0.012850667039553324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,1536,0.022314666046036616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,1536,0.011713777979214987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,1536,0.012034667034943899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,2560,0.012418666647540199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,2048,0.010678222609890832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,1024,0.009156444834338294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,1024,0.009825777676370408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,1536,0.009008000294367472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,768,0.018233777748213876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,768,0.008007110820876228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,768,0.008980444735950893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,6144,0.035747554567125105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,512,0.016827555166350473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,512,0.006815111057625876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,512,0.008323555191357931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,1024,0.007931555310885111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,768,0.007348444726732042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,256,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,5120,0.032953778902689614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,256,0.008001777860853408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,4096,0.030511998467975195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,128,0.004853333450025982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,128,0.007663110891977946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,512,0.007181333170996771
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,64,0.004122666600677702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,64,0.007647111184067196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,7168,32,0.0041617775956789655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,7168,32,0.007652444144090016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,256,0.00692622239391009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,65536,0.27134932412041557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,7168,128,0.006871999965773688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,65536,0.20332622528076172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,16384,0.0639537771542867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,16384,0.06593066453933716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,16384,0.05436444282531738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,3072,0.027085334062576294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,65536,0.12427288956112331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,12288,0.05025777882999844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,12288,0.0426879988776313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,2560,0.025769778423839148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,16384,0.03678044345643785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,10240,0.04216088851292928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,10240,0.0374017788304223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,12288,0.029232889413833618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,8192,0.0395671096112993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,8192,0.033913777934180364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,8192,0.03169599837727017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,10240,0.025506666964954797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,7168,0.03641600079006619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,7168,0.02954044606950548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,7168,0.028811554114023846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,1024,0.01903733279969957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,8192,0.022282666630215112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,6144,0.02557955516709222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,6144,0.025948445002237957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,7168,0.019928000039524503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,5120,0.031082666582531396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,5120,0.021486222743988037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,5120,0.023346667488416035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,256,0.017214222086800467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,6144,0.018047110901938546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,7168,128,0.015156444576051501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,4096,0.01759111053413815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,4096,0.019894222418467205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,5120,0.016249777542220224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,3584,0.02682311170630985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,3584,0.015228443675571017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,3584,0.018509333332379658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,65536,0.19204266866048178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,3072,0.013617777989970313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,3072,0.016751999656359356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,4096,0.014688889185587565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,2560,0.024475556280877855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,2560,0.01202044470442666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,2560,0.015375999940766228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,3584,0.013668444421556262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,2048,0.022737777895397607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,2048,0.010374222364690568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,2048,0.013611555927329592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,3072,0.012748444245921241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,2560,0.011734222372372946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,1536,0.008947555389669206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,1536,0.011596444580290051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,12288,0.05043377810054355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,10240,0.04440444376733568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,1024,0.007585778004593319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,1024,0.009463999834325578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,2048,0.010105777945783403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,768,0.017824888229370117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,768,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,768,0.008954666554927826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,1536,0.008578666382365758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,1024,0.00792711145348019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,512,0.004920888692140579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,512,0.008163555628723567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,768,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,512,0.006959999601046245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,256,0.004219555606444676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,6144,0.03397422366672092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,256,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,256,0.006976889239417181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,128,0.003944888710975647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,4096,0.028909332222408716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,128,0.007673777639865875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,64,0.003633777714437909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,64,0.0073831114504072405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,6144,32,0.0036071112586392295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,6144,32,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,3072,0.025576000412305195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,65536,0.23886312378777397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,6144,128,0.006791111081838608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,65536,0.19410933388604057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,65536,0.12139911121792263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,16384,0.06137777699364556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,1536,0.019879110985332064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,16384,0.05338577760590447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,1024,0.01754577789041731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,16384,0.03512622250450982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,12288,0.04639377858903673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,12288,0.04174222217665779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,10240,0.043502221504847206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,10240,0.039536890056398176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,10240,0.03639288743336996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,512,0.016419554750124615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,12288,0.02831288841035631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,8192,0.03169599837727017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,8192,0.03102311160829332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,10240,0.02440533373090956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,256,0.01686844395266639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,7168,0.027951111396153767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,7168,0.027797334724002417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,8192,0.02142400046189626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,6144,0.033069332440694175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,6144,0.024139554964171514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,6144,128,0.015441776977645027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,6144,0.02532800038655599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,7168,0.019183110859658983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,5120,0.02049066623051961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,5120,0.022760889596409265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,6144,0.017536888519922893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,4096,0.027874665127860174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,4096,0.017328888177871704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,4096,0.019311111834314134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,5120,0.015909334023793537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,65536,0.18781067265404594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,4096,0.01447111037042406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,3584,0.01574133336544037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,3584,0.0181688881582684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,16384,0.059692442417144775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,3072,0.014767999450365702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,3072,0.016583111551072862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,3584,0.013332444760534497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,3072,0.012313777373896705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,2560,0.012758221891191272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,12288,0.04905866583188375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,2560,0.014988443917698331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,2048,0.02166844407717387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,2048,0.011184000306659274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,2048,0.013160000244776407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,2560,0.010991999672518836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,2048,0.009568000005351173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,1536,0.009367110828558603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,1536,0.011386666860845355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,7168,0.03537599907981025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,8192,0.038520889149771795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,1024,0.01683200067943997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,1024,0.009316444396972656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,1024,0.0075457774930530125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,1536,0.008340444829728868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,768,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,768,0.008633777499198914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,1024,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,5120,0.03064889046880934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,512,0.0052675555149714155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,512,0.007999999655617608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,768,0.007251555720965068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,256,0.01648533344268799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,256,0.004396444393528832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,256,0.007976888782448238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,3584,0.025853332546022203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,128,0.01516177753607432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,512,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,128,0.0041831110914548235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,128,0.007653333246707916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,64,0.0038968887594011095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,64,0.007663110891977946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,5120,32,0.003580444388919406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,5120,32,0.007663110891977946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,256,0.006611555400821898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,65536,0.1571848922305637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,65536,0.18191022343105742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,3072,0.02473066747188568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,65536,0.12722488244374594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,16384,0.05100711186726888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,16384,0.04632355438338386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,16384,0.03424533208211263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,12288,0.042493333419164024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,5120,128,0.006628444625271692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,12288,0.03552800085809495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,12288,0.02755733331044515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,65536,0.07695200045903523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,10240,0.030016889174779255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,16384,0.024471999870406255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,10240,0.024440889557202656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,8192,0.03405155407057868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,8192,0.024831111232439678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,8192,0.020949333906173706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,12288,0.02034222251839108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,7168,0.03137955400678847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,7168,0.021370665894614324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,7168,0.019304000669055514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,2560,0.02363555630048116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,6144,0.029253333806991577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,6144,0.018768888380792405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,6144,0.01811822255452474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,10240,0.017772444420390658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,5120,0.02700000007947286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,5120,0.016182222300105624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,5120,0.01644000079896715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,8192,0.0158933334880405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,4096,0.02500533395343357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,4096,0.013777777552604675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,4096,0.014140443669425117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,7168,0.014661333627170987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,1536,0.018731556004948087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,3584,0.012636444634861417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,6144,0.013665777941544851
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,3584,0.013556444810496436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,3072,0.022437334060668945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,3072,0.011544889046086205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,3072,0.01257777793539895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,5120,0.012406222522258759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,4096,0.01127022256453832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,2560,0.00998222248421775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,2560,0.011367110742463006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,3584,0.010479110810491772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,3072,0.009282666775915358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,2048,0.009062222308582729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,2048,0.009672889278994666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,2560,0.00866844422287411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,1536,0.016855110724767048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,1536,0.007489778101444244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,1536,0.008718222379684448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,768,0.01721244388156467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,2048,0.00792266676823298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,1024,0.006028444402747684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,1024,0.015876443849669564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,1024,0.007627555893527136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,1536,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,5120,512,0.015492444237073263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,768,0.0052533331844541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,768,0.0073315559162033936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,512,0.014484445254007975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,512,0.004518222063779831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,512,0.0069431112044387394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,1024,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,768,0.006481777876615524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,256,0.0042133331298828125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,256,0.006561777657932705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,512,0.006172444257471297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,256,0.005938666562239329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,128,0.0038844446341196695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,128,0.006881777611043718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,10240,0.03794044587347243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,64,0.0035546666218174826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,64,0.006587555425034628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,4096,32,0.0038613333470291565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,4096,32,0.006635555376609166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,65536,0.15522400538126627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,65536,0.16211822297838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,4096,128,0.005990222096443176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,65536,0.11846755610571967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,16384,0.05195822318394979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,16384,0.04249688982963562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,16384,0.033534222178988986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,65536,0.07526666588253446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,12288,0.0421706669860416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,3584,0.02369155486424764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,12288,0.0329075555006663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,12288,0.027108444107903376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,16384,0.024246222443050806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,10240,0.02755644420782725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,10240,0.023853333459960088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,12288,0.01996799972322252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,2560,0.021013332737816706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,8192,0.02294933299223582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,10240,0.01756266587310367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,8192,0.020786666207843356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,2048,0.01893422173129188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,7168,0.02033511135313246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,7168,0.018989332848125033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,8192,0.015592889653311836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,6144,0.02903999884923299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,6144,0.01790577835506863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,6144,0.017551110850440133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,768,0.01684177749686771
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,7168,0.01440355512830946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,5120,0.015611555841233997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,5120,0.016020443704393175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,6144,0.013242666920026144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,5120,0.01241422196229299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,4096,0.013431111143694984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,4096,0.013779555757840475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,256,0.01516711049609714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,3584,0.022966222630606756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,3584,0.012426666915416718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,3584,0.01312444441848331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,4096,0.01072266697883606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,3072,0.021241777473025854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,3072,0.011173333558771344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,3072,0.012082666986518435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,3584,0.009684444301658207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,4096,128,0.014757333530320061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,2560,0.009979555176364051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,2560,0.010680889089902243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,3072,0.008998221821255153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,2560,0.00814400033818351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,2048,0.008601778083377415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,2048,0.009743111001120673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,2048,0.0075777777367168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,10240,0.03729066583845351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,1536,0.0070675553547011475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,1536,0.008353778057628209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,8192,0.03341066506173875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,1024,0.015846222639083862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,1024,0.005527110977305307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,1024,0.007734222544564142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,1536,0.007053333024183909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,1024,0.006609777609507243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,768,0.00490044429898262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,7168,0.031207111146714952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,768,0.007292444507280986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,512,0.014463111758232117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,512,0.004538666870858935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,512,0.0069439998931354946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,5120,0.026760889424218073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,768,0.006308444258239534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,256,0.003896000070704354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,256,0.006607999818192587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,512,0.006213333457708359
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,4096,0.024490666058328416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,128,0.0038533334930737815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,128,0.006554666492674086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,256,0.005966222120655908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,64,0.00350577798154619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,64,0.006292444550328785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3584,32,0.0035377778112888336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3584,32,0.006616000086069107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3584,128,0.005881777654091517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,65536,0.1472782161500719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,65536,0.1101493305630154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,65536,0.07404088973999023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,16384,0.04895377821392483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,2560,0.0199955552816391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,16384,0.036334223217434354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,16384,0.032235556178622775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,2048,0.01805688937505086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,12288,0.0276826669772466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,12288,0.026912000444200303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,16384,0.023704888092146978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,12288,0.01960177719593048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,1536,0.01664355562792884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,10240,0.02311022248533037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,10240,0.023758222659428913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,10240,0.017330666383107502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,8192,0.01835555500454373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,8192,0.020564445190959506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,8192,0.015437333120240105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,768,0.016935999194780987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,7168,0.016100444727473788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,7168,0.018529777725537617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,7168,0.014133334159851074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,256,0.014807111687130399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,6144,0.014460444450378418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,6144,0.017305778132544625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3584,128,0.015115555789735583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,5120,0.025855110751258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,5120,0.01239911135700014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,5120,0.015835555063353646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,6144,0.013154666456911298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,5120,0.011711111499203576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,4096,0.010795555180973478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,4096,0.013820444544156393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,65536,0.14986311064826116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,4096,0.01034311122364468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,3584,0.022008889251285132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,3584,0.009815110928482479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,3584,0.012838222086429596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,3584,0.009538667069541084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,3072,0.008996444443861643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,3072,0.011630222201347351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,3072,0.008951110972298516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,12288,0.04049955474005805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,2560,0.019201777047581144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,2560,0.008599110775523717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,2560,0.010388444695207808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,2560,0.00797333319981893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,2048,0.007530666887760162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,2048,0.009327111144860586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,10240,0.03611822260750665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,1536,0.016176000237464905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,1536,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,1536,0.008272888759771982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,8192,0.03194311261177063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,1024,0.015154666370815702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,1024,0.00481599983241823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,1024,0.007606222397751278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,2048,0.007298666569921706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,768,0.01684711045689053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,768,0.004230222354332606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,768,0.007131555842028723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,1536,0.0069351109365622205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,1024,0.006291555447710886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,512,0.004209777961174647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,512,0.006767999794748094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,768,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,512,0.005959110955397288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,256,0.003504000190231535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,256,0.006574222197135289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,7168,0.029606223106384277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,6144,0.02803022331661648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,256,0.005903999838564131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,128,0.0034933334423436057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,128,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,64,0.003156444471743372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,64,0.006260444306664997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,3072,32,0.003495111233658261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,3072,32,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,65536,0.14774755636850992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,65536,0.13192089398701987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,3072,128,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,65536,0.10493599706225926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,16384,0.04765333400832283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,16384,0.03671911027696397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,4096,0.023689778314696416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,16384,0.031884445084465876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,65536,0.07285600238376193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,12288,0.028031110763549805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,12288,0.026425777210129633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,16384,0.023382221659024555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,10240,0.03535022338231405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,10240,0.02369244396686554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,10240,0.023380445109473333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,12288,0.019243554936514962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,8192,0.03194133440653483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,8192,0.019556444552209642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,8192,0.01998666591114468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,10240,0.017196445001496207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,7168,0.029378665818108454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,7168,0.017867555220921833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,7168,0.01846844454606374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,8192,0.01516266663869222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,6144,0.027792887555228338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,6144,0.015612444943851896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,3072,0.0203484445810318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,6144,0.01686755485004849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,5120,0.02564711040920681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,5120,0.013643555343151093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,7168,0.013752000199423896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,5120,0.015488000379668342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,6144,0.01260088880856832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,4096,0.021996445126003687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,4096,0.010625777973069085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,4096,0.013444444371594323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,5120,0.011140444212489657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,2048,0.01718755563100179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,3584,0.009497777455382878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,3584,0.012308444413873883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,4096,0.009703999592198266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,3072,0.019261333677503798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,3072,0.008781332936551835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,3072,0.011121778024567498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,2560,0.017861333158281114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,2560,0.007976888782448238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,3584,0.009343110852771336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,2560,0.00998755544424057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,3072,0.00832622249921163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,2048,0.007296000089910295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,2048,0.009000000026490953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,2560,0.007962666451931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,1536,0.015828443898095023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,1536,0.0058666666348775225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,1536,0.008329778081840938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,2048,0.007489778101444244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,1024,0.014895111322402954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,1024,0.004924444274769889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,1024,0.007671111159854465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,1536,0.007050666544172499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,512,0.014162666267818876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,768,0.004543111142185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,768,0.007270221908887227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,1024,0.006321777900060018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,256,0.014811555544535318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,768,0.006131555471155379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,512,0.004407111141416761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,512,0.0068862222962909276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,256,0.015219555960761176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,256,0.0038986665507157645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,256,0.006248888870080312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,512,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,256,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,128,0.00387999994887246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,128,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,3072,128,0.014818666709793938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,64,0.0035377778112888336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,64,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2560,32,0.0035422220826148987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2560,32,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2560,128,0.00573244442542394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,65536,0.0925111108356052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,65536,0.10319822364383274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,12288,0.03963911202218797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,16384,0.04718666606479221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,65536,0.07139466868506537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,16384,0.029113776153988306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,16384,0.03153955605294969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,16384,0.02272622287273407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,12288,0.02222577730814616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,12288,0.025789333714379206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,12288,0.018775999546051025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,3584,0.02121155626244015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,10240,0.018575110369258456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,10240,0.0229093333085378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,10240,0.01668355531162686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,8192,0.031211555004119873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,8192,0.01567555632856157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,8192,0.019846222466892667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,2048,0.016492444607946608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,7168,0.028667555914984807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,7168,0.014799111419253878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,7168,0.018217777212460835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,8192,0.014780445231331719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,7168,0.013412444127930535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,6144,0.01349511080318027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,6144,0.016857778032620747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,6144,0.011747555600272285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,768,0.016508445143699646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,5120,0.011617778076065911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,5120,0.015114666687117683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,5120,0.010488000180986194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,4096,0.021298666795094807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,4096,0.010025777750545079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,4096,0.012746666868527731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,512,0.013807111316257052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,3584,0.019263111882739596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,3584,0.009577777650621202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,3584,0.011757333245542316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,4096,0.009687110781669617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,3072,0.01851377718978458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,3072,0.008995555341243744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,3072,0.010574222438865237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,3584,0.008811555802822113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,2560,0.017479111750920612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,2560,0.0075937774446275495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,2560,0.009716444545321995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2560,128,0.014498665928840637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,3072,0.00832622249921163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,2048,0.006616000086069107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,2048,0.01648000048266517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,2048,0.008951110972298516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,2560,0.007728888756699032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,1536,0.015513777732849121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,1536,0.00592533333433999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,1536,0.008044444852405125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,2048,0.007304000357786815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,1024,0.014746665954589844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,1024,0.004912000149488449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,1024,0.007297777467303806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,65536,0.1475128862592909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,1536,0.006713777780532837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,768,0.00453599997692638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,768,0.007304000357786815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,1024,0.006291555447710886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,12288,0.039176887936062284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,768,0.005947555518812603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,512,0.0041946665280395085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,512,0.006612444503439798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,256,0.003551111039188173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,256,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,512,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,128,0.01479022204875946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,256,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,128,0.0031742221779293488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,128,0.006296000132958095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,64,0.0032239999208185407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,64,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,2048,32,0.0034977777136696708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,2048,32,0.006272000157170826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,10240,0.03474311033884684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,65536,0.08160710997051664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,65536,0.12489599651760525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,2048,128,0.005568000177542369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,65536,0.10258844163682725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,16384,0.024893333514531452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,16384,0.031089779403474595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,65536,0.07005688879224989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,16384,0.022384888596004907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,12288,0.01943911115328471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,12288,0.025753777888086107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,12288,0.01846222248342302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,6144,0.02682222260369195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,10240,0.017701332767804463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,10240,0.022628444764349196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,10240,0.01622222198380364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,8192,0.028014222780863445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,8192,0.015457777513398064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,8192,0.019371555911170114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,5120,0.024367999699380662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,8192,0.014092443717850579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,7168,0.013735999663670858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,7168,0.01795377830664317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,7168,0.012496888637542725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,6144,0.024657777614063684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,6144,0.012320000264379712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,6144,0.016698666744761996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,6144,0.011253332926167382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,5120,0.021532444490326777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,5120,0.010885333021481832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,5120,0.014771555860837301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,5120,0.010119111173682744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,512,0.013447999954223633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,4096,0.0099653336736891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,4096,0.012084444363911947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,768,0.016358221570650738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,4096,0.009513777991135916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,3584,0.008946666287051307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,3584,0.01129333343770769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,2048,256,0.014820444915029736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,3584,0.008786666724416945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,3072,0.008265777594513362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,3072,0.010312889185216691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,3072,0.008344000412358178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,2560,0.017403556240929496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,2560,0.0068897778789202375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,2560,0.009720889230569204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,16384,0.04162844353251987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,2048,0.015933333171738517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,2048,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,2048,0.008856000171767341
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,2560,0.007662222617202335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,1536,0.01515111161602868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,1536,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,1536,0.00796622203456031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,2048,0.007258666886223688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,12288,0.035319109757741295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,1024,0.004560889055331548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,1024,0.007343110938866933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,1536,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,10240,0.03122488988770379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,768,0.003955555458863576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,1024,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,768,0.0069679998689227635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,768,0.0059368887709246735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,512,0.0038986665507157645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,512,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,512,0.005622222191757626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,256,0.015072888798183866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,256,0.003567999849716822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,256,0.006396444307433234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,7168,0.026329777306980554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,256,0.005657777604129579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,128,0.0032275555034478507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,128,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1536,128,0.005628444254398346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,64,0.0032097777972618737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,64,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1536,32,0.0031635556370019913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1536,32,0.00627377794848548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,4096,0.019223110543357003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,65536,0.05814933114581638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,65536,0.10223199923833211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,65536,0.06851110855738322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,3584,0.018283555905024212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,16384,0.019862223002645705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,16384,0.030619555049472388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,16384,0.021948445174429152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,3072,0.01753777762254079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,12288,0.01684799955950843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,12288,0.02533333334657881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,12288,0.01804266704453362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,1024,0.014428445034556918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,10240,0.01551644504070282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,768,0.015831111205948722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,10240,0.022337777747048274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,8192,0.026572444372706946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,8192,0.013776889277829064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,8192,0.01921599937809838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,10240,0.015260444747077094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,512,0.013069333301650153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,8192,0.013396444420019785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,7168,0.012952889005343119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,7168,0.017514665921529133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,6144,0.022335999541812476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,6144,0.011392888923486074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,6144,0.015736889508035447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,7168,0.011426666544543372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,6144,0.01074044406414032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,5120,0.010029333333174387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,5120,0.013797333670987023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,5120,0.010027555955780877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,4096,0.0188755558596717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,4096,0.008618666893906063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,4096,0.011759111450778114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,4096,0.009327111144860586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,3584,0.01814311080508762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,3584,0.00794933322403166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,3584,0.011033777561452655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,3584,0.00867022242810991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1536,128,0.014140443669425117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,3072,0.007072889142566257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,3072,0.010312889185216691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,3072,0.008269333177142674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,65536,0.12190488974253337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,2560,0.006633777585294511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,2560,0.009386666946940953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,2560,0.007660444411966536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,16384,0.04083733426200019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,2048,0.005903999838564131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,2048,0.008627555436558193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,2048,0.007263999846246507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,12288,0.034508443540996976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,1536,0.005380444642570283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,1536,0.008007999923494127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,1536,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,10240,0.030787554052140977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,1024,0.004395555704832077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,1024,0.013801777528391944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,1024,0.007311111523045435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,7168,0.02438844409253862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,1024,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,768,0.00397511116332478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,768,0.0069511110583941145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,512,0.012764444781674279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,512,0.003912000192536248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,512,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,768,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,512,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,256,0.003539555602603488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,256,0.006640000061856375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,5120,0.02051822178893619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,256,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,128,0.0032293332947625052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,128,0.006305777778228124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,1024,128,0.005621333503060871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,64,0.003240888938307762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,64,0.00630044440428416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,1024,32,0.0031911111954185697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,1024,32,0.006296000132958095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,65536,0.11728355619642471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,65536,0.044884443283081055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,65536,0.10270222028096516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,65536,0.06616710954242282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,16384,0.039352890517976545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,16384,0.016103999482260812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,3072,0.016878222425778706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,16384,0.030573334958818223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,12288,0.033442666133244835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,12288,0.014472888575659858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,16384,0.021708443760871887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,12288,0.025309332542949255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,2560,0.016180444094869826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,12288,0.017571555243598092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,10240,0.015115555789735583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,10240,0.022314666046036616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,10240,0.015025777949227227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,8192,0.013797333670987023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,8192,0.01869244376818339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,8192,0.013233777549531726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,7168,0.023440000083711412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,7168,0.012422222230169507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,7168,0.017111111018392775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,7168,0.011374221907721626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,1536,0.014560000763999091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,2048,0.015488889482286243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,6144,0.02168088820245531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,6144,0.011168888873524137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,6144,0.015133332875039844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,5120,0.01000533335738712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,5120,0.013378666506873237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,6144,0.010679110884666443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,5120,0.010020444790522257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,4096,0.008351999852392409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,4096,0.011756444142924415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,4096,0.009347555538018545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,768,0.0151582227812873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,256,0.015096889601813423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,3584,0.007868444754017724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,3584,0.01090222183201048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,3584,0.00869422240389718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,3072,0.006969778074158563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,3072,0.010062221851613786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,1024,128,0.01418133411142561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,3072,0.008231999973456064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,2560,0.006295111030340195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,2560,0.009353777600659264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,2560,0.007672888537247975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,10240,0.029354665014478896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,2048,0.00555288874440723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,2048,0.00866844422287411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,2048,0.007259555160999298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,1536,0.014474666780895658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,1536,0.00490399988161193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,1536,0.007993777592976889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,1536,0.006664888726340399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,1024,0.013737777868906656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,1024,0.0041751112374994485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,1024,0.007240888973077138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,1024,0.006246222390068903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,8192,0.02520888878239526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,768,0.0038951109680864546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,768,0.0069679998689227635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,768,0.005948444621430502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,5120,0.019911110401153564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,512,0.003559111307064692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,512,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,512,0.005920889063013925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,4096,0.01888444523016612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,256,0.014934221903483072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,256,0.0032035555276605818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,256,0.006243555496136348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,128,0.014129777749379476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,128,0.002855111120475663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,128,0.006592000110281839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,256,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,64,0.0028320000403457214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,64,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,768,32,0.0028311111446883944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,768,128,0.005561777700980504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,768,32,0.0063182223174307085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,65536,0.03425511055522495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,65536,0.11386222309536403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,65536,0.1023520032564799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,3584,0.01791466606987847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,16384,0.03898755709330241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,65536,0.06732266479068331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,16384,0.015488000379668342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,16384,0.030326220724317763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,12288,0.01365244471364551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,12288,0.024958221448792353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,16384,0.02163733376397027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,12288,0.016819554898473952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,10240,0.012466666599114736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,10240,0.021929777330822412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,10240,0.014087999860445658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,8192,0.02407466702991062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,8192,0.011330666641394297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,8192,0.017887110511461895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,8192,0.012395555774370829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,3072,0.01754755609565311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,7168,0.010764444039927589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,7168,0.016348444753222995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,7168,0.011391110718250275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,2560,0.016149333781666227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,6144,0.010040889183680216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,6144,0.014426666829321118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,6144,0.01069866700304879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,2048,0.015506666567590503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,5120,0.009009777671760982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,5120,0.013100444442696042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,5120,0.009966221948464712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,768,0.015160888433456421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,4096,0.00793511089351442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,4096,0.011516444385051727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,4096,0.009307555854320526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,3584,0.016846223009957206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,3584,0.007294221884674496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,3584,0.010672888822025724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,3584,0.008662222160233391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,3072,0.016215110818545025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,3072,0.006559999866618051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,3072,0.010212444596820408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,768,512,0.01278133359220293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,3072,0.008151999778217739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,2560,0.005903111149867375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,2560,0.009296000003814697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,2560,0.0075866662793689305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,10240,0.027849776877297297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,2048,0.005579555614127054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,2048,0.00833155545923445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,2048,0.007248000138335758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,12288,0.03256266646915012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,1536,0.004649777793222004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,1536,0.008045333127180735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,1536,0.006633777585294511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,1024,0.013436444103717804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,1024,0.0042408886882993905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,1024,0.007057777709431118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,1024,0.00627377794848548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,768,0.0155102229780621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,7168,0.02241244415442149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,768,0.0038862224254343244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,768,0.0069439998931354946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,512,0.012410666379663678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,512,0.003522666792074839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,768,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,512,0.006640000061856375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,256,0.014779556128713818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,256,0.003152888889114062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,512,0.005670222143332164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,256,0.006592000110281839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,128,0.013086222112178802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,128,0.00286311118139161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,256,0.005623111294375525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,128,0.006588444527652528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,6144,0.021567111214001972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,64,0.0028133332315418455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,64,0.006280888699822956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,512,32,0.0028693332440323303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,512,32,0.006576889091067844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,512,128,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,65536,0.024076445235146418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,65536,0.11225422223409016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,65536,0.10156799687279595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,16384,0.01310755560795466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,65536,0.06605599986182319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,16384,0.03017510970433553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,12288,0.03007022208637661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,16384,0.020807999703619216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,12288,0.011751111182901593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,12288,0.02499288817246755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,12288,0.015791111522250705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,10240,0.010371555884679159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,10240,0.021336000826623704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,10240,0.014003554979960123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,5120,0.01923199991385142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,8192,0.02355466617478265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,8192,0.009363555245929295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,8192,0.017218665944205392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,7168,0.02193333374129401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,8192,0.012438221938080259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,7168,0.008461332983440822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,7168,0.015786666009161208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,7168,0.011425777441925473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,6144,0.007649777664078607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,6144,0.014495111174053617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,6144,0.01073066641887029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,5120,0.018925334016482036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,5120,0.007428444094128079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,5120,0.01310755560795466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,5120,0.009951999617947472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,4096,0.01754844519827101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,4096,0.007731555236710443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,4096,0.011403555671374003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,4096,0.009191999832789103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,4096,0.017634666628307767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,3584,0.007630222373538547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,3584,0.010407999985747868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,3584,0.008380444513426887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,3072,0.015867556134859722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,3072,0.006612444503439798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,3072,0.010045333041085137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,3072,0.008010666403505538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,2560,0.015154666370815702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,2560,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,2560,0.00922400007645289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,2560,0.007313778003056844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,2048,0.014481777946154276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,2048,0.005203555441564984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,2048,0.008401778009202745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,2560,0.01519466605451372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,2048,0.007037333316273159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,1536,0.013112000293201871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,1536,0.004567111117972268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,1536,0.008022222254011366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,1536,0.006642666541867786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,1024,0.01277599980433782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,1024,0.003882666842805015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,1024,0.007304000357786815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,1024,0.006253333141406377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,768,0.003551111039188173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,768,0.006918222126033571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,512,1536,0.014089778065681458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,768,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,512,0.01204711116022534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,512,0.0034924443397257063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,512,0.006291555447710886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,512,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,256,0.003141333245568805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,256,0.006110222389300664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,256,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,128,0.011358222199810876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,128,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,128,0.006292444550328785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,256,128,0.005554666535721884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,64,0.0025173332542181015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,64,0.006244444598754247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,256,32,0.0026604444202449587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,256,32,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,65536,0.11093244287702773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,65536,0.020844444632530212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,65536,0.10154400269190471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,65536,0.06195555792914497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,10240,0.02628799941804674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,16384,0.037026666932635836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,16384,0.010480889015727572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,16384,0.030171556605233088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,16384,0.01947022146648831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,12288,0.029306666718588933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,12288,0.009114666945404476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,12288,0.024381332927280005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,12288,0.015811555915408664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,10240,0.025807999902301367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,10240,0.009356444080670675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,10240,0.02074399921629164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,8192,0.022858666049109563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,10240,0.013878222141000958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,8192,0.008443555898136562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,8192,0.017222222354676988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,8192,0.01238222254647149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,7168,0.00831200016869439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,7168,0.015834665960735746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,7168,0.011375111010339526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,6144,0.019897777173254225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,6144,0.020115555988417733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,6144,0.007611555357774098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,6144,0.014121777481502958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,6144,0.010688888529936472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,5120,0.008039111064540016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,5120,0.0127368892232577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,5120,0.009673777553770278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,4096,0.01650311052799225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,4096,0.0076044441925154785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,4096,0.011347555451922946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,4096,0.009017777939637503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,3584,0.016563556260532804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,3584,0.007290666302045186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,3584,0.010491555763615502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,3584,0.008640888664457533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,2048,0.014128888646761576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,3072,0.006792000184456508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,3072,0.009741333623727163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,3072,0.007983110845088959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,2560,0.01483022173245748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,2560,0.005917333480384614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,2560,0.00905511114332411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,2560,0.007310222420427535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,2048,0.0136808885468377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,2048,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,2048,0.008354666332403818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,2048,0.006962666908899943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,768,0.014144889182514615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,1536,0.004498666773239772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,1536,0.008015111088752747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,1536,0.006643555644485686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,256,256,0.012724444270133972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,1024,0.003839111162556542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,1024,0.007302222152551015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,1024,0.006228444476922353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,768,0.013791110780504016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,768,0.003516444315512975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,768,0.007085333267847697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,768,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,16384,0.03644977675543891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,512,0.0031999999450312722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,512,0.00655733338660664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,512,0.005616000129116907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,256,0.011067555182509951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,256,0.003152888889114062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,256,0.006224000205596288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,7168,0.021294222937689886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,256,0.005577777822812398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,128,0.011058666639857821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,128,0.002867555452717675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,128,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,384,128,128,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,64,0.002502222235004107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,64,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,128,32,0.002521777732504739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,128,32,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,65536,0.017496888836224873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,16384,0.010049777726332346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,65536,0.10152710808648004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,16384,0.03014577759636773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,12288,0.008436444732877944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,12288,0.0240346673462126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,5120,0.018485332528750103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,10240,0.009291555318567488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,10240,0.020627554919984605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,8192,0.008303110798199972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,8192,0.01680444512102339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,7168,0.00796888851457172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,7168,0.015778667396969266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,6144,0.007335111498832703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,6144,0.014427555931939019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,5120,0.006989333364698622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,5120,0.01275022245115704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,4096,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,4096,0.011375111010339526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,3584,0.006561777657932705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,3584,0.01069777790043089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,3072,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,3072,0.009673777553770278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,2560,0.006597333484225803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,2560,0.009317333499590555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,2048,0.00627377794848548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,2048,0.008608000146018134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,1536,0.004519999855094486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,1536,0.007633777956167857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,1024,0.003814222084151374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,1024,0.006949333681000604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,768,0.003504000190231535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,768,0.0069102222720781965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,512,0.0034497777620951333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,512,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,256,0.006246222390068903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,3584,0.015626667274369132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,256,0.0028560000161329904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,128,0.0028133332315418455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,128,0.006252444452709622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,64,0.0027991111079851785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,64,0.006262222097979651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,64,32,0.002817777709828483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,64,32,0.005971555494599872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,65536,0.017528888252046373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,16384,0.00963466697269016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,65536,0.09964355495240952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,16384,0.02978844443957011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,12288,0.008423999779754216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,12288,0.023175110419591267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,10240,0.00795022232664956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,10240,0.020241777102152508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,8192,0.008343111309740279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,8192,0.016837333639462788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,7168,0.007984889050324758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,7168,0.015460444821251763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,6144,0.007616000043021307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,6144,0.013951111170980664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,5120,0.007249777515729268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,5120,0.012876444392734103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,3584,0.006559111177921295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,4096,0.006611555400821898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,4096,0.011018666956159802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,3584,0.010704889065689512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,3072,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,3072,0.00983199973901113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,2048,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,2560,0.006561777657932705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,2560,0.009318222602208456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,2048,0.008635555704434713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,1536,0.004519999855094486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,3072,0.015511110424995422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,1536,0.007643555601437886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,1024,0.0038364442686239877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,512,0.0031644445326593188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,1024,0.0069724445541699724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,768,0.0035662220584021676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,768,0.006952889263629913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,512,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,128,0.006261333409282897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,256,0.0028293333533737394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,64,0.006284444282452266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,256,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,128,0.0028311111446883944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,64,0.0026595555245876312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,384,32,32,0.0028275555620590844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,384,32,32,0.00629688882165485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,16384,0.5410382482740614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,12288,0.23773243692186144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,16384,0.2830017672644721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,12288,0.41579556465148926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,12288,0.2107315593295627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,1536,0.012420444852775998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,10240,0.2959066761864556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,10240,0.1761937803692288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,1024,0.011756444142924415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,384,128,512,0.011702222128709158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,8192,0.2428577740987142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,8192,0.14317954911126032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,7168,0.15920444329579672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,7168,0.20643556118011475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,7168,0.12540622552235922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,10240,0.2041235499911838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,16384,0.2994240125020345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,6144,0.1825253301196628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,6144,0.10811466640896267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,8192,0.17681154939863417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,16384,0.14421599441104466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,5120,0.1514808866712782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,5120,0.09085066450966729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,12288,0.1101564433839586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,4096,0.11441777812110053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,4096,0.12250755892859565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,4096,0.07603466510772705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,10240,0.09222933318879868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,3584,0.10623821947309707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,3584,0.1069928871260749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,3584,0.06676799721188016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,8192,0.07519644498825073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,7168,0.0665280024210612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,3072,0.09214399920569526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,3072,0.058352887630462646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,6144,0.14459110630883112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,2560,0.09141688876681858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,2560,0.07797333266999987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,2560,0.05015733175807529
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,5120,0.12775822480519614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,2048,0.08409155739678277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,2048,0.06356088982688056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,2048,0.04186755418777466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,6144,0.058392001522911914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,3072,0.09872177574369644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,1536,0.048842665221956044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,1536,0.032744887802335955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,5120,0.049526221222347684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,1024,0.0686275561650594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,1024,0.03466044531928168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,1024,0.027601778507232666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,4096,0.04204444421662224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,3584,0.037117332220077515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,768,0.027132444911532935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,768,0.025295999315049913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,3072,0.033365332418017916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,1536,0.07657333215077718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,512,0.019913777709007263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,512,0.022893332772784766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,2560,0.029696000946892634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,256,0.0274008893304401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,256,0.013549333645237817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,256,0.02056799994574653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,2048,0.025918222135967676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,128,0.02437422176202138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,128,0.011911999848153857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,128,0.01982311076588101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,768,0.03527377711402045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,64,0.009674666656388177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,64,0.017141333884663053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,65536,32,0.010029333333174387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,65536,32,0.01697688963678148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,1536,0.02164266672399309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,65536,0.4727448887295193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,65536,0.29178134600321454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,65536,512,0.06042755312389798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,16384,0.09088089068730672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,16384,0.11722044150034587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,1024,0.01793244481086731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,16384,0.07585510942671034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,12288,0.07466310924953885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,12288,0.08898311191134983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,12288,0.05804444683922661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,65536,0.15066933631896973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,10240,0.06562755505243938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,768,0.016153777639071148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,10240,0.07577955722808838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,10240,0.04940711127387153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,8192,0.05840622054206001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,8192,0.06020977762010363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,8192,0.04084088736110263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,16384,0.042725334564844765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,7168,0.05331733491685656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,7168,0.05375822385152181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,7168,0.036031110419167414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,12288,0.0336444444126553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,65536,0.2980106671651204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,6144,0.04610133171081543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,6144,0.031954665978749595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,10240,0.028620445066028174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,5120,0.04453511039415995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,512,0.014548444085650973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,5120,0.03930222325854831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,5120,0.02758399976624383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,256,0.01367200000418557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,4096,0.040525333748923406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,4096,0.03240799903869629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,8192,0.024685333172480266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,4096,0.02293511066171858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,65536,128,0.013222222526868185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,3584,0.028742220666673448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,3584,0.02102311121092902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,7168,0.022304889228608873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,3072,0.03644444545110067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,3072,0.02528711160024007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,3072,0.019256888164414298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,6144,0.02019999921321869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,2560,0.0210524449745814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,2560,0.017079111602571275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,5120,0.018053332964579265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,2048,0.032560000816980995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,2048,0.01738755570517646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,2048,0.015167999598715039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,4096,0.016132444143295288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,3584,0.014888000157144336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,1536,0.01419911119672987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,1536,0.013431111143694984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,3072,0.01408266690042284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,1024,0.02794933319091797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,1024,0.011252444651391772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,1024,0.012224888636006249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,6144,0.049293332629733615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,768,0.022525333695941504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,768,0.009694221946928237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,768,0.011304000185595619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,2560,0.013134222063753339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,2048,0.011929777761300405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,512,0.0079297779334916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,512,0.009375111096435124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,1536,0.01055555542310079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,256,0.018559111489189994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,256,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,256,0.008657777474986183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,1024,0.009681777821646797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,128,0.016162667009565566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,128,0.004906666775544484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,128,0.008289777570300633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,3584,0.038352890147103205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,64,0.004884444591071871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,64,0.00830577810605367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,16384,32,0.0052764444715446895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,16384,32,0.008681777450773451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,768,0.008656889200210571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,65536,0.3566293451521132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,65536,0.25701512230767143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,2560,0.03467999895413717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,16384,0.0817999972237481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,65536,0.2336853345235189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,16384,0.09239911370807224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,16384,0.05805866585837471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,512,0.007435555259386699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,12288,0.06997866763008966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,12288,0.07140533129374187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,12288,0.04548622171084086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,256,0.007224889265166388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,10240,0.058211558394961886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,1536,0.030763556559880573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,10240,0.038579556677076556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,16384,128,0.0069146665434042616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,8192,0.04686133397950066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,8192,0.032610668076409235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,65536,0.13487199942270914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,7168,0.0473235547542572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,7168,0.04134133458137512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,7168,0.028559999333487615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,16384,0.039465778403811984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,12288,0.03161866797341241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,6144,0.03562666641341315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,6144,0.025568889247046575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,16384,512,0.02566577825281355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,5120,0.03963466816478305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,5120,0.030477331744299993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,5120,0.022648000054889258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,10240,0.026672000686327618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,4096,0.03655111127429538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,4096,0.025274667474958632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,4096,0.018980445133315194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,8192,0.023524444964196947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,7168,0.020948444803555805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,3584,0.022135110365019903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,3584,0.017557332913080852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,6144,0.01908177799648709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,3072,0.03270399901602004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,3072,0.019193778435389202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,3072,0.016172443826993305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,5120,0.017177777157889474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,2560,0.030921777089436848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,2560,0.01646399994691213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,2560,0.014701333310869006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,4096,0.015391111373901367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,2048,0.029109332296583388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,2048,0.014147554834683737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,2048,0.01310488912794325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,3584,0.014307555225160388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,1536,0.027221333649423387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,1536,0.01107022249036365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,1536,0.011848889291286469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,10240,0.05825155311160617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,3072,0.013507555756303998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,1024,0.009986666341622671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,1024,0.01036711119943195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,2560,0.0125937776433097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,8192,0.05191555619239807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,768,0.008590222232871586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,768,0.009489778015348647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,2048,0.011301333705584208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,512,0.020243555307388306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,512,0.00674222202764617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,512,0.007991111112965478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,1536,0.009929777847396003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,6144,0.043776889642079674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,256,0.005559999909665849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,256,0.007634667058785756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,1024,0.008325333396593729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,128,0.016136889656384785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,128,0.004600888739029567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,128,0.007656888829337225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,3584,0.03438933359252082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,64,0.00452888881166776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,64,0.007629333270920648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,12288,32,0.004879999905824661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,12288,32,0.007616889145639207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,768,0.007317333585686154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,65536,0.2972382174597846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,65536,0.20594488249884713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,512,0.006980444822046492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,16384,0.0676399999194675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,16384,0.07574311229917738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,65536,0.20290754901038277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,16384,0.05373422304789225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,256,0.006637333167923822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,12288,0.05513777666621738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,12288,0.057201776239607066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,12288,0.041393776734670006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,10240,0.04883644315931532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,10240,0.047480000389946826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,12288,128,0.006629333313968446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,10240,0.035403556293911405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,65536,0.13138933976491293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,8192,0.03778577844301859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,16384,0.03820000092188517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,8192,0.030518223841985066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,12288,0.030537777476840552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,7168,0.03347288899951511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,7168,0.027048889133665297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,10240,0.02606222199069129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,1024,0.024806222981876794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,6144,0.028922667105992634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,6144,0.02437422176202138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,768,0.02103022237618764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,8192,0.02254666719171736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,5120,0.02463111115826501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,5120,0.021659556362364028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,7168,0.020288000504175823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,6144,0.018353778454992507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,4096,0.020756444997257657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,4096,0.018207111292415194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,12288,256,0.017688888642523024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,3584,0.030265778303146362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,3584,0.01779822177357144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,3584,0.01687111126052009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,5120,0.016576889488432143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,4096,0.015050666199790107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,3072,0.01585422290696038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,3072,0.015560888581805758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,3584,0.013897778259383308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,8192,0.04388622111744351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,2560,0.013423110875818463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,2560,0.014355555176734924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,3072,0.013037333057986366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,7168,0.04021955529848734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,2048,0.011819555527634092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,2048,0.013041777743233575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,6144,0.03717511230044895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,1536,0.024451555477248296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,1536,0.010045333041085137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,1536,0.011411555939250521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,2560,0.012059555285506778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,5120,0.034373333056767784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,1024,0.021931555536058214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,1024,0.008287111090289222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,1024,0.009352888498041365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,2048,0.011014222270912595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,768,0.007243555453088548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,768,0.008603555460770925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,1536,0.008623110751310984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,512,0.01893422173129188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,4096,0.03162489003605313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,512,0.005547555370463266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,512,0.008221333225568136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,1024,0.007880888879299164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,256,0.004543999830881755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,256,0.007592889169851939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,768,0.006948444578382704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,3072,0.028802666399214003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,128,0.015164444843928019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,128,0.004232888834344016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,128,0.007640889121426477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,512,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,64,0.0038515557017591265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,64,0.00758844448460473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,10240,32,0.0041928887367248535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,10240,32,0.007656888829337225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,65536,0.24336444007025826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,2560,0.02717777755525377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,65536,0.16130399703979492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,16384,0.07419377565383911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,16384,0.0605964461962382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,256,0.006512889017661412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,16384,0.04331555631425646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,10240,128,0.0062977779242727495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,12288,0.04727910955746969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,12288,0.03243822190496657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,2048,0.025617778301239014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,65536,0.08334310849507649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,10240,0.039453334278530545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,10240,0.028320001231299505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,16384,0.026241777671708003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,12288,0.020959110723601446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,8192,0.03268266717592875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,8192,0.024588444166713293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,10240,0.018561777141359117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,768,0.0192248887485928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,7168,0.028453333510292902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,7168,0.0218551109234492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,8192,0.016514667206340365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,10240,256,0.01721333298418257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,6144,0.025042666329277888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,6144,0.019882667395803664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,7168,0.014919999572965833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,6144,0.014097778333557976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,5120,0.021306667062971327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,5120,0.017873777283562552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,65536,0.1886195606655545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,4096,0.029491557015313044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,4096,0.017486222916179232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,4096,0.015622221761279635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,12288,0.050873776276906334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,3584,0.028014222780863445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,3584,0.01568266749382019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,3584,0.014665777484575907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,5120,0.01311644415060679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,3072,0.026593777868482802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,3072,0.014139556222491793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,3072,0.013444444371594323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,4096,0.011898666620254517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,2560,0.025514667232831318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,2560,0.01239911135700014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,10240,0.044406221972571484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,2560,0.012068444656001197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,2048,0.02420355545149909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,2048,0.010696888797812991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,2048,0.01129955550034841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,8192,0.040411555104785495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,1536,0.022283555732833013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,1536,0.008980444735950893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,1536,0.00999111102686988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,3584,0.011077333655622272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,3072,0.010704889065689512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,1024,0.007287999822033777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,1024,0.008660444782839881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,2560,0.009655111365848118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,2048,0.008175111479229396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,768,0.005655111124118169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,768,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,7168,0.03679733475049337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,1536,0.007297777467303806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,512,0.004974222017659081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,512,0.007242666350470648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,1024,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,6144,0.03457511133617825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,256,0.0042275554604000514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,256,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,768,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,512,0.005988444305128521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,128,0.003842666745185852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,128,0.0069120000633928515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,5120,0.03182399935192532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,64,0.0038337777886125776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,64,0.006631111105283101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,8192,32,0.0038897775941424896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,8192,32,0.006660444455014334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,65536,0.18428977330525717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,256,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,65536,0.22053421868218315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,65536,0.14406222767300075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,8192,128,0.005919111271699269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,16384,0.05566222137875027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,16384,0.039042668210135564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,65536,0.07952355676227145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,16384,0.025464889076020982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,12288,0.04273600048489041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,12288,0.03104177779621548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,1024,0.01942133406798045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,768,0.017861333158281114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,10240,0.03610488772392273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,10240,0.02723466687732273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,512,0.01722311145729489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,12288,0.02084622283776601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,8192,0.02920000089539422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,8192,0.02353422178162469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,10240,0.018176888426144917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,256,0.016546666622161865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,7168,0.03563822309176127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,7168,0.02569688856601715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,7168,0.021244444780879553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,8192,128,0.015501333607567681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,6144,0.03337422344419692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,6144,0.022112000319692824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,6144,0.01924088928434584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,8192,0.016169778174824186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,5120,0.019162666466501024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,5120,0.030802667140960693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,5120,0.017320000463061862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,4096,0.028764443265067205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,4096,0.015944888194402058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,4096,0.015135111080275642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,7168,0.014815110299322339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,3584,0.026748443643252056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,3584,0.014146667387750415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,6144,0.013770666387346057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,3584,0.014462222655614218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,5120,0.012429333395428128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,3072,0.012893333203262754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,3072,0.013329777452680798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,4096,0.011622221933470832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,2560,0.024262222978803847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,2560,0.011043555206722684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,2560,0.012413333687517377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,16384,0.05985244115193685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,2048,0.023078221413824294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,2048,0.01017955525053872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,2048,0.010868444210953183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,3584,0.010838222172525195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,3072,0.009959999885823991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,1536,0.008632000121805403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,1536,0.009155555731720394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,2560,0.008793777889675563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,1024,0.017719111508793302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,12288,0.049386666880713574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,1024,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,1024,0.007985778152942657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,2048,0.007994666695594788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,768,0.005786666853560342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,768,0.0075635554061995606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,1536,0.0068764446510208984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,512,0.016186666157510545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,512,0.005038222091065513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,512,0.007287110719415877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,1024,0.006562666760550604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,256,0.017183999220530193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,256,0.0042462220622433555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,256,0.0069120000633928515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,10240,0.04379822148217095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,768,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,128,0.003936888857020272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,128,0.0069715554515520734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,8192,0.03901777664820353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,64,0.003629333443111844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,64,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,7168,32,0.0038275557259718576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,512,0.005895111295912001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,7168,32,0.006933333145247565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,65536,0.19275377856360543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,65536,0.15006844202677408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,256,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,65536,0.1332355605231391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,16384,0.05137777659628126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,7168,128,0.005884444548024072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,16384,0.04907288816240099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,16384,0.03772799836264716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,12288,0.043288889858457774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,12288,0.03852799865934584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,12288,0.0306328899330563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,10240,0.038740442858801946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,10240,0.031832890378104314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,10240,0.026867555247412786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,65536,0.07677066988415189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,16384,0.024703999360402424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,8192,0.026921777261628047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,8192,0.022954665952258643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,12288,0.02022133270899455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,7168,0.03178044491344028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,7168,0.02348355617788103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,7168,0.020668443706300523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,10240,0.017647999856207106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,8192,0.015542222393883599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,3072,0.025415110919210646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,6144,0.020426667398876615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,6144,0.019283556275897555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,5120,0.0273991111252043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,5120,0.017500445246696472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,5120,0.017296888762050204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,7168,0.014342221948835584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,1536,0.02125155594613817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,6144,0.013202666408485837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,4096,0.015057777365048727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,4096,0.015032000011867948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,3584,0.024296000599861145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,3584,0.013359111216333179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,3584,0.013988445202509562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,5120,0.012356444365448423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,768,0.017308443784713745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,3072,0.012178666889667511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,3072,0.013059555656380124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,4096,0.011346666349305047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,2560,0.022405332989162866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,3584,0.010248888697889116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,2560,0.01072355525361167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,2560,0.012072000238630505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,7168,128,0.014867555763986377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,2048,0.009303111169073317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,2048,0.010660444696744284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,3072,0.00906844437122345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,1536,0.018887110882335238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,1536,0.007982222570313348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,1536,0.009015111459626092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,2560,0.00833422193924586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,2048,0.007301333049933116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,1024,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,1024,0.007941332956155142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,1536,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,1024,0.00638133328821924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,768,0.005538666827811136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,768,0.007602666815121968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,8192,0.034062223301993474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,6144,0.029799110359615747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,512,0.004628444297446145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,512,0.00692622239391009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,768,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,256,0.016065778003798593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,256,0.0042364444169733255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,256,0.006679111056857639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,512,0.005938666562239329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,128,0.015115555789735583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,128,0.0039013334446483185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,4096,0.025803556044896443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,128,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,64,0.0035439998739295532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,64,0.006584888945023219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,6144,32,0.0035662220584021676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,6144,32,0.0068959999415609576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,65536,0.14706044726901585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,256,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,3072,0.02370044423474206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,65536,0.16873244444529215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,65536,0.12695111168755427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,16384,0.04935822221967909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,16384,0.044044444958368935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,16384,0.0365955564710829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,6144,128,0.005895111295912001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,12288,0.0336844457520379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,12288,0.029335111379623413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,65536,0.07455378108554415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,16384,0.02409066590997908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,10240,0.028347555134031508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,10240,0.025799110531806946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,2048,0.02090577781200409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,8192,0.03317244516478644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,8192,0.023567110300064087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,12288,0.01963911122745938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,8192,0.022191999687088862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,10240,0.01738133364253574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,7168,0.020678222179412842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,7168,0.019997333486874897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,8192,0.015296889675988091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,1024,0.016533333394262526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,6144,0.017772444420390658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,6144,0.018564444449212816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,7168,0.013969777358902825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,768,0.01717333330048455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,5120,0.015457777513398064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,5120,0.01685599982738495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,6144,0.013027555412716336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,6144,512,0.015517334143320719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,4096,0.01384800010257297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,4096,0.014142221874660917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,5120,0.011787555283970304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,4096,0.010382222632567087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,3584,0.012144889268610211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,3584,0.013835555149449242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,3584,0.009472889204819998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,12288,0.04122844338417053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,3072,0.011067555182509951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,3072,0.012714666624863943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,10240,0.03680266605483161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,3072,0.009012444151772393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,2560,0.009654222263230218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,2560,0.011239999698268043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,7168,0.030340443054835003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,2560,0.007844444778230455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,2048,0.008618666893906063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,6144,0.02887822190920512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,2048,0.00962933318482505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,5120,0.026364444030655756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,1536,0.017488888568348356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,1536,0.007558222446176741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,1536,0.008492444124486711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,2048,0.007088888850477006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,1024,0.016188444362746347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,1024,0.0058746664888328975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,1024,0.007655111451943715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,4096,0.024894222617149353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,768,0.004931555440028508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,768,0.007282666862010956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,1536,0.0069368887278768755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,1024,0.006232000059551663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,512,0.0041884444653987885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,512,0.006966222491529252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,3584,0.023704888092146978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,768,0.006246222390068903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,256,0.0038906666967603895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,256,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,512,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,128,0.015112888481881885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,128,0.0038559999730851916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,128,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,256,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,64,0.0035502223504914176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,64,0.006813333266311222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,5120,32,0.003548444559176763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,5120,32,0.006647111227114995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,3072,0.0224906669722663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,65536,0.12830577956305608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,65536,0.13981866836547852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,5120,128,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,65536,0.08643733130560981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,16384,0.0365626679526435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,16384,0.02527911133236355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,65536,0.07233688566419813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,2560,0.021132444341977436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,12288,0.027722666660944622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,16384,0.023031999667485554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,12288,0.02095555596881443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,10240,0.034713778230879046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,10240,0.023217777411142986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,10240,0.01882755590809716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,12288,0.018887110882335238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,8192,0.031614220804638334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,8192,0.01891644464598762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,8192,0.016873778568373788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,10240,0.016827555166350473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,2048,0.01961955593691932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,7168,0.016935111747847665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,7168,0.015299555328157214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,8192,0.014449778530332776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,6144,0.027236445082558527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,6144,0.014930667148696052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,6144,0.01439377831088172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,7168,0.013400889105266996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,6144,0.01237155579858356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,5120,0.012958221965365939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,5120,0.013162666724787818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,768,0.016695999436908297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,4096,0.02365422248840332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,512,0.015140444040298462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,4096,0.011579555769761404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,4096,0.011402666568756104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,3584,0.022557333111763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,3584,0.010724444356229572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,3584,0.011002666420406766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,5120,0.011232888533009423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,3072,0.020229332976871066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,3072,0.009645333720578088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,3072,0.009995555712117089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,4096,0.009306666751702627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,2560,0.018535999788178336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,2560,0.00868622213602066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,2560,0.009017777939637503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,3584,0.008485332959228093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,2048,0.017503110898865592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,2048,0.007921777665615082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,2048,0.007607111500369177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,5120,256,0.016708445217874315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,3072,0.00795022232664956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,1536,0.006575999988449945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,1536,0.0069395556218094295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,2560,0.007594666547245449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,2048,0.007153777612580194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,1024,0.005247999810510212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,1024,0.006619555668698416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,1536,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,1024,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,768,0.004564444637960858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,768,0.006566222343179915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,16384,0.046558221181233726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,512,0.01479288935661316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,512,0.0041964443193541635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,512,0.006254222244024277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,768,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,256,0.014847111370828418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,256,0.0038355555799272326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,256,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,512,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,12288,0.038872000243928694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,256,0.005565333283609814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,7168,0.0292195545302497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,128,0.003612444632583194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,128,0.0058746664888328975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,64,0.003544888976547453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,64,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,4096,32,0.0035662220584021676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,4096,32,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,65536,0.11488089296552871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,65536,0.081313775645362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,4096,128,0.005566222386227713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,16384,0.04596977763705783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,16384,0.03539911243650649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,16384,0.025009777810838487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,65536,0.07112711005740695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,16384,0.022647110952271357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,5120,0.025402666793929204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,12288,0.029342220889197454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,12288,0.020595555504163105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,10240,0.0341635545094808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,10240,0.024650666448805068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,10240,0.01854933301607768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,12288,0.018395556343926322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,8192,0.031210667557186548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,8192,0.020631111330456205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,8192,0.016415999995337594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,10240,0.01646933290693495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,8192,0.014464888307783337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,7168,0.01826844447188907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,7168,0.015116444892353482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,1024,0.015457777513398064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,1536,0.016233777006467182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,6144,0.026743110683229234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,6144,0.016359110673268635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,6144,0.013902222116788229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,5120,0.02512888941499922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,5120,0.014304889572991265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,5120,0.012770666844315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,7168,0.013003555436929068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,4096,0.021935110290845234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,4096,0.011235555840863122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,4096,0.010905777414639791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,768,0.016584888100624084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,3584,0.02051555613676707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,3584,0.01017955525053872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,3584,0.01036000003417333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,6144,0.011970666547616323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,5120,0.01073333372672399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,3072,0.009184000392754873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,3072,0.00965244405799442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,4096,0.00958933350112703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,2560,0.018533334136009216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,2560,0.008273777862389883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,2560,0.008645333349704742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,3584,0.00833333366447025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,2048,0.016576889488432143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,2048,0.007267555428875818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,2048,0.007598222129874759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,3072,0.007928888830873702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,1536,0.016149333781666227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,1536,0.006248000181383557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,1536,0.0069422221018208405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,2560,0.007293333609898885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,2048,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,1024,0.005221333354711533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,1024,0.006689777804745569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,4096,128,0.014471999473041959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,1536,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,768,0.016538666354285348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,768,0.004568888909286923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,768,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,1024,0.006256888724035687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,512,0.004249777644872665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,512,0.006167111297448476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,65536,0.13795467217763266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,256,0.015436444017622205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,768,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,256,0.0038311113086011675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,256,0.005943999936183293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,128,0.015119999647140503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,128,0.0038444445365005066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,128,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,512,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,64,0.0035146665241983202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,64,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3584,32,0.003569777641031477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3584,32,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,256,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,65536,0.10355644755893284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,65536,0.13603199852837458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3584,128,0.005541333307822545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,12288,0.03860622313287523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,16384,0.03195200032658047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,65536,0.07631911171807183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,16384,0.024022221565246582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,12288,0.038007110357284546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,12288,0.022806222240130108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,12288,0.0199991116921107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,65536,0.06994133525424533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,16384,0.02238311039076911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,10240,0.033646222617891096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,10240,0.01923199991385142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,10240,0.01811911165714264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,12288,0.018231110440360177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,8192,0.030455999904208716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,8192,0.01607111096382141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,8192,0.01583377851380242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,7168,0.028189334604475234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,7168,0.014395554860432943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,10240,0.016181333197487723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,7168,0.014449778530332776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,6144,0.02664799988269806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,6144,0.013040000365840064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,6144,0.013731555806265937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,8192,0.014152000347773233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,7168,0.028549333413441975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,5120,0.011747555600272285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,5120,0.01275466630856196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,7168,0.012812444733248817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,4096,0.022608889473809138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,4096,0.01036533299419615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,4096,0.010783111055692038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,6144,0.011732444167137146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,3584,0.01958666741847992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,3584,0.009343110852771336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,3584,0.00962222201956643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,5120,0.00979022185007731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,3072,0.018812444474962022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,3072,0.0086666668454806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,3072,0.009023110899660323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,4096,0.008642666869693333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,2560,0.01768977774514092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,2560,0.0075804442167282104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,2560,0.008273777862389883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,3584,0.008279110822412703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,2048,0.016182222300105624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,2048,0.0069004446268081665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,2048,0.007277333074145847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,3072,0.007659555309348636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,1536,0.015613334046469795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,1536,0.005955555372767978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,1536,0.006990222467316522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,2560,0.0074337778819931885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,1024,0.014811555544535318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,1024,0.004701333327425851
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,1024,0.006291555447710886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,2048,0.0069360000391801195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,3072,0.01958577831586202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,768,0.004454222404294544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,768,0.006294222341643439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,1536,0.006631111105283101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,512,0.013773333695199756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,512,0.003979555434650845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,512,0.006258666515350342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,1024,0.006225777582989798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,256,0.015370666980743408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,256,0.003571555432346132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,256,0.005945777727497949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,768,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,128,0.015124445160230001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,128,0.0032151111712058387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,128,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,512,0.005585777676767773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,1024,0.01517599986659156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,64,0.003185777821474605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,64,0.006227555374304454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,3072,32,0.0032586666444937387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,3072,32,0.0058782220714622075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,65536,0.08339822292327881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,256,0.005528000079923206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,65536,0.07332711087332831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,3072,128,0.005568000177542369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,16384,0.029076443778143987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,16384,0.023334221707450017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,65536,0.06898222366968791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3584,512,0.014444443914625378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,12288,0.03733777668741014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,12288,0.02310666607485877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,12288,0.0195777780479855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,16384,0.021864000293943617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,10240,0.020980444219377305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,10240,0.01753866672515869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,12288,0.017987555927700467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,8192,0.029904001288943823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,8192,0.015468445089128284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,8192,0.015425778097576566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,10240,0.015847111741701763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,7168,0.02776355544726054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,7168,0.014112000664075216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,7168,0.014131555954615274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,16384,0.04549244377348158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,6144,0.02574311031235589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,6144,0.0125928885406918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,8192,0.013791110780504016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,6144,0.013405332962671915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,5120,0.02387733260790507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,5120,0.011410666836632622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,5120,0.012093333734406365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,7168,0.012322666744391123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,6144,0.011056889262464313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,5120,0.009361777868535783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,4096,0.009886222581068674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,4096,0.009958222508430481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,4096,0.00869422240389718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,3584,0.009006222089131674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,3584,0.009650666680600908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,5120,0.024392000503010217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,3072,0.018174222773975797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,3072,0.008415111237102086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,3072,0.008622222476535374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,3584,0.008325333396593729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,2560,0.017165333032608032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,2560,0.007505777809354994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,2560,0.007893333004580604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,3072,0.007639110916190677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,2048,0.015848888291252982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,2048,0.006588444527652528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,2048,0.007250666618347168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,2560,0.0074346669846110875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,1536,0.015461333923869662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,1536,0.005843555761708154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,1536,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,2048,0.006908444480763541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,1024,0.014149333039919535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,1024,0.004582222137186262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,1024,0.006590222318967183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,1536,0.006568888823191325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,768,0.016502221425374348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,768,0.0042275554604000514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,768,0.006245333287451003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,3072,768,0.016419554750124615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,512,0.013473778135246701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,512,0.003939555750952827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,512,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,1024,0.005968889014588461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,256,0.015120888749758402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,256,0.0035048888789282907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,256,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,768,0.005938666562239329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,512,0.005664000080691443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,128,0.0031715554909573663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,128,0.005947555518812603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,256,0.005566222386227713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,64,0.003167111012670729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,64,0.006258666515350342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2560,32,0.0031964443624019623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2560,32,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,65536,0.1107306679089864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2560,128,0.005582222094138463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,65536,0.06719822353786893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,65536,0.0712053312195672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,16384,0.04460888769891527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,16384,0.022965333527988855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,65536,0.06708088848325941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,65536,0.13548266887664795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,16384,0.022620444496472675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,12288,0.03272622161441379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,12288,0.020969778299331665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,12288,0.018979556030697294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,10240,0.02920088834232754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,10240,0.03333955672052171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,10240,0.017136888371573556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,10240,0.017263111140992906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,16384,0.02158222264713711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,8192,0.014521777629852295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,12288,0.01755555636352963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,8192,0.014774221513006421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,10240,0.015605333778593274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,7168,0.013203555511103736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,7168,0.014053333136770459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,8192,0.013169777890046438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,6144,0.02293955617480808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,6144,0.011964444484975604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,6144,0.013002666334311167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,7168,0.011375111010339526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,5120,0.020278222031063504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,5120,0.01070666644308302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,5120,0.01185155577129788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,6144,0.010394666757848527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,5120,0.009327111144860586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,4096,0.009508444203270806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,4096,0.009369778136412302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,4096,0.008642666869693333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,4096,0.021514667405022517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,3584,0.00868266655339135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,3584,0.009005332986513773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,3584,0.008348444269763099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,3072,0.016543999314308167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,3072,0.008166222108734978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,3072,0.008002666963471307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,3584,0.018631110588709515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,2560,0.016581333345837064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,2560,0.006980444822046492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,2560,0.007623111208279927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,3072,0.007888000044557784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,2048,0.0151582227812873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,2048,0.00627377794848548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,2048,0.007072889142566257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,2560,0.007297777467303806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,2048,0.006931555353932911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,1536,0.014896889527638754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,1536,0.005295111073387994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,1536,0.0069271110826068455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,1024,0.013442666994200813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,1024,0.004518222063779831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,1024,0.006258666515350342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,1536,0.006383111079533895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,768,0.01647822227742937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,768,0.00388711111413108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,768,0.006305777778228124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,1024,0.0062977779242727495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,768,0.005945777727497949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2560,128,0.014829332629839579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,512,0.003638222399685118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,512,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,256,0.015120888749758402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,256,0.0031848889258172777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,256,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,512,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,128,0.014832889040311178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,128,0.003195555466744635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,256,0.005571555760171678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,128,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,64,0.003212444484233856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,64,0.005945777727497949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,2048,32,0.0032151111712058387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,2048,32,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,16384,0.03859022259712219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,65536,0.1096000009112888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,65536,0.05355911122428047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,2048,128,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,65536,0.06874310970306396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,16384,0.038219554556740656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,16384,0.019710222880045574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,16384,0.021928888228204515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,12288,0.032636443773905434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,12288,0.016861332787407767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,65536,0.06606488757663302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,12288,0.01871288816134135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,16384,0.021307556165589228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,10240,0.015200000670221118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,10240,0.016965332958433364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,12288,0.017253332667880587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,8192,0.02711733347839779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,8192,0.014539556370841132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,8192,0.014856888188256158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,10240,0.01443822185198466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,7168,0.024656888511445787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,7168,0.025015999873479206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,7168,0.012840888566441007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,7168,0.013487111363146039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,6144,0.021328889661365088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,6144,0.011755555868148804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,8192,0.01237777786122428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,6144,0.012381333443853589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,7168,0.010796444283591377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,5120,0.010371555884679159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,5120,0.010575110713640848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,6144,0.010028444230556488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,5120,0.009389333426952362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,4096,0.00868000007337994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,4096,0.008904889225959778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,4096,0.018762666318151686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,3584,0.017216889394654166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,3584,0.00793688909875022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,4096,0.00867022242810991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,3584,0.008295111358165741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,3072,0.01684711045689053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,3072,0.007303111255168915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,3072,0.008046222229798635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,3584,0.008037333687146505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,2560,0.016228444046444364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,2560,0.0063324446479479475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,2560,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,3072,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,2048,0.015174221661355762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,2048,0.006162666612201267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,2048,0.006934222247865465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,2560,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,2048,0.0069351109365622205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,3584,0.01755466726091173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,1536,0.005211555709441503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,1536,0.006721777634488211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,1536,0.0063573333124319715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,1024,0.0045422220395671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,1024,0.006229333165619109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,1024,0.006063999815119638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,768,0.016206221448050607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,768,0.004254222330119875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,768,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,768,0.005727111051479976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,2048,512,0.012835555606418185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,512,0.003927111211750242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,512,0.005981333139869902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,512,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,256,0.014492443866199918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,256,0.003555555724435382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,256,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,10240,0.028936889436509874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,128,0.014163555370436775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,8192,0.025958221819665697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,128,0.0032186667538351486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,128,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,64,0.0031911111954185697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,256,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,64,0.005940444353553984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1536,32,0.003192000091075897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1536,32,0.005884444548024072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,65536,0.03846933444341024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,65536,0.0667475528187222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1536,128,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,16384,0.03649333450529311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,16384,0.016530666086408827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,16384,0.021691555778185528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,65536,0.0650213360786438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,16384,0.02054666644997067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,5120,0.01957155598534478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,12288,0.014148443937301636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,12288,0.018453333112928603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,4096,0.018012444178263348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,10240,0.02718399961789449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,10240,0.012716444830099741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,10240,0.01647200021478865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,8192,0.023152000374264185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,12288,0.016033778587977093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,8192,0.01400888959566752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,8192,0.013791110780504016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,7168,0.021851556168662176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,10240,0.013074666261672974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,7168,0.012791111237472959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,7168,0.012680889003806643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,8192,0.011405333048767514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,6144,0.020464888877338834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,6144,0.011304000185595619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,6144,0.0107022225856781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,5120,0.018918222851223417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,7168,0.010732444624106089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,5120,0.00997866690158844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,5120,0.009698666632175446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,4096,0.017553778158293832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,6144,0.010016000105275048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,4096,0.008345777789751688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,4096,0.008672888908121321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,3584,0.016812443733215332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,5120,0.009298666483826106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,3584,0.007625777688291337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,3584,0.008427555362383524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,4096,0.008452444440788692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,3072,0.006752000086837345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,3072,0.007654222349325816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,3584,0.007989333735571967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,2560,0.015673778123325773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,2560,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,2560,0.007368889119890001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,3072,0.007635555333561367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,2048,0.014816888504558139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,2048,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,2048,0.006940444310506185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,2560,0.007142222589916653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,2048,0.00666933341158761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,1536,0.004885333279768626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,1536,0.006634666687912411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,1536,0.014801777071423002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,1536,0.006291555447710886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,1024,0.013434666726324292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,1024,0.0041786668201287585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,1024,0.00629066675901413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,1024,0.013444444371594323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,768,0.003867555409669876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,768,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,1024,0.005953777581453323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,512,0.012440000143316058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,512,0.003547555456558863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,768,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,512,0.005943999936183293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,512,0.00554577757914861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,256,0.0031964443624019623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,256,0.005983110931184556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,256,0.005244444641802046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,128,0.014134221606784396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1536,512,0.013415111435784234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,128,0.0028577778074476454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,128,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,64,0.0028364445186323593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,64,0.00591822216908137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,1024,32,0.0028320000403457214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,1024,32,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,1024,128,0.0052666668262746595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,65536,0.03233777814441257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,65536,0.0655040012465583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,65536,0.06439910994635688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,65536,0.10249955786599053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,16384,0.01368622233470281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,16384,0.021385777327749465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,12288,0.030861334668265447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,16384,0.020222221811612446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,12288,0.029493331909179688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,12288,0.013007111019558378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,12288,0.0181697772608863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,12288,0.014534221755133735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,10240,0.02667733364635044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,10240,0.012034667034943899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,10240,0.01611822181277805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,10240,0.013094222380055321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,8192,0.023006222314304773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,8192,0.011191999746693505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,8192,0.01237511138121287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,8192,0.011439999772442712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,7168,0.010658666491508484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,7168,0.011418666276666852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,3072,0.016540444559521146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,7168,0.01072088877360026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,6144,0.009915555516878763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,6144,0.0103493332862854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,6144,0.010060444474220276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,1536,0.014083556003040738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,5120,0.009005332986513773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,5120,0.009697777529557547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,5120,0.009293333523803288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,4096,0.017330666383107502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,4096,0.007652444144090016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,4096,0.008610666626029545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,4096,0.008291555775536431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,3584,0.016554666890038382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,3584,0.007287999822033777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,3584,0.00832622249921163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,768,0.015832000308566622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,3584,0.007982222570313348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,3072,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,3072,0.016535111599498324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,3072,0.00793333351612091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,2560,0.015492444237073263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,2560,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,2560,0.007634667058785756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,3072,0.007667555577225155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,1024,256,0.014495111174053617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,2048,0.0052666668262746595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,2560,0.00702400008837382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,2048,0.006968888971540663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,1536,0.014117333624098035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,1536,0.004946666873163647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,1536,0.006796444455782573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,2048,0.006633777585294511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,1024,0.013169777890046438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,1024,0.0041404445138242514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,1024,0.006468444648716185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,1536,0.00647288892004225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,65536,0.10153066449695164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,768,0.0038373333712418876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,1024,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,768,0.006245333287451003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,512,0.013075555364290873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,512,0.00349866681628757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,512,0.006224000205596288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,768,0.005567111074924469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,512,0.005583999885453119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,256,0.0035182221068276297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,256,0.005683555371231503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,16384,0.03643466697798835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,128,0.013800000150998434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,256,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,128,0.0028808888875775864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,128,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,64,0.0028213332924577924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,64,0.005913777897755305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,768,32,0.0028293333533737394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,768,32,0.005960000058015187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,768,128,0.0052666668262746595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,65536,0.025960000024901494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,65536,0.06466488705741034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,65536,0.06434933344523112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,7168,0.021677333447668288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,16384,0.012151111331250934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,16384,0.021464889248212177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,16384,0.019107555349667866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,6144,0.020259555843141343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,12288,0.013915555344687568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,12288,0.017108443710539076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,12288,0.01440000037352244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,5120,0.01885777711868286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,10240,0.012449777788586087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,10240,0.024693333440356787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,10240,0.014128888646761576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,8192,0.021984888447655573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,8192,0.011351111034552256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,10240,0.01276000009642707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,8192,0.011688888900809817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,8192,0.011687999798191918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,7168,0.010318222145239512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,7168,0.011002666420406766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,7168,0.01056177748574151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,6144,0.019567999574873183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,6144,0.009569777382744683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,6144,0.010310221877362993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,6144,0.009900444083743626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,2048,0.01499111122555203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,5120,0.008343111309740279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,5120,0.009729777773221334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,768,0.015488889482286243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,768,256,0.014436443646748861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,4096,0.01681155628628201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,4096,0.007658667034573025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,5120,0.009353777600659264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,4096,0.008646222452322641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,4096,0.00850044439236323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,3584,0.007261333366235097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,3584,0.007996444072988298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,3584,0.00798044436507755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,3072,0.015182221929232279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,3072,0.00646577775478363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,3072,0.007666666474607255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,65536,0.10013688935173883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,2560,0.01481599940194024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,3072,0.007660444411966536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,2560,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,2560,0.007594666547245449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,2048,0.0052595556610160405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,2560,0.007306666837798224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,2048,0.007037333316273159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,2048,0.0069022224181228215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,1536,0.004575110971927643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,1536,0.0069297779765393995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,16384,0.03506044546763102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,1536,0.006291555447710886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,1024,0.0038444445365005066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,1024,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,1024,0.005955555372767978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,12288,0.029375112719006006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,768,0.00350577798154619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,768,0.006224888894293044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,768,0.005906666732496685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,512,0.012085333466529846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,512,0.0035297779573334586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,512,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,7168,0.02057422200838725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,256,0.013428444663683573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,256,0.0032071111102898917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,512,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,256,0.0058142224119769196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,128,0.012696888711717395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,128,0.002801777794957161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,256,0.005592888842026393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,128,0.005895111295912001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,64,0.00286666676402092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,64,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,512,128,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,512,32,0.0027911110470692315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,512,32,0.0058711109062035876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,65536,0.01971733404530419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,65536,0.0996133353975084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,65536,0.06445244285795423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,16384,0.033675554725858904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,65536,0.06327644321653578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,5120,0.01852799952030182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,16384,0.011063999599880643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,16384,0.019705777366956074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,12288,0.026838221483760413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,12288,0.009660444325870937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,12288,0.015449777245521545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,16384,0.017260443833139207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,10240,0.009697777529557547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,12288,0.014142221874660917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,10240,0.013421333498424955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,8192,0.02199733257293701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,10240,0.013041777743233575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,8192,0.0083137775460879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,8192,0.011359111302428775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,7168,0.01993333299954732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,8192,0.01145244472556644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,7168,0.007908444437715743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,7168,0.011010666688283285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,6144,0.0192275560564465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,7168,0.010652444428867765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,6144,0.007622222105662028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,6144,0.009972444838947719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,6144,0.009746666583749983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,5120,0.00797333319981893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,5120,0.009668444593747457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,5120,0.009283555878533257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,3584,0.015972443752818637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,4096,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,4096,0.008640888664457533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,4096,0.008289777570300633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,3584,0.015493333339691162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,3584,0.006949333681000604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,3584,0.00795555528667238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,3584,0.00793422261873881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,3072,0.014537778165605334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,3072,0.006575111299753189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,3072,0.007711999946170383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,3072,0.007463110817803278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,2560,0.01441333360142178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,2560,0.00591822216908137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,2560,0.007263110743628607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,2560,0.006960888703664144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,2048,0.014157333307796054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,2048,0.005218666460778978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,2048,0.006979555719428592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,1536,0.013815110756291283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,2048,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,1024,0.012773333324326409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,1536,0.004576000074545543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,1536,0.006593777901596493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,512,768,0.015114666687117683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,1536,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,1024,0.01239111108912362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,1024,0.004182222402758068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,1024,0.006588444527652528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,768,0.01444533301724328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,1024,0.005940444353553984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,768,0.0035111109415690103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,768,0.0060159998635451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,512,0.011184000306659274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,768,0.005719111197524601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,512,0.003208000005947219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,512,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,512,0.005592888842026393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,256,0.003181333343187968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,256,0.005945777727497949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,256,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,128,0.011067555182509951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,128,0.0028568889117903183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,128,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,64,0.0028808888875775864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,256,128,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,64,0.005616888817813661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,256,32,0.002875555513633622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,256,32,0.005573333137565189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,65536,0.016521778371598985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,65536,0.0632515549659729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,65536,0.061850666999816895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,16384,0.03195910983615451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,16384,0.010034666293197209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,16384,0.018900444110234577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,16384,0.01720711092154185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,12288,0.026242666774325903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,12288,0.008355555435021719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,12288,0.01480977733929952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,12288,0.014175110393100314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,10240,0.024002666274706524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,10240,0.009032888544930352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,10240,0.013109332985348172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,10240,0.01276000009642707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,8192,0.021617778473430212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,8192,0.008052444292439355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,8192,0.011703111231327057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,8192,0.01146577795346578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,5120,0.017502221796247695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,7168,0.007647111184067196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,7168,0.010777778095669217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,7168,0.010657777388890585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,4096,0.016513778103722464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,6144,0.007293333609898885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,6144,0.010334221853150262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,6144,0.009708444277445475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,2048,0.013421333498424955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,5120,0.008023999631404877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,5120,0.009335111412737105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,5120,0.009299555586444007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,4096,0.015503999259736804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,4096,0.007278222176763747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,4096,0.008336000144481659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,4096,0.008343111309740279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,3584,0.015124445160230001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,3584,0.006931555353932911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,3584,0.00795111142926746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,1536,0.01346577786737018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,3584,0.00793333351612091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,3072,0.015186667442321777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,3072,0.006610666712125142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,3072,0.00794933322403166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,3072,0.00738933351304796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,2560,0.014500444134076437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,2560,0.0063440000845326324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,2560,0.007381333245171442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,2560,0.007050666544172499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,2048,0.005924444645643234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,256,256,0.012746666868527731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,2048,0.00692622239391009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,1536,0.012724444270133972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,2048,0.006803555621041193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,1536,0.0045422220395671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,1536,0.006577777779764599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,1536,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,1024,0.0038533334930737815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,1024,0.006265777680608962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,1024,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,768,0.013406222065289816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,768,0.0035573335157500375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,768,0.0059573331640826324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,768,0.005566222386227713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,65536,0.09849866893556382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,512,0.003351111171974076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,512,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,512,0.00546488869521353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,10240,0.023319111929999456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,256,0.0028817777832349143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,256,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,256,0.005261333452330695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,7168,0.0199608885579639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,128,0.0026337777574857077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,128,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,6144,0.018187556001875136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,256,128,128,0.005583999885453119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,64,0.0025235555238193935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,64,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,128,32,0.002809777855873108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,128,32,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,65536,0.01553244392077128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,5120,0.017536888519922893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,65536,0.06262399752934773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,16384,0.009668444593747457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,16384,0.018208000395033095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,12288,0.008277333445019191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,12288,0.014139556222491793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,10240,0.00795111142926746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,10240,0.01310755560795466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,8192,0.007285333342022366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,8192,0.011739555332395764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,7168,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,7168,0.010686222049925061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,6144,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,6144,0.010095111197895473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,5120,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,5120,0.009469333622190688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,4096,0.006569777925809224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,4096,0.008271999657154083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,3584,0.006600889066855113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,3584,0.008284444610277811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,3072,0.006506666541099548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,2560,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,3072,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,2560,0.006963555183675554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,2048,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,2048,0.006945778098371293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,1536,0.004559111264016894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,1536,0.006595555692911148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,1024,0.0038568890757030914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,1024,0.00629066675901413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,768,0.003506666670242945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,768,0.006280000011126201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,512,0.003144888828198115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,256,0.0028479999552170434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,512,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,128,0.0028008888992998335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,256,0.005728000154097875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,128,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,64,0.00254133323000537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,64,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,64,32,0.0024871110088295406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,64,32,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,65536,0.015439111325475903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,65536,0.06151555644141304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,16384,0.008981333010726506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,16384,0.01720533271630605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,12288,0.008303110798199972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,12288,0.014732443624072604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,10240,0.013400000002649097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,10240,0.00775733341773351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,8192,0.007295110987292395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,8192,0.011392888923486074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,7168,0.0068959999415609576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,7168,0.010989333192507425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,6144,0.006575111299753189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,6144,0.00998577806684706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,5120,0.006562666760550604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,5120,0.009531555904282464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,4096,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,4096,0.008467555873923832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,3584,0.006288888967699475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,3584,0.007655999726719326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,3072,0.006224000205596288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,3072,0.007640000018808577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,2560,0.0063368889192740125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,2560,0.007312888900438945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,2048,0.006903111106819577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,2048,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,1536,0.004531555705600315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,1536,0.006566222343179915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,1024,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,1024,0.0038728887836138406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,768,0.0035208890007601846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,768,0.006246222390068903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,512,0.003144888828198115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,512,0.005777777896987067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,256,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,128,0.0028453332682450614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,256,0.0028577778074476454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,128,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,64,0.0025439999169773525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,64,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,256,32,32,0.0028213332924577924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,256,32,32,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,16384,0.4161884519788954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,16384,0.29533065689934623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,16384,0.27110134230719674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,12288,0.3217182159423828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,2048,0.013438222308953604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,1024,0.012372444073359171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,12288,0.19906311564975313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,512,0.011025778121418424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,10240,0.2593759960598416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,10240,0.2111119959089491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,256,0.011398221883508893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,10240,0.1703102191289266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,8192,0.2100364499621921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,8192,0.17422133021884492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,8192,0.13834933439890543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,7168,0.15569155746036106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,7168,0.18497688240475127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,7168,0.12204889456431071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,6144,0.14254844188690186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,6144,0.15875021616617838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,6144,0.10625511407852173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,256,128,128,0.010027555955780877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,12288,0.23452533615960014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,5120,0.1332533359527588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,5120,0.09022666348351373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,5120,0.12587111526065403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,16384,0.13925600051879883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,4096,0.10768800311618382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,4096,0.07215466764238146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,12288,0.10564711358812119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,3584,0.10470133357577854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,3584,0.09472355577680801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,3584,0.0639537771542867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,10240,0.08892799748314752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,8192,0.07242577605777316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,3072,0.0824506680170695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,3072,0.0561457806163364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,7168,0.06447555621465047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,2560,0.08989689085218643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,2560,0.06923555665545993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,2560,0.04859822326236301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,6144,0.056127998563978404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,2048,0.08306133084826998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,2048,0.05605866511662801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,2048,0.040064887868033514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,4096,0.11253155602349176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,1536,0.07531555493672688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,1536,0.043113778034845986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,1536,0.031152000029881794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,3072,0.09739822149276733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,1024,0.06772266493903266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,1024,0.030930668115615845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,1024,0.025568889247046575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,5120,0.04778666628731621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,768,0.034394668208228216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,768,0.024272888898849487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,768,0.02370044423474206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,4096,0.04011911153793335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,512,0.059642665916019015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,512,0.01760444376203749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,512,0.021419554948806763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,3584,0.035424888134002686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,3072,0.03198488884501987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,256,0.012811555630630918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,256,0.019588443968031142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,2560,0.028012444575627644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,128,0.024343111448817786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,128,0.01071911139620675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,128,0.018195556269751657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,2048,0.024813334147135418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,64,0.009470221896966299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,64,0.017200888858901132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,65536,32,0.01071911139620675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,65536,32,0.017065778374671936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,65536,0.2900613413916694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,65536,0.41606489817301434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,65536,0.2800862259334988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,1536,0.020622221959961783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,65536,256,0.027245332797368366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,16384,0.10536444187164307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,65536,0.1476915544933743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,16384,0.07372355461120605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,12288,0.07311999797821045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,12288,0.08055466413497925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,12288,0.05628266599443224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,1024,0.017086222767829895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,10240,0.06440622276730008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,10240,0.06781244277954102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,10240,0.04800533254941305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,768,0.015662221444977652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,16384,0.04187822341918945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,8192,0.05609689156214396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,8192,0.04017511010169983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,512,0.014147554834683737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,7168,0.05177422364552816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,7168,0.048519111341900296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,12288,0.03288799855444167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,7168,0.034854223330815635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,6144,0.048152890470292836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,6144,0.04204622242185804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,6144,0.03088888857099745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,10240,0.028537776735093858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,5120,0.0439626673857371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,5120,0.036017778846952654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,5120,0.0269804447889328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,256,0.01311555587583118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,4096,0.040018667777379356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,65536,128,0.013055110971132914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,4096,0.02960977620548672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,16384,0.08940000004238552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,4096,0.022445332672860887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,3584,0.038006222910351224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,3584,0.026485333840052288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,3584,0.020488889680968392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,8192,0.02438933319515652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,3072,0.02344177828894721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,3072,0.018831999765502084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,7168,0.02195288903183407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,2560,0.03449599941571554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,2560,0.019567110472255282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,2560,0.016905777984195285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,6144,0.020112888680564035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,2048,0.032116442918777466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,2048,0.016318221886952717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,2048,0.015001777145597668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,8192,0.057413332992129855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,5120,0.017903111047214933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,1536,0.013517333401574029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,1536,0.013238222234778933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,4096,0.0162453336848153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,3584,0.014663999279340109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,1024,0.027793778313530818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,1024,0.010659555594126383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,1024,0.012070222033394707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,3072,0.013781333135233985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,768,0.009022222624884712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,768,0.011042666931947073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,2560,0.012711111042234631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,2048,0.01181422256761127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,512,0.00740622232357661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,512,0.009602666729026372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,1536,0.010382222632567087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,256,0.018056000272432964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,256,0.005245333330498801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,256,0.008363555702898238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,1024,0.008976000050703684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,128,0.016297777493794758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,128,0.004212444441186057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,128,0.008285333712895712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,768,0.0079644446571668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,64,0.004910222358173794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,64,0.008379555410808986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,16384,32,0.004899555610285865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,16384,32,0.008601778083377415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,65536,0.24925600157843697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,512,0.007268444531493717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,65536,0.3103235562642415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,3072,0.03609155615170797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,65536,0.2157804436153836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,16384,0.08018133375379774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,16384,0.07823733488718669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,16384,0.05730755461586846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,256,0.006948444578382704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,12288,0.0600151088502672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,12288,0.04365066687266032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,1536,0.03011200163099501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,16384,128,0.0069022224181228215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,10240,0.05769866704940796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,10240,0.05076533224847582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,10240,0.037660443120532565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,8192,0.05091555582152473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,8192,0.042287998729281955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,8192,0.03152711192766825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,65536,0.13370045026143393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,768,0.022333333889643352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,7168,0.03672444489267137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,7168,0.028223110569847956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,16384,0.039501332574420504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,6144,0.04262755645645989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,6144,0.03142933381928338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,6144,0.025120000044504803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,16384,512,0.02437777817249298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,5120,0.03912355502446493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,5120,0.02699822187423706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,5120,0.022097777989175584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,12288,0.03107822272512648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,4096,0.03592888845337762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,4096,0.022825777530670166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,4096,0.018575999471876357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,10240,0.026711111267407734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,8192,0.023335999912685815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,3584,0.02019377715057797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,3584,0.017198221551047433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,7168,0.020847999387317236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,6144,0.018965333700180054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,3072,0.01752000053723653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,3072,0.01609511176745097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,5120,0.01699911057949066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,2560,0.03047555685043335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,2560,0.01480977733929952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,2560,0.014499555031458536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,4096,0.015385778413878547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,3584,0.014369777507252164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,2048,0.012734221915404001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,2048,0.013037333057986366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,3072,0.013407111167907715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,12288,0.0652666687965393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,1536,0.026758222116364375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,1536,0.01107022249036365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,1536,0.01202755586968528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,2560,0.012078222301271228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,1024,0.02365066607793172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,1024,0.009009777671760982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,1024,0.010040889183680216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,768,0.020575111111005146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,768,0.007628444168302748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,768,0.008948444492287105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,7168,0.04588977826966179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,2048,0.011128889189826118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,512,0.020279111133681405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,512,0.005903111149867375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,512,0.008344888687133789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,256,0.017551110850440133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,256,0.004572444491916233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,256,0.007628444168302748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,1536,0.009368889033794403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,128,0.015855999456511605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,128,0.0038933331767717996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,128,0.007659555309348636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,1024,0.0075342224703894714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,64,0.0036311112344264984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,64,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,12288,32,0.0038542221817705366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,12288,32,0.007605333295133378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,3584,0.03360088997417026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,768,0.007288888924651676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,65536,0.2935555511050754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,65536,0.19787111547258165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,3072,0.03199822372860379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,16384,0.07515733109580146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,16384,0.052292442984051175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,512,0.006918222126033571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,256,0.006633777585294511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,12288,0.05766310956743029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,12288,0.04068177938461304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,12288,128,0.006664888726340399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,65536,0.1287226676940918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,10240,0.048282666338814624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,12288,2048,0.029150221082899306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,10240,0.03494666682349311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,16384,0.03809422254562378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,8192,0.039670222335391574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,8192,0.030397332376903955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,12288,0.029930667744742498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,10240,0.025739555557568867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,7168,0.034782220919926964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,7168,0.02664799988269806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,8192,0.02260711126857334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,6144,0.03638044330808852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,6144,0.030584888325797185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,6144,0.024067555864651997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,7168,0.02016177773475647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,6144,0.018552889426549275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,5120,0.026203556193245783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,5120,0.02146400014559428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,16384,0.06554755237367418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,65536,0.19786222775777182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,4096,0.03141244583659702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,4096,0.018029333816634286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,4096,0.02199199961291419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,3584,0.029791110091739233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,3584,0.018950222267044913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,3584,0.01687733332316081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,12288,0.05410933494567871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,3072,0.028076443407270644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,3072,0.017270222306251526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,3072,0.015523556205961438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,10240,0.04769777920511034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,5120,0.01668444441424476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,2560,0.01478844384352366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,2560,0.014256889621416727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,4096,0.014762666490342883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,2048,0.025395555628670588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,2048,0.012914666699038612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,8192,0.04253155655331082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,3584,0.013726222018400827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,2048,0.012814222110642327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,1536,0.02404533326625824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,1536,0.010666666759385003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,1536,0.011344888971911537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,1024,0.02093244426780277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,1024,0.00889511075284746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,1024,0.009362666971153682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,3072,0.012984889249006907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,768,0.018567999203999836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,768,0.0069679998689227635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,768,0.008691555923885768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,7168,0.038607110579808555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,2560,0.012059555285506778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,512,0.00564177789621883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,512,0.007995555798212687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,2048,0.010089777410030365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,256,0.01735111077626546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,256,0.004862222406599256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,256,0.007321777443091075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,1536,0.008460444708665213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,1024,0.007368000017272101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,128,0.01612711118327247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,128,0.004202666795916027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,128,0.007313778003056844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,768,0.00703022215101454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,64,0.0038319999972979226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,64,0.007301333049933116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,10240,32,0.003858666867017746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,10240,32,0.007652444144090016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,65536,0.18118845091925728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,65536,0.21054932806226942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,65536,0.14777866999308267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,512,0.006576889091067844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,16384,0.054604444238874644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,16384,0.040285332335366145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,256,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,12288,0.049858666128582425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,12288,0.04248266749911838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,10240,128,0.006593777901596493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,12288,0.03173066510094537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,5120,0.03358488943841722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,10240,0.03502666619088914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,10240,0.02790133489502801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,65536,0.08199910985098945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,16384,0.026037333740128413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,8192,0.02939200070169237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,8192,0.02425777746571435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,12288,0.021073778470357258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,7168,0.03601333167817857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,7168,0.025733333494928148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,7168,0.021546666820844013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,10240,0.01850844422976176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,6144,0.03382577829890781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,6144,0.022384888596004907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,6144,0.019874667127927143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,2560,0.0265964451763365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,5120,0.030982222821977403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,8192,0.016534222496880423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,5120,0.0195448895295461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,5120,0.017652443713612027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,4096,0.028584000137117174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,4096,0.016172443826993305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,4096,0.015488889482286243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,7168,0.014875556031862894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,3584,0.0277164445983039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,3584,0.01441422270403968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,3584,0.014184888866212634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,6144,0.013802666631009845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,10240,512,0.01891911029815674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,3072,0.013399110900031196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,3072,0.013100444442696042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,5120,0.012780444489585029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,4096,0.011686222420798408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,2560,0.011729777687125735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,2560,0.012040889097584618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,3584,0.011058666639857821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,3072,0.01035377797153261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,2048,0.010032888915803697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,2048,0.011254222028785281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,16384,0.06031911240683662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,2560,0.009304888546466827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,1536,0.021475555168257818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,1536,0.008473777936564552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,1536,0.010031999813185798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,1024,0.018753778603341844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,1024,0.006909333169460297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,1024,0.007954667011896769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,10240,0.04375999834802416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,2048,0.007732444339328342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,768,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,768,0.007640889121426477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,8192,0.0388008885913425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,1536,0.00692622239391009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,512,0.017217778497272067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,512,0.004512000001139111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,512,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,256,0.016774222254753113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,1024,0.006648889018429651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,256,0.0038906666967603895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,256,0.0069120000633928515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,128,0.015452444553375244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,128,0.003626666549179289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,128,0.006755555669466655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,768,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,64,0.0038746665749284956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,64,0.006723555425802867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,8192,32,0.0038888889054457345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,8192,32,0.006994666324721442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,65536,0.17610843976338705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,512,0.005954666684071223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,65536,0.1939635541703966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,256,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,16384,0.05816000037723117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,65536,0.13940710491604275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,8192,128,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,16384,0.05060088965627882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,16384,0.038374221987194486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,12288,0.04748355680041843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,12288,0.03925333420435587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,12288,0.030563556485705908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,3072,0.026297777891159058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,10240,0.04251377781232198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,65536,0.07896711428960164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,10240,0.032446222172843084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,10240,0.02674399978584713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,8192,0.03801600138346354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,16384,0.025399999486075506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,8192,0.027681777874628704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,8192,0.023415999280081853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,7168,0.0348328881793552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,7168,0.02425777746571435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,12288,0.020637333393096924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,7168,0.02102577851878272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,6144,0.0211182220114602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,6144,0.01908177799648709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,2560,0.024783111280865137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,5120,0.029959999852710303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,5120,0.018458666072951425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,5120,0.01735466718673706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,2048,0.023810666468408372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,4096,0.02780533168050978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,4096,0.016030222177505493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,4096,0.015099555253982544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,10240,0.018007111218240526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,3584,0.026483555634816486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,3584,0.013873777455753751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,3584,0.013948443863126965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,8192,0.015929778416951496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,7168,0.014471999473041959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,3072,0.012400888734393649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,3072,0.012999999854299756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,6144,0.013735999663670858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,5120,0.012672888735930124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,2560,0.011112000379297467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,2560,0.012089778151777057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,4096,0.011415999796655444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,2048,0.022673777408070032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,2048,0.009702222214804756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,2048,0.010613333019945355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,3584,0.010497777826256223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,8192,768,0.017498667041460674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,1536,0.008351999852392409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,1536,0.009304888546466827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,3072,0.009693333672152625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,2560,0.00817599975400501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,1024,0.006286222073766921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,1024,0.007984889050324758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,2048,0.007600888609886169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,768,0.01723377737734053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,768,0.005544888890451855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,768,0.007271111011505127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,1536,0.0069155556460221606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,1024,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,512,0.00423555572827657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,512,0.007341333561473423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,768,0.006227555374304454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,256,0.01683644453684489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,256,0.0038542221817705366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,256,0.006911110960774952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,512,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,256,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,128,0.003855111284388436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,128,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,6144,0.03274044394493103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,64,0.0032231110251612137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,64,0.006684444430801604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,7168,32,0.003559999995761447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,7168,32,0.006861333217885759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,65536,0.14408532778422037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,65536,0.19144977463616264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,65536,0.13349778122372097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,7168,128,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,2560,0.023634667197863262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,3072,0.024988444315062627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,16384,0.04962044292026096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,16384,0.03728533453411526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,65536,0.07624266544977824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,12288,0.036862221029069685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,12288,0.030004445049497817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,16384,0.024664888779322307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,1536,0.01982311076588101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,1024,0.017851556340853374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,10240,0.031088001198238794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,10240,0.02624799973434872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,8192,0.03309155503908793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,8192,0.025983111725913152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,8192,0.02238044473859999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,12288,0.020000888241661918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,7168,0.030710223648283217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,7168,0.02247200078434414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,7168,0.020650666620996263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,512,0.017140444782045152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,6144,0.029215110672844782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,6144,0.01983644399378035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,6144,0.018887110882335238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,10240,0.01759022143152025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,5120,0.027087999714745417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,5120,0.017309332887331646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,5120,0.016960889101028442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,8192,0.015442666080262927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,7168,0.01418133411142561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,4096,0.025105777713987563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,4096,0.01443199978934394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,4096,0.01482399966981676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,6144,0.013183111117945777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,3584,0.024358222881952923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,3584,0.012768000364303589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,3584,0.013628444737858243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,3072,0.023003555006451074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,3072,0.011753777662913004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,5120,0.012087999946541257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,3072,0.012821333275900947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,4096,0.010986666712496014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,2560,0.010188444621033138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,2560,0.012015110916561551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,3584,0.00978311151266098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,3072,0.008577778107590145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,7168,128,0.01514400045077006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,2048,0.009415110780133141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,2048,0.010442666709423065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,1536,0.01818933255142636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,1536,0.008001777860853408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,1536,0.008925333619117737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,2560,0.00796622203456031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,1024,0.01681600014368693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,1024,0.006299555715587404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,1024,0.007671111159854465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,2048,0.007256000406212277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,1536,0.0069751110341813825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,768,0.0053431110249625305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,768,0.007326222128338284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,16384,0.04946400059594048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,512,0.016143111719025504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,512,0.004200889004601372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,512,0.007296000089910295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,1024,0.006253333141406377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,12288,0.040735999743143715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,256,0.003896000070704354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,256,0.006955555743641323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,768,0.006263111200597551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,512,0.006009777800904379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,10240,0.03679733475049337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,128,0.0034933334423436057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,128,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,64,0.0031662223239739737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,64,0.006903999795516332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,6144,32,0.0032151111712058387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,6144,32,0.006594666590293248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,65536,0.13993599679734972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,256,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,65536,0.1569466723336114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,65536,0.12408888339996338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,6144,128,0.0058871110280354815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,16384,0.04707733458942837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,16384,0.036308444208569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,65536,0.07460533248053657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,12288,0.03986400034692552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,12288,0.037109331952200994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,12288,0.028905779123306274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,16384,0.02370133333735996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,12288,0.019367999500698514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,10240,0.031234665049446955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,10240,0.02551199992497762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,2560,0.022314666046036616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,8192,0.03218222326702542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,8192,0.02570488883389367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,8192,0.02165777815712823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,2048,0.019633778267436557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,7168,0.029948443174362183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,7168,0.02295200030008952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,7168,0.02016088863213857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,10240,0.017086222767829895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,6144,0.028071112102932397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,6144,0.020439111524158053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,6144,0.018215111560291715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,8192,0.015026667051845126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,5120,0.025905778010686237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,5120,0.017076444294717576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,5120,0.016705777910020616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,7168,0.013763555222087435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,4096,0.024715556038750544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,4096,0.014225777652528552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,4096,0.014485332700941296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,768,0.016879111528396606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,6144,0.012739555703269111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,3584,0.023112000690566167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,3584,0.013059555656380124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,3584,0.013380444712109037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,3072,0.021549334128697712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,3072,0.011884444289737277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,3072,0.012322666744391123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,5120,0.011743110915025076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,2560,0.020617778102556866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,2560,0.010033778018421596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,2560,0.011422221859296164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,4096,0.010048000348938836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,256,0.016864889197879367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,2048,0.0185253338681327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,2048,0.008960888617568547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,2048,0.009639999932712978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,3584,0.00906399968597624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,1536,0.01717866626050737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,1536,0.007939555578761632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,1536,0.008612444831265343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,3072,0.008620444271299574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,1024,0.015796444482273526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,1024,0.005903111149867375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,1024,0.007618666523032718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,2560,0.007654222349325816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,768,0.016862221890025668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,768,0.004587555511130227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,768,0.007490666376219855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,6144,128,0.014811555544535318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,2048,0.007240888973077138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,512,0.00423200014564726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,512,0.00723555518521203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,1536,0.006755555669466655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,256,0.003882666842805015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,256,0.006760889043410619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,1024,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,128,0.01482399966981676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,128,0.0035048888789282907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,128,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,768,0.005980444451173146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,64,0.0032239999208185407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,64,0.006263999889294307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,5120,32,0.003240888938307762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,5120,32,0.006782222125265334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,16384,0.04743822084532844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,512,0.005879999862776862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,65536,0.12100088596343994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,65536,0.13186666700575086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,65536,0.08385688728756374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,16384,0.0362702210744222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,16384,0.025058666865030926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,256,0.005913777897755305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,5120,128,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,12288,0.031165334913465712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,12288,0.02087733315096961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,65536,0.07155733638339572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,10240,0.03340177734692892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,10240,0.026710222164789837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,16384,0.02286400066481696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,10240,0.018619555566045973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,10240,0.03594400154219733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,8192,0.030436442957984075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,8192,0.02141333288616604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,8192,0.016542222764756944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,7168,0.027976890405019123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,7168,0.019824888971116807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,12288,0.018763555420769584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,7168,0.015082667271296183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,10240,0.016528889536857605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,6144,0.016495111915800307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,6144,0.014166222678290473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,8192,0.014478221535682678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,5120,0.025008888708220586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,5120,0.01443377799457974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,7168,0.013269333375824822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,5120,0.01316177762216992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,4096,0.023023110296991136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,4096,0.01293511109219657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,4096,0.011393778026103973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,6144,0.012304888831244575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,3584,0.02219911085234748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,3584,0.009665778113736046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,3584,0.010686222049925061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,5120,0.010446222292052375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,3072,0.0196426659822464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,3072,0.0087333329849773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,3072,0.009687110781669617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,4096,0.009653333160612319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,2560,0.018554665976100497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,2560,0.00794400026400884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,2560,0.008970666262838576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,3584,0.008658666577604082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,3072,0.00794222205877304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,2048,0.006703111032644908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,2048,0.007867555651399825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,2560,0.007446222007274628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,1536,0.01614755557643043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,1536,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,1536,0.00692622239391009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,2048,0.006959999601046245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,1536,0.006590222318967183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,1024,0.0046693334976832075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,1024,0.006642666541867786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,256,0.016492444607946608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,1024,0.005956444475385878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,768,0.016798221402698092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,5120,512,0.015498666299713982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,768,0.004210666649871402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,768,0.00629688882165485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,512,0.014785778191354541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,512,0.0038737778862317405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,512,0.006239999913507038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,768,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,256,0.0035582222044467926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,256,0.005988444305128521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,512,0.00573955559068256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,12288,0.037575999895731606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,128,0.014468444718254937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,128,0.0032115555885765287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,128,0.005906666732496685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,16384,0.04515199859937032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,64,0.0033084443873829315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,64,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,4096,32,0.0033004445334275565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,4096,32,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,256,0.00554577757914861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,65536,0.12908977932400173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,65536,0.07827822367350261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,4096,128,0.005600000007285013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,16384,0.04439999990993076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,16384,0.040185779333114624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,16384,0.02424088948302799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,12288,0.03727644350793626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,12288,0.02818044357829624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,12288,0.020226667324701946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,6144,0.026447110705905493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,65536,0.07041866911782159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,10240,0.03322488731808133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,10240,0.023976888921525743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,10240,0.018197332819302876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,16384,0.022579555710156757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,8192,0.029888888200124104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,8192,0.020433778564135235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,8192,0.01589866644806332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,7168,0.027624888552559748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,7168,0.018552889426549275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,12288,0.018449778358141582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,7168,0.014918223023414612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,10240,0.016340444485346477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,6144,0.016143111719025504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,6144,0.013564444250530668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,8192,0.01441333360142178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,7168,0.01309511148267322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,5120,0.01459111107720269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,5120,0.01277955538696713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,6144,0.011687111523416309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,2048,0.01687733332316081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,4096,0.012352888782819113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,4096,0.011125333607196808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,5120,0.010476444330480365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,3584,0.020908445119857788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,3584,0.012720888687504662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,3584,0.0103511114915212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,4096,0.008948444492287105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,3584,0.00830666638082928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,3072,0.008647999829716152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,1024,0.015167999598715039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,3072,0.009352888498041365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,3072,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,2560,0.007592000067234039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,2560,0.008283555507659912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,2560,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,4096,256,0.015507555670208402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,2048,0.016833777228991192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,2048,0.006577777779764599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,2048,0.007594666547245449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,1536,0.015456888410780164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,1536,0.005551110953092575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,1536,0.0069395556218094295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,2048,0.0069919998447100324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,1024,0.01513155632548862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,1024,0.004568888909286923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,1024,0.006583111153708563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,1536,0.006578666468461354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,768,0.016599110431141324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,768,0.004061333421203825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,768,0.006305777778228124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,1024,0.005947555518812603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,512,0.014375999569892883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,512,0.003633777714437909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,512,0.006258666515350342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,768,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,65536,0.13111733065711126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,256,0.015169777803950839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,256,0.003547555456558863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,256,0.006193777753247156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,128,0.01477777792347802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,128,0.0031644445326593188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,128,0.005938666562239329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,512,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,64,0.0032151111712058387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,64,0.005912000106440649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,256,0.005618666609128316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3584,32,0.003167111012670729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3584,32,0.00591555568906996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3584,128,0.00553688903649648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,65536,0.10275377829869588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,65536,0.12928533554077148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,16384,0.04371466570430332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,65536,0.07488711012734307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,16384,0.03048355711830987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,16384,0.023590222001075745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,6144,0.02606222199069129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,12288,0.02202933364444309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,65536,0.06947288910547893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,12288,0.0196435550848643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,16384,0.02204533252451155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,10240,0.03281866510709127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,10240,0.018238221605618794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,10240,0.017858665850427415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,5120,0.02437422176202138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,4096,0.022299556268586054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,12288,0.018106665876176622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,8192,0.015321777926550971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,8192,0.01566755606068505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,7168,0.027095999982621934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,7168,0.01442311041884952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,7168,0.014464888307783337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,10240,0.015981333123313058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,8192,0.01383733335468504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,6144,0.012233778006500669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,6144,0.013439111411571503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,5120,0.023984889189402264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,3072,0.018931556079122756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,7168,0.012307555311255984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,5120,0.011063999599880643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,5120,0.012415111064910889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,4096,0.00996977753109402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,4096,0.010660444696744284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,6144,0.011075555450386472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3584,2560,0.01721955504682329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,3584,0.019160888261265226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,3584,0.00923111124171151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,3584,0.009682666924264696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,5120,0.009777777724795872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,3072,0.008665777742862701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,3072,0.0177875558535258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,3072,0.008998221821255153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,4096,0.008868444297048781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,2560,0.00758133331934611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,2560,0.00794933322403166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,3584,0.00831733312871721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,2048,0.015843555331230164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,3072,0.00793511089351442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,2048,0.0068986668354935115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,2048,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,2560,0.007626666790909237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,1536,0.005510222166776657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,1536,0.0069573331210348345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,2048,0.0069120000633928515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,1024,0.014528888795110913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,1024,0.004536889079544279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,1024,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,1536,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,768,0.016541333662139047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,768,0.0041680000722408295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,768,0.006215999937719769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,1024,0.006206222292449739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,768,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,512,0.003871110992299186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,512,0.005923555543025334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,512,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,256,0.015152888165579902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,256,0.0035128887328836652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,256,0.006232000059551663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,256,0.005531555662552516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,12288,0.03671377897262573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,128,0.0031582222630580268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,128,0.005882666756709416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,3072,128,0.00553955551650789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,64,0.0028728888266616394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,64,0.005924444645643234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,3072,32,0.0028204443968004654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,3072,32,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,65536,0.1277137729856703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,65536,0.08031111293368869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,65536,0.0718453327814738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,8192,0.029439111550649006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,6144,0.025624000363879736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,16384,0.029735111527972754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,16384,0.02300888962215847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,65536,0.06765421893861559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,12288,0.035775999228159584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,12288,0.02420177724626329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,12288,0.019142222073343065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,16384,0.021706667211320665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,4096,0.0217866665787167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,10240,0.01678222252262963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,10240,0.017245334055688646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,12288,0.017792888813548617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,8192,0.028594666057162817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,8192,0.014829332629839579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,8192,0.015188443991872998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,2560,0.016910221841600206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,10240,0.01586844523747762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,7168,0.01309511148267322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,7168,0.014131555954615274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,1536,0.015461333923869662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,6144,0.02530577778816223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,6144,0.011744889120260874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,6144,0.012994666894276937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,8192,0.013400889105266996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,7168,0.011883555187119378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,5120,0.01074222226937612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,5120,0.012347555822796293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,6144,0.010278222461541494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,5120,0.009354666703277165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,4096,0.02000177734427982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,512,0.013789333403110504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,4096,0.009358222285906473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,4096,0.010022222167915767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,3584,0.0180711117055681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,3584,0.008602666358153025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,3584,0.00906044410334693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,4096,0.008635555704434713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,3072,0.008014221986134848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,3584,0.008295999632941352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,3072,0.008559111091825698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,3072,0.007722666694058313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,2560,0.007313778003056844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,2560,0.007778666913509369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,3072,128,0.015095111396577625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,2560,0.007312888900438945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,2048,0.01613333324591319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,2048,0.006226666685607698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,2048,0.007242666350470648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,1536,0.01514488955338796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,1536,0.004870222260554631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,1536,0.0069422221018208405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,2048,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,1536,0.006622222148709827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,1024,0.004212444441186057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,1024,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,1024,0.006212444355090459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,768,0.01683555543422699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,768,0.003987555702527364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,768,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,768,0.005923555543025334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,512,0.013759999639458127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,512,0.0038319999972979226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,512,0.00599911105301645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,16384,0.04304000073009067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,256,0.014534221755133735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,256,0.003559111307064692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,256,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,512,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,128,0.014928888943460254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,128,0.0032035555276605818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,10240,0.03222844335767958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,128,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,64,0.00314666661951277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,64,0.0061048890153567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2560,32,0.0031999999450312722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2560,32,0.005879111174080107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,256,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,65536,0.06350666946834989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,65536,0.0693697796927558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2560,128,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,65536,0.06651200188530816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,16384,0.023657777243190344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,7168,0.026684444811609056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,16384,0.022263111339675054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,5120,0.02275111112329695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,12288,0.031756444109810725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,12288,0.019328888919618394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,12288,0.018611555298169453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,16384,0.02142133315404256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,10240,0.017609778377744887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,12288,0.017529777354664274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,10240,0.01683555543422699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,8192,0.025556445121765137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,8192,0.014207111464606391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,8192,0.014832889040311178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,10240,0.01498044364982181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,7168,0.02401599950260586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,7168,0.012823999755912356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,7168,0.013760888742076026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,8192,0.0124177775449223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,3072,0.017208000024159748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,6144,0.011619555453459421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,6144,0.012801777985360889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,7168,0.011112000379297467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,5120,0.019663110375404358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,5120,0.010383110907342698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,6144,0.009815110928482479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,5120,0.011390222443474663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,4096,0.018186666899257235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,4096,0.008980444735950893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,4096,0.009311111436949836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,5120,0.00927111092540953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,2560,0.01686844395266639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,3584,0.0173999998304579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,3584,0.008371555142932469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,3584,0.008660444782839881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,3072,0.01683288812637329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,4096,0.008613333106040955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,3072,0.00758844448460473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,3072,0.008278222547637092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,3584,0.008086221913496653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,2560,0.006468444648716185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,2560,0.007989333735571967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,3072,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,2048,0.015227556228637695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,2048,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,2048,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,2560,0.007329777710967594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2560,1024,0.01407911048995124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,2048,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,1536,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,1536,0.006960888703664144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,1536,0.0064586665895250105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,1024,0.00425155543618732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,1024,0.006792888873153263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,1024,0.006044444524579578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,768,0.015864888827006023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,768,0.004040889028045866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,768,0.00600888869828648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,768,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,512,0.013447110851605734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,512,0.003903999924659729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,512,0.005956444475385878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,512,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,65536,0.10432800319459702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,256,0.0032400000426504347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,256,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,16384,0.03705777724583944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,128,0.014783110883500842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,256,0.00553688903649648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,128,0.0032088889016045462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,128,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,64,0.002879111096262932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,64,0.005919111271699269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,2048,32,0.003160888950030009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,2048,32,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,2048,128,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,65536,0.055049777030944824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,65536,0.0674720009167989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,10240,0.028391109572516546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,16384,0.036327110396491155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,16384,0.019377777973810833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,65536,0.06522488594055176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,16384,0.021760889225535925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,16384,0.021024000313546922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,12288,0.016237333416938782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,12288,0.018728888697094388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,12288,0.01682577696111467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,6144,0.021624000536070928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,10240,0.028149333265092637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,10240,0.015477332803938123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,10240,0.01641688909795549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,8192,0.024672889047198828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,8192,0.01408266690042284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,8192,0.014412444498803882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,10240,0.013631111217869652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,7168,0.02238133384121789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,7168,0.01276711126168569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,7168,0.013091555900043912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,8192,0.011759999725553723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,6144,0.020624889267815482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,6144,0.011251555548773872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,6144,0.01180266671710544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,2560,0.01607022186120351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,7168,0.010499555203649731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,5120,0.019263111882739596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,5120,0.010658666491508484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,5120,0.01032266683048672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,4096,0.01756000022093455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,4096,0.008966222405433655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,6144,0.009660444325870937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,4096,0.008967999782827165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,3584,0.016753777861595154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,3584,0.008344000412358178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,5120,0.009091555244392818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,3584,0.008346666892369589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,3072,0.016189333465364244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,3072,0.007249777515729268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,4096,0.008630221916569604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,3072,0.00793599999613232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,2560,0.015798222687509324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,3584,0.008015111088752747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,2560,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,2560,0.007684444387753804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,1536,0.014136888914638095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,2048,0.014807111687130399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,2048,0.00554044461912579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,2048,0.007306666837798224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,3072,0.007648888561460707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,1536,0.014144000079896716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,1536,0.005048000150256687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,1536,0.0069066666894488865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,1024,0.013439999686347114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,2560,0.007250666618347168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,1024,0.004233777936961916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,1024,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,2048,0.006959111326270633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,768,0.0038355555799272326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,768,0.005954666684071223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,1536,0.0063866666621632045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,1024,0.014078223043017916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,512,0.013139555851618448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,512,0.00350222239891688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,1024,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,512,0.005991999887757831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,256,0.003144888828198115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,256,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,768,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,512,0.005567111074924469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,256,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,128,0.0031537777847713898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,128,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1536,128,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,64,0.0028320000403457214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,64,0.005705777969625261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1536,32,0.002814222127199173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1536,32,0.0058702222175068325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,65536,0.09540888998243545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,65536,0.04047466648949517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,65536,0.0653182201915317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,16384,0.034903112385008074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,16384,0.015615110596021017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,16384,0.021367111139827307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,65536,0.06451644500096639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,12288,0.029534220695495605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,12288,0.013910222384664746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,12288,0.017892445127169292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,16384,0.02017333275742001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,12288,0.014864888456132678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,10240,0.012768000364303589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,10240,0.015944888194402058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,2048,256,0.015262222952312894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,10240,0.012720888687504662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,8192,0.01167377746767468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,8192,0.012867555850081973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,8192,0.0114702218108707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,65536,0.10272800260119969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,7168,0.011101333631409539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,7168,0.02095199955834283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,7168,0.011791110866599612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,6144,0.01990133358372582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,6144,0.009992000129487779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,6144,0.010644444160991244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,7168,0.010339555641015371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,5120,0.01888088881969452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,5120,0.009530666801664565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,5120,0.009680888719028896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,6144,0.009644444617960189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,5120,0.008971555365456475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,4096,0.008280889027648503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,4096,0.008667555948098501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,4096,0.00832622249921163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,12288,0.031034668286641438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,3584,0.007316444483068254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,3584,0.008288000192907121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,3584,0.008016000191370646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,768,0.015171556009186639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,3072,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,3072,0.008010666403505538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,3072,0.007609777980380588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,2560,0.015389333168665567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,128,0.014787554740905762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,2560,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,2560,0.0075839997993575195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,2048,0.01481599940194024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1536,256,0.01481066644191742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,2048,0.005221333354711533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,2048,0.006983111302057902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,1536,0.014109333356221518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,2560,0.007194666398896112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,1536,0.00462311092350218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,1536,0.006904888898134232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,1024,0.013120888835854001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,2048,0.006969778074158563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,1024,0.0042364444169733255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,1024,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,1536,0.0063164445261160536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,768,0.0038808890514903595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,768,0.006221333311663733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,1024,0.00592533333433999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,768,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,512,0.003546666767862108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,512,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,512,0.0053182223604785065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,10240,0.02567111121283637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,256,0.0031786666562159858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,256,0.005674666828579373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,256,0.005387555393907759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,8192,0.023016888234350417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,128,0.0029440000653266907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,128,0.005630222045713001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,1024,128,0.005371555685997009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,64,0.0028524444335036804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,64,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,1024,32,0.0029226665695508323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,4096,0.017383999294704862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,1024,32,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,65536,0.031839112440745033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,65536,0.09498933288786147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,65536,0.06462399827109443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,16384,0.035016000270843506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,16384,0.013474666410022311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,65536,0.06393066379759046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,16384,0.021140444609853957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,12288,0.02828977836502923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,12288,0.011648000114493899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,12288,0.017877333694034152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,3584,0.01647644407219357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,16384,0.018509333332379658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,12288,0.014544889330863953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,10240,0.013445333474212222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,10240,0.015508444772826301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,10240,0.012694222231705984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,8192,0.02231377859910329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,8192,0.012258666257063547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,8192,0.011688888900809817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,8192,0.011404444773991903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,7168,0.020968000094095867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,7168,0.011669333610269757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,7168,0.011601777540312873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,7168,0.010318222145239512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,6144,0.01055111073785358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,3072,0.01605333387851715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,6144,0.01016355554262797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,5120,0.018551111221313477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,5120,0.009317333499590555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,5120,0.009648000200589498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,6144,0.009703111317422655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,5120,0.009009777671760982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,4096,0.00832711077398724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,4096,0.008527999950779809
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,4096,0.008323555191357931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,3584,0.01647200021478865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,3584,0.00796888851457172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,3584,0.008264000217119852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,3584,0.00793422261873881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,768,0.015785778562227886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,3072,0.0063031112982167145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,3072,0.007987555530336168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,3072,0.007659555309348636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,512,0.012720888687504662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,2560,0.01511111193233066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,2560,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,2560,0.007636444436179266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,2560,0.007105777661005656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,2048,0.005271111097600725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,2048,0.006970666348934174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,2048,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,256,0.01447111037042406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,1536,0.004905777672926585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,1536,0.006616000086069107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,1536,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,1024,128,0.01460622251033783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,1024,0.0041991112132867174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,1024,0.006597333484225803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,1024,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,768,0.015804444750150044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,768,0.003556444413132138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,768,0.006248888870080312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,768,0.0058337777025169795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,512,0.013178666432698568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,512,0.003601777884695265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,512,0.00591555568906996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,512,0.00554933316177792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,10240,0.024695111645592585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,256,0.0031999999450312722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,256,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,256,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,6144,0.019905777441130746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,128,0.0028568889117903183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,128,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,768,128,0.005559999909665849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,64,0.0028924445311228433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,64,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,768,32,0.0026577777332729767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,768,32,0.005959110955397288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,4096,0.017197334104114108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,65536,0.025851555996470984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,65536,0.06400089131461249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,3072,0.015846222639083862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,65536,0.06421866681840685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,16384,0.013758222262064615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,16384,0.020247111717859905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,16384,0.017840888765123155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,12288,0.026362667481104534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,12288,0.012418666647540199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,12288,0.01591999994383918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,12288,0.014188443620999655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,10240,0.023687111006842718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,10240,0.011258666714032492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,10240,0.013791999883121915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,10240,0.012597333225939008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,8192,0.021550221575631037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,8192,0.010036444498433007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,8192,0.011759999725553723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,8192,0.011159111228254108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,7168,0.020259555843141343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,7168,0.009339555270142024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,7168,0.011031111081441244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,7168,0.010382222632567087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,6144,0.018926221463415358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,6144,0.008015111088752747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,6144,0.010327999790509542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,2048,0.01483022173245748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,5120,0.01795111099878947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,5120,0.007648888561460707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,6144,0.00943288869327969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,5120,0.009934222532643212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,5120,0.008982222113344405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,4096,0.016493333710564505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,4096,0.007666666474607255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,4096,0.008640888664457533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,1536,0.013773333695199756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,4096,0.008615111311276754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,3584,0.015792888071801927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,3584,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,3584,0.008043555749787224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,3584,0.008294222255547842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,3072,0.0064195555945237475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,3072,0.007998222278224098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,3072,0.007730666961934831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,2560,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,2560,0.007313778003056844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,1024,0.013109332985348172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,2560,0.007290666302045186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,2048,0.0052800000541739995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,2048,0.014131555954615274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,2048,0.0069297779765393995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,1536,0.013095999757448832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,1536,0.004576888763242298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,2048,0.00664533343580034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,1536,0.00656088896923595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,1024,0.01277511070171992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,1024,0.0038986665507157645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,1536,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,1024,0.006286222073766921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,1024,0.0059164443777667145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,768,0.003568888952334722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,768,0.006251555350091722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,768,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,512,0.012100444071822695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,512,0.003488000068399641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,512,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,512,0.005429333282841577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,256,0.015114666687117683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,768,128,0.014122666584120857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,256,0.003169777699642711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,256,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,128,0.01308444473478529
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,128,0.0028488888508743713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,256,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,128,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,64,0.0025111111915773815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,64,0.005923555543025334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,512,32,0.0027964444210131965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,512,128,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,512,32,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,65536,0.021007110675175984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,65536,0.09310133589638604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,65536,0.06326399909125434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,65536,0.06288888719346789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,16384,0.010026666853162978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,16384,0.01852799952030182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,16384,0.01716977854569753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,16384,0.032560000816980995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,12288,0.008998221821255153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,12288,0.01442311041884952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,12288,0.014110222458839417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,65536,0.09381066428290473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,10240,0.009368889033794403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,10240,0.013222222526868185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,10240,0.01255555533700519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,8192,0.020958221620983545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,8192,0.00846400029129452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,8192,0.011689778003427716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,8192,0.011395555403497485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,7168,0.019705777366956074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,7168,0.007995555798212687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,7168,0.010743111371994019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,7168,0.010392888552612728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,3072,0.015121777852376303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,6144,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,6144,0.01017155581050449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,2560,0.014817777607176038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,6144,0.009347555538018545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,5120,0.008277333445019191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,5120,0.009353777600659264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,5120,0.008959111240175035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,768,0.014472888575659858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,4096,0.007317333585686154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,4096,0.008332444561852349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,4096,0.008314666648705801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,512,256,0.014133334159851074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,3584,0.007019555403126611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,3584,0.007968000239796108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,16384,0.03065955638885498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,3584,0.007970666719807519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,3072,0.0069644442862934535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,3072,0.00757244477669398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,3072,0.0074782222509384155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,2560,0.014119111829333834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,2560,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,2560,0.007296889192528195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,12288,0.02587644424703386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,2560,0.006968888971540663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,2048,0.004929777648713854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,2048,0.006646222124497096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,10240,0.023279110590616863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,2048,0.006787555499209299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,1536,0.004559111264016894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,1536,0.006285333385070165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,1536,0.006298666612969504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,6144,0.018580444984965853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,1024,0.00398577791121271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,1024,0.006559999866618051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,1024,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,768,0.014567111929257711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,768,0.0036684444381131064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,768,0.00629688882165485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,768,0.0057555557125144536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,5120,0.017497777938842773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,512,0.0031662223239739737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,512,0.005927111125654644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,512,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,256,0.012796444197495779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,256,0.0033031110134389666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,256,0.005969777703285217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,4096,0.015806222955385845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,128,0.011015111373530494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,256,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,128,0.0026622222115596137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,128,0.005884444548024072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,64,0.0028151110228565005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,256,128,0.005205333232879639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,64,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,256,32,0.0028355556229750314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,256,32,0.005919999960396025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,65536,0.015692444311247934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,65536,0.09240888886981541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,65536,0.06292000081804064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,65536,0.0613466633690728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,16384,0.009674666656388177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,16384,0.017160000072585214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,16384,0.017264000243610807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,3072,0.01514400045077006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,3584,0.015275556180212231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,12288,0.00830577810605367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,12288,0.014148443937301636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,10240,0.022271111607551575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,12288,0.014110222458839417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,10240,0.009082666701740688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,10240,0.013056000073750814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,10240,0.012445333103338877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,8192,0.020845333735148113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,8192,0.008344888687133789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,8192,0.011352889239788055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,7168,0.01959466603067186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,8192,0.01108000013563368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,7168,0.007658667034573025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,7168,0.01072088877360026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,7168,0.010371555884679159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,6144,0.01757955551147461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,6144,0.00739822205570009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,6144,0.010064000056849586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,6144,0.009372444616423713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,5120,0.016528889536857605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,5120,0.007107555866241455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,5120,0.009359111388524374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,5120,0.008995555341243744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,4096,0.006625777731339137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,4096,0.008607111043400234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,2048,0.013864888913101621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,4096,0.008600888980759515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,3584,0.006605333338181178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,3584,0.007978666987684038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,3584,0.008042666647169325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,3072,0.014485332700941296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,3072,0.006274666637182236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,3072,0.00794577764140235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,3072,0.007635555333561367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,2560,0.013818666338920593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,2560,0.006555555595291986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,2560,0.007626666790909237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,2560,0.007304888632562425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,1536,0.01242844429281023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,2048,0.00630666646692488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,2048,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,2048,0.006736889067623351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,1536,0.012226666841242047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,1536,0.0042088888585567474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,1536,0.006559999866618051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,1536,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,1024,0.011400000088744693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,1024,0.0038568890757030914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,1024,0.006225777582989798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,1024,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,1024,0.011821333732869891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,768,0.003549333247873518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,768,0.006260444306664997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,768,0.005577777822812398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,512,0.010737777584128909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,512,0.0032391111469931076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,512,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,512,0.005409777578380372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,256,0.011047999891969891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,256,0.003137777869900068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,256,0.005589333259397083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,256,512,0.012072888513406118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,256,0.005445333404673471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,128,0.0028773333049482773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,128,0.0107004443804423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,128,0.0059075554211934405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,64,0.002519999941190084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,192,128,128,0.0052355556852287715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,64,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,128,32,0.0028320000403457214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,128,32,0.005948444621430502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,65536,0.015088889333936902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,65536,0.061230222384134926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,16384,0.009008888569143083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,16384,0.017218665944205392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,12288,0.00796888851457172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,12288,0.014669333895047506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,10240,0.007287999822033777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,10240,0.01312800000111262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,8192,0.006970666348934174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,8192,0.011411555939250521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,7168,0.006602666858169768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,7168,0.011054221954610614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,6144,0.006609777609507243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,6144,0.010038221875826517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,5120,0.006000888844331105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,5120,0.009724444813198512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,4096,0.005912888795137405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,4096,0.008610666626029545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,3584,0.00646577775478363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,3584,0.007985778152942657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,3072,0.006295111030340195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,3072,0.008005333443482717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,2560,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,2560,0.007319110963079665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,2048,0.006264888991912206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,2048,0.0069448889957533936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,1536,0.004572444491916233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,1536,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,1024,0.0038337777886125776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,1024,0.006578666468461354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,768,0.0035439998739295532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,768,0.006055999961164262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,512,0.0032213332338465583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,512,0.005950222412745158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,256,0.002847111059559716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,256,0.0059262220230367445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,128,0.0028444443725877335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,128,0.005913777897755305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,64,0.0028497777465316984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,64,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,64,32,0.00254844439526399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,64,32,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,16384,0.030347555875778198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,65536,0.01463733282354143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,16384,0.008627555436558193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,65536,0.06032088730070326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,16384,0.017180444465743173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,12288,0.007827555967701806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,12288,0.014473777678277759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,10240,0.007276444799370236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,10240,0.013105777402718862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,8192,0.006976889239417181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,8192,0.011257777611414591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,7168,0.0067528887755341
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,7168,0.011046222514576383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,6144,0.006333333336644703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,6144,0.010031999813185798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,5120,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,5120,0.009356444080670675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,4096,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,4096,0.008349333372380998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,3584,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,3584,0.008307555483447181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,3072,0.006280888699822956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,12288,0.02507466740078396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,3072,0.007607999775144789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,2560,0.006284444282452266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,2560,0.007301333049933116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,2048,0.006053333481152852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,2048,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,1536,0.004239999999602635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,1536,0.006802666518423293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,1024,0.003846222327815162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,1024,0.005940444353553984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,768,0.003559111307064692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,768,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,512,0.0031902222997612427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,512,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,256,0.0029484445436133277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,256,0.005948444621430502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,128,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,64,0.005587555468082428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,128,0.005635555419656966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,64,0.002532444480392668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,192,32,32,0.00249777775671747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,192,32,32,0.005767999837795894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,16384,0.3751795556810167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,16384,0.29320089022318524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,16384,0.266848882039388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,12288,0.23788266711764863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,12288,0.2879386742909749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,12288,0.1962195502387153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,3584,0.015448000695970325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,4096,0.015513777732849121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,10240,0.20274221897125244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,10240,0.24408978886074492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,2048,0.013442666994200813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,10240,0.16896266407436797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,8192,0.19970044824812147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,8192,0.18144533369276258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,8192,0.13408977455563015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,7168,0.1740586625205146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,7168,0.1187102264828152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,192,128,768,0.01276533305644989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,6144,0.14992177486419678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,6144,0.10377599795659383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,7168,0.15449599424997965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,6144,0.14121154944101968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,5120,0.1258808904223972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,5120,0.08659555514653523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,16384,0.13816000355614558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,12288,0.10370755857891506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,4096,0.10333422157499526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,4096,0.0708177751964993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,10240,0.08773244751824273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,3584,0.1038373311360677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,3584,0.0907484425438775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,3584,0.06272444460127088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,8192,0.07162400086720784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,7168,0.06300710969501071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,3072,0.07799022065268622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,3072,0.054690665668911405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,6144,0.055050667789247304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,2560,0.08920088741514419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,2560,0.06491288873884413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,2560,0.0466595557000902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,5120,0.12463911374409993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,2048,0.08245155546400282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,2048,0.05293955405553182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,2048,0.038719111018710665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,4096,0.11201511489020453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,1536,0.07471822367774116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,1536,0.04086844457520379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,1536,0.030407110850016277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,3072,0.0964862240685357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,1024,0.06718666685952081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,1024,0.029776000314288672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,1024,0.02498666610982683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,5120,0.04666399955749512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,4096,0.03929155402713352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,768,0.022989332675933838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,768,0.023964444796244305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,3584,0.03478577733039856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,3072,0.031253334548738264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,512,0.016598222984208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,512,0.02126577827665541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,2560,0.027792887555228338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,256,0.02760444415940179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,256,0.010686222049925061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,256,0.019507555498017203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,2048,0.02422222163942125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,768,0.03405422303411696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,128,0.009358222285906473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,128,0.017616889543003507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,1536,0.02019999921321869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,64,0.008311111066076491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,64,0.016530666086408827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,65536,32,0.008653333617581261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,65536,32,0.01698666645420922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,65536,0.285349342558119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,65536,0.3981920083363851
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,512,0.05975111325581869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,65536,0.28037956025865346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,1024,0.01687022215790219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,16384,0.1036542190445794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,16384,0.10263911220762466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,16384,0.07191377878189087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,65536,0.14552977350023058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,12288,0.0787644452518887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,12288,0.05571733580695259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,768,0.015204444527626038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,10240,0.06365422407786052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,10240,0.06565599971347384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,65536,128,0.024718221690919664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,10240,0.0473653342988756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,16384,0.041983111037148364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,8192,0.05679555733998617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,8192,0.05412977933883667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,8192,0.03922399878501892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,512,0.014022222823566861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,7168,0.04761688907941183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,7168,0.03436266713672214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,12288,0.03280622098180983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,6144,0.0484426650736067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,6144,0.041285332706239485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,256,0.012817777693271637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,6144,0.030589332183202107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,10240,0.028120001157124836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,5120,0.03499199946721395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,5120,0.026484444737434387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,8192,0.024540444215138752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,4096,0.03968000080850389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,4096,0.02921600143114726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,12288,0.07232444153891669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,4096,0.022107554806603327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,65536,128,0.01254488858911726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,3584,0.03746666510899862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,3584,0.025789333714379206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,3584,0.020255999432669747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,7168,0.021751999855041504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,3072,0.022672888305452134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,3072,0.018567999203999836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,6144,0.019868445065286424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,5120,0.017863111363516915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,2560,0.019270223047998216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,2560,0.016768000192112394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,4096,0.016011555989583332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,7168,0.051817778084013194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,2048,0.01593244406912062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,3584,0.014717333846622042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,2048,0.015040889382362366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,1536,0.030058665408028498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,1536,0.013172444370057849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,1536,0.013033777475357056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,3072,0.013765333427323235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,2560,0.012873777912722694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,1024,0.01054488867521286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,1024,0.011745777395036487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,5120,0.043284446001052856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,2048,0.011696000066068439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,768,0.008952000074916417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,768,0.011067555182509951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,1536,0.01036000003417333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,512,0.0248346659872267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,512,0.007291555404663086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,512,0.009386666946940953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,3072,0.035826666487587824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,256,0.018239999810854595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,256,0.004883555488453971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,256,0.008625778059164682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,1024,0.008632000121805403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,2560,0.03379999929004245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,128,0.004419555680619346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,128,0.008254222571849823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,768,0.007617777420414819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,64,0.0038382220599386427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,64,0.008455110920800103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,16384,32,0.003850666599141227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,16384,32,0.008281778130266402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,512,0.00700711127784517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,65536,0.2941519949171278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,65536,0.21076622274186876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,2048,0.03238755464553833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,256,0.0069324444565508105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,16384,0.07522400220235188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,16384,0.056141330136193164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,1024,0.028259555498758953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,12288,0.06459289126925997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,12288,0.05861155854331123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,16384,128,0.007041778001520369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,12288,0.04322133461634318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,768,0.022328888376553852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,10240,0.04899644520547655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,10240,0.03696711195839776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,65536,0.13298043939802381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,8192,0.05047733253902859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,8192,0.041327999697791204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,16384,0.03917510973082648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,8192,0.031025777260462444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,7168,0.04574222366015116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,7168,0.03581600056754218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,7168,0.027540443672074214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,12288,0.03118488854832119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,6144,0.04230666822857327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,6144,0.03086400032043457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,6144,0.024824889169798955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,16384,128,0.016488000750541687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,10240,0.02641333308484819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,5120,0.03834933373663161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,5120,0.026846221751636926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,5120,0.02196888956758711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,8192,0.022913777165942725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,4096,0.023048889305856492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,4096,0.018353778454992507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,7168,0.02053244411945343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,3584,0.03352800011634827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,3584,0.0196115556690428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,3584,0.01720355616675483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,6144,0.018896889355447557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,3072,0.031903111272388034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,3072,0.01754844519827101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,3072,0.015585778488053216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,65536,0.24643201298183867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,5120,0.01681155628628201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,2560,0.014940443966123792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,2560,0.01425333321094513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,4096,0.015244444211324057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,2048,0.028891556792789038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,2048,0.012747555143303342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,2048,0.012860444684823355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,16384,0.07873244418038262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,3584,0.013995556367768182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,1536,0.026692445079485577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,1536,0.011020444333553314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,1536,0.011711999773979187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,3072,0.01314311143424776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,1024,0.024474667178259954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,1024,0.00923111124171151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,1024,0.009840889109505547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,10240,0.05685333410898844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,768,0.0076044441925154785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,768,0.01959111127588484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,768,0.009408888717492422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,512,0.020247111717859905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,512,0.005584888988071018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,512,0.008375999828179678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,2560,0.01219733307758967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,256,0.01755288905567593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,256,0.00453599997692638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,256,0.007989333735571967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,2048,0.011117333339320289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,128,0.016169778174824186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,128,0.004250666747490565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,128,0.007315555380450354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,1536,0.008907555705971187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,64,0.003930666794379552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,64,0.007607111500369177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,12288,32,0.004234666625658671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,12288,32,0.007663110891977946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,1024,0.007607111500369177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,768,0.007252444823582967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,512,0.006980444822046492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,65536,0.26087111896938747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,65536,0.1928622192806668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,16384,0.06698489189147949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,4096,0.03563288847605387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,16384,0.051591111554039844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,256,0.006689777804745569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,12288,0.05333333214124044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,12288,0.05236088898446825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,12288,0.040130668216281466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,10240,0.04639910989337497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,10240,0.04353688822852241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,10240,0.03456444541613261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,12288,128,0.006611555400821898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,8192,0.04220088985231188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,8192,0.03664088911480374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,8192,0.029706666866938274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,65536,0.13000710805257162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,7168,0.03839644458558824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,7168,0.03189333279927572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,7168,0.026408000124825373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,16384,0.03881066706445482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,12288,2560,0.030172444052166406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,6144,0.036489778094821505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,6144,0.027748444014125403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,6144,0.02388266722361247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,12288,0.029789331886503432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,5120,0.03312266535229153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,5120,0.0239955551094479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,5120,0.020998222960366145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,4096,0.031180444690916274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,4096,0.020922667450375024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,4096,0.017779555585649278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,10240,0.025736000802781846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,3584,0.029339555237028334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,3584,0.017837334010336135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,3584,0.016889777448442247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,8192,0.022413333257039387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,7168,0.020076443751653034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,3072,0.016154666741689045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,3072,0.015274667077594332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,6144,0.018410666121376883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,5120,0.01647733317481147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,2560,0.01391999920209249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,2560,0.013968888256284924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,4096,0.014753777119848462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,2048,0.02574311031235589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,2048,0.01202311118443807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,2048,0.012712000144852532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,3584,0.01372977760103014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,1536,0.02397866712676154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,1536,0.010194666683673859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,1536,0.011047110789351992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,3072,0.012829333543777466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,1024,0.020960888928837244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,1024,0.008433777425024245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,1024,0.009324444664849175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,2560,0.011511999699804517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,768,0.018585777944988675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,768,0.006851555572615729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,768,0.008972444468074376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,2048,0.010015111002657149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,512,0.018943111101786297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,512,0.005577777822812398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,16384,0.0643706652853224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,512,0.007923555870850882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,1536,0.008309333688682979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,65536,0.19324889447953966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,256,0.004530666602982415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,256,0.007636444436179266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,128,0.015809777710172866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,128,0.003858666867017746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,128,0.007270221908887227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,1024,0.007320889168315464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,64,0.0038906666967603895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,64,0.007301333049933116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,768,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,10240,32,0.0038773334688610504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,10240,32,0.007497777541478475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,65536,0.19791911707984078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,65536,0.1773866679933336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,512,0.006569777925809224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,256,0.006295111030340195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,65536,0.14783467186821833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,16384,0.05102133419778612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,16384,0.039664887719684176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,12288,0.048604445325003735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,12288,0.03984266519546509
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,12288,0.031145777967241075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,10240,128,0.006583111153708563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,10240,0.04300622145334879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,10240,0.033075554503334895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,65536,0.0808275540669759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,10240,0.027719111906157598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,8192,0.03867288761668735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,8192,0.027791112661361694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,3072,0.028323554330401953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,16384,0.025795555777019922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,8192,0.023686221904224817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,7168,0.024495111571417913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,7168,0.02150933278931512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,12288,0.02095111045572493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,2560,0.026712000370025635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,6144,0.021418665846188862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,6144,0.01924800044960446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,10240,0.018473777506086562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,5120,0.030588444736268785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,8192,0.01621866722901662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,5120,0.018588445252842374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,5120,0.017567111386193168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,4096,0.028578665521409776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,7168,0.014762666490342883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,4096,0.015895111693276297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,4096,0.01514488955338796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,3584,0.026802667313151892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,3584,0.014089778065681458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,3584,0.01402666668097178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,6144,0.013807111316257052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,5120,0.012664889295895895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,3072,0.012688000169065265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,10240,256,0.01754133403301239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,4096,0.011577777564525604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,3072,0.013054221868515015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,2560,0.011056889262464313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,2560,0.024527110987239416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,2560,0.012129777835475074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,3584,0.010840000377760993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,2048,0.009693333672152625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,2048,0.011033777561452655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,3072,0.00980711148844825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,1536,0.021885333789719477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,2560,0.008440888590282863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,1536,0.008238222036096785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,1536,0.009320889082219865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,1024,0.018594667315483093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,1024,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,1024,0.007990222010347579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,2048,0.007625777688291337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,16384,0.058897780047522656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,768,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,768,0.008041777544551426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,1536,0.0069617778062820435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,512,0.017384888397322763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,512,0.004517333375083076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,1024,0.006533333410819371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,512,0.007311111523045435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,256,0.017214222086800467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,256,0.0038862224254343244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,256,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,768,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,6144,0.03353600038422479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,7168,0.03568533394071791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,128,0.0035573335157500375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,128,0.006610666712125142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,64,0.0032213332338465583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,64,0.006945778098371293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,8192,32,0.003527111063400904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,8192,32,0.0069466663731469055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,65536,0.1726906696955363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,512,0.006158222340875202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,65536,0.17995111147562662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,256,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,65536,0.13708089457617864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,16384,0.05758844481574165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,16384,0.04691822330156962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,16384,0.03827022181616889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,12288,0.04761688907941183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,12288,0.036390221781200834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,12288,0.030399110582139757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,8192,128,0.00591111100382275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,3072,0.026038222842746313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,10240,0.04188088907135857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,10240,0.03044355577892727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,10240,0.026748443643252056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,65536,0.07794221904542711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,8192,0.037455999188952975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,8192,0.0262080000506507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,16384,0.025148444705539282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,8192,0.022988445229000513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,12288,0.020261334048377145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,7168,0.022678222921159532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,7168,0.02033244404527876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,6144,0.032814221249686346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,6144,0.019888889458444383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,2048,0.023502222365803186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,6144,0.018921777606010437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,5120,0.029711110724343195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,5120,0.017180444465743173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,10240,0.01794933279355367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,5120,0.016975111431545682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,4096,0.02738222148683336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,4096,0.015140444040298462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,4096,0.014569777581426831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,8192,0.015986666083335876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,3584,0.026403556267420452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,3584,0.013144888811641268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,3584,0.014006222287813822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,7168,0.014471999473041959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,3072,0.02533866630660163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,3072,0.01201244443655014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,3072,0.012904889053768583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,6144,0.013463111387358772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,2560,0.023687111006842718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,2560,0.010666666759385003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,5120,0.012530666258600024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,2560,0.011765333513418833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,768,0.01796444422668881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,2048,0.02239022155602773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,2048,0.009356444080670675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,2048,0.010372444159454769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,1536,0.01939911146958669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,4096,0.011071110765139261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,1536,0.008018666671382057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,1536,0.008950221869680617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,1024,0.018170666363504197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,1024,0.005932444499598608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,1024,0.008276444342401292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,3584,0.010312889185216691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,768,0.018217777212460835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,768,0.004952888935804367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,768,0.008007110820876228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,3072,0.009328888522254096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,512,0.016901334126790363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,512,0.004230222354332606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,512,0.007288888924651676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,2560,0.008288000192907121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,256,0.017222222354676988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,2048,0.007487110793590546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,256,0.003927999900446998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,256,0.0069360000391801195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,8192,128,0.014857777290874057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,128,0.015171556009186639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,128,0.003549333247873518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,128,0.006650666809744305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,1536,0.006798222247097228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,64,0.0032417778339650896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,64,0.006948444578382704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,7168,32,0.0034888887570963967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,7168,32,0.0069271110826068455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,1024,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,65536,0.15800266795688206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,768,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,65536,0.1399946610132853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,65536,0.1280719969007704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,16384,0.048680000834994845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,16384,0.04284266630808512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,16384,0.03674666749106513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,12288,0.04079733292261759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,12288,0.0335822237862481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,12288,0.02949955397182041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,10240,0.03601333167817857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,10240,0.028176890479193792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,10240,0.026076444321208533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,512,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,8192,0.03289599882231818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,8192,0.024311110377311707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,256,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,8192,0.022299556268586054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,7168,0.030452443493737116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,7168,128,0.005881777654091517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,7168,0.02115999990039402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,7168,0.020241777102152508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,65536,0.07567644119262695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,16384,0.0244195560614268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,6144,0.01869777838389079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,6144,0.018567999203999836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,5120,0.02678311036692725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,12288,0.01963200006220076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,5120,0.016063110695944894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,5120,0.016594666573736403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,10240,0.017514665921529133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,4096,0.025032000409232244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,4096,0.01422044469250573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,4096,0.014270222849316068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,8192,0.015452444553375244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,3584,0.02363288899262746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,3584,0.01239733315176434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,3584,0.013560888667901358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,7168,0.014073777529928418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,3072,0.023054222265879314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,3072,0.011433777709801992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,3072,0.01275288893116845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,6144,0.013117333253224691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,2560,0.021873777111371357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,2560,0.011375111010339526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,5120,0.011974222130245633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,2560,0.011744889120260874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,2048,0.019581332802772522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,2048,0.010204444328943888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,2048,0.010040000081062317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,7168,7168,0.03460799985461765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,1536,0.017881777551439073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,1536,0.008652444514963362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,1536,0.008664888640244802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,4096,0.010703110860453712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,1024,0.016864889197879367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,1024,0.006591111007663939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,1024,0.007685333490371704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,768,0.017601778109868366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,768,0.005428444594144821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,3584,0.009651555783218807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,768,0.007640889121426477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,512,0.016187555260128446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,512,0.004660444541109933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,512,0.007304888632562425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,3072,0.00848444468445248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,256,0.016546666622161865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,256,0.003879111260175705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,256,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,2560,0.00794577764140235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,128,0.014834667245546976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,128,0.0034959999223550162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,128,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,2048,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,64,0.0035031110876136353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,64,0.006288888967699475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,6144,32,0.003540444291300244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,1536,0.006611555400821898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,6144,32,0.00674222202764617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,1024,0.006247111078765657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,65536,0.1427217721939087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,65536,0.13632622030046251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,768,0.006238222122192383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,16384,0.0466302235921224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,16384,0.046105778879589505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,65536,0.12252888414594863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,16384,0.035692443450291954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,512,0.006010666489601135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,12288,0.039480888181262545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,12288,0.036822223001056247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,12288,0.029076443778143987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,10240,0.03497066762712266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,10240,0.03052266769938999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,10240,0.02527644402450985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,256,0.005865777946180767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,8192,0.03187111020088196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,8192,0.025776000486479864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,8192,0.021382222572962444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,6144,128,0.005888888819350137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,7168,0.029712888929578993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,7168,0.022459555003378127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,7168,0.01958577831586202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,6144,0.0281333327293396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,6144,0.01997066703107622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,6144,0.01811733345190684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,65536,0.0732933349079556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,5120,0.02603911028967963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,16384,0.02365866634580824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,5120,0.017165333032608032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,5120,0.016495111915800307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,4096,0.024268445041444566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,12288,0.01907911068863339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,4096,0.014484445254007975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,4096,0.014276444911956787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,10240,0.016929777132140267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,3584,0.022663111488024395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,3584,0.013009777499569787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,3584,0.013516444298956128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,8192,0.015039111177126566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,3072,0.011678222152921887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,3072,0.012496000362767113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,7168,0.013441777891582914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,2560,0.020390222469965618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,2560,0.010492444038391113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,2560,0.0107022225856781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,6144,0.012659555508030785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,2048,0.018224888377719455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,2048,0.009316444396972656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,2048,0.009002666506502364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,5120,0.01146044416560067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,4096,0.009639999932712978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,1536,0.017031111650996737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,1536,0.007655999726719326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,3584,0.008894222478071848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,1536,0.008284444610277811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,3072,0.008268444074524773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,1024,0.00582755563987626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,1024,0.007854222423500484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,2560,0.007537778053018782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,768,0.017557332913080852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,768,0.004879999905824661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,768,0.007378666765160031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,2048,0.007091555330488417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,512,0.015490666031837463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,512,0.004466666529575984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,512,0.006983111302057902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,1536,0.006612444503439798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,6144,6144,0.02884977724817064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,256,0.003887999802827835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,256,0.006607111129495833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,128,0.014501333236694336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,1024,0.006230222268237009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,128,0.003544888976547453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,128,0.006630222416586346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,64,0.003160888950030009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,64,0.006599999964237213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,5120,32,0.003181333343187968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,768,0.006048888795905643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,5120,32,0.006598222172922558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,65536,0.10587466425365871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,65536,0.08236533403396606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,512,0.005945777727497949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,16384,0.04381777842839559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,16384,0.03429155548413595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,256,0.005838222387764189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,16384,0.025126222107145522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,12288,0.03706399930848016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,5120,128,0.005882666756709416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,12288,0.027221333649423387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,12288,0.02072000006834666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,10240,0.032913777563307024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,10240,0.020495999190542433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,10240,0.018568888306617737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,65536,0.0711617800924513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,8192,0.03000533249643114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,8192,0.017215111189418368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,16384,0.02267733381854163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,8192,0.01644711030854119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,7168,0.027826666831970215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,7168,0.015112888481881885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,7168,0.015138667490747241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,12288,0.018567999203999836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,6144,0.02641155487961239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,6144,0.013797333670987023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,6144,0.013800000150998434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,10240,0.01640977793269687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,5120,0.024111110303137038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,5120,0.01203733351495531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,5120,0.013069333301650153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,8192,0.014475555883513557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,4096,0.023246222072177466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,4096,0.011060444845093621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,4096,0.011375111010339526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,7168,0.013162666724787818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,3584,0.021984888447655573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,3584,0.009400000174840292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,3584,0.01072266697883606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,6144,0.012065778175989786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,3072,0.01958044370015462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,3072,0.008632889224423302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,3072,0.010055111514197456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,5120,0.010400888820489248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,2560,0.017672888106769986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,2560,0.00776088900036282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,2560,0.008664000365469191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,4096,0.00903911143541336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,2048,0.01698755555682712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,2048,0.006911110960774952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,2048,0.007655111451943715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,3584,0.00833777752187517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,1536,0.016135111451148987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,1536,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,1536,0.006949333681000604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,3072,0.007999999655617608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,1024,0.0151582227812873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,1024,0.004286222159862518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,1024,0.0064382221963670515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,2560,0.007302222152551015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,768,0.017159110969967313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,768,0.004182222402758068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,768,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,2048,0.006960888703664144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,1536,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,512,0.0035182221068276297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,512,0.006215999937719769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,1024,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,3072,0.02102488941616482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,256,0.0032071111102898917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,256,0.006008000009589725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,768,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,512,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,128,0.014803555276658801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,1024,0.016195555528004963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,128,0.0032026666320032547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,128,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,64,0.002879111096262932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,64,0.006226666685607698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,4096,32,0.0028355556229750314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,4096,32,0.005943999936183293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,256,0.005579555614127054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,65536,0.09746222363577949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,65536,0.12741155094570583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,65536,0.07714311281840007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,16384,0.043416887521743774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,16384,0.031249778138266668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,4096,128,0.00555288874440723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,16384,0.024241778585645888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,12288,0.03666311171319749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,12288,0.024875554773542616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,12288,0.019882667395803664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,65536,0.06974133518007067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,10240,0.021943999661339655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,10240,0.018221333622932434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,5120,256,0.015240000353919135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,8192,0.029492444462246362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,8192,0.016619554824299283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,16384,0.022318222456508215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,8192,0.015829333000712924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,7168,0.02756533357832167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,7168,0.014763555592960782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,12288,0.018237334158685472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,7168,0.014744889405038623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,6144,0.0262444449795617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,6144,0.013075555364290873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,65536,0.12790132893456355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,6144,0.013697778185208639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,5120,0.023823110593689814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,10240,0.016228444046444364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,5120,0.011728888584507836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,5120,0.012688000169065265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,4096,0.02238133384121789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,8192,0.01427111029624939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,4096,0.01202488856183158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,4096,0.011163555913501315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,3584,0.019868445065286424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,3584,0.010762666662534079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,7168,0.012793777717484368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,3584,0.01035911093155543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,3072,0.018218666315078735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,3072,0.008532444636027018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,3072,0.009177777502271865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,6144,0.011367999845080905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,2560,0.017156443662113614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,2560,0.007473777565691207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,5120,0.010020444790522257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,2560,0.00796088907453749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,2048,0.01646399994691213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,2048,0.006400888992680444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,4096,0.008787555827034844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,2048,0.007305777735180325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,1536,0.015457777513398064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,1536,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,3584,0.008303110798199972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,1536,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,3072,0.007979555262459649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,1024,0.0042088888585567474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,1024,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,2560,0.007302222152551015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,2048,0.006976889239417181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,768,0.0041866666740841335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,768,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,1536,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,512,0.014780445231331719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,512,0.003858666867017746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,512,0.006252444452709622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,1024,0.006216889040337668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,256,0.015503111812803479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,256,0.0032017777363459268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,256,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,768,0.005923555543025334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,128,0.014936889211336771
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,128,0.0032888888898822996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,128,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,512,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,64,0.0028355556229750314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,64,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3584,32,0.0029315555261241067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3584,32,0.006245333287451003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,256,0.005532444351249271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,65536,0.12579911284976536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,65536,0.08328711324267916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,65536,0.07396711243523492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3584,128,0.005537777725193236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,16384,0.042818665504455566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,16384,0.026683555708991155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,16384,0.02353422178162469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,12288,0.03602844476699829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,12288,0.022908444205919903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,12288,0.019490665859646268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,65536,0.06854399707582262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,16384,0.021951110826598272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,10240,0.03201599915822347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,512,0.0148026661740409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,10240,0.023023110296991136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,10240,0.01755466726091173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,12288,0.017886221408843994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,8192,0.01956266661485036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,8192,0.01555199921131134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,4096,256,0.01494488947921329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,7168,0.017304889029926725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,7168,0.014374223020341663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,10240,0.01585155559910668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,6144,0.02558044426971012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,6144,0.015448888142903646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,6144,0.013091555900043912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,8192,0.013795555465751223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,5120,0.02297244469324748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,5120,0.014095111025704278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,5120,0.012071111136012606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,7168,0.012251555919647217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,4096,0.020873778396182593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,4096,0.011744000017642975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,4096,0.010390222072601318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,6144,0.011054221954610614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,3584,0.02012355625629425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,3584,0.010693333215183683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,3584,0.010028444230556488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,10240,0.03279111120435927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,5120,0.009534222384293875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,3072,0.018138666947682697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,3072,0.00978311151266098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,3072,0.008725333544943068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,4096,0.008672888908121321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,2560,0.008804444637563493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,2560,0.007631999750932057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,3584,0.008338666624493068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,3072,0.007924444145626491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,2048,0.007384888827800751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,2048,0.007118222614129384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,2560,0.007321777443091075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,1536,0.015152000718646579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,1536,0.0061919999619325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,1536,0.006626666833957036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,2048,0.006959111326270633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,1024,0.014748444159825643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,1024,0.004893333133724001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,1024,0.00656444455186526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,1536,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,1024,0.006205333189831839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,768,0.00453244439429707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,1024,0.014817777607176038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,768,0.006251555350091722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,512,0.014447111222479077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,512,0.00387999994887246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,512,0.006224000205596288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,768,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,256,0.01500266624821557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3584,768,0.01698933376206292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,256,0.003516444315512975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,256,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,512,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,128,0.003145777723855443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,128,0.014488000008794995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,128,0.00592533333433999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,64,0.00286666676402092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,64,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,3072,32,0.0028702221396896574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,3072,32,0.005932444499598608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,65536,0.12459111213684082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,65536,0.07100266880459256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,256,0.005526222288608551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,65536,0.0708835588561164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,3072,128,0.005583111196756363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,16384,0.04217511084344652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,16384,0.02401688860522376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,16384,0.02271911170747545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,12288,0.03568799959288703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,12288,0.02021422154373593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,12288,0.019183999962276883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,65536,0.06730844577153523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,10240,0.031545778115590416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,10240,0.017874666386180453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,16384,0.021560000048743352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,10240,0.01735911104414198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,8192,0.0280693338976966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,8192,0.016324443949593436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,8192,0.014975999792416891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,12288,0.017700443665186565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,10240,0.015496888094478183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,7168,0.014543111125628153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,7168,0.014110222458839417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,8192,0.013402666482660504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,7168,0.011356444822417365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,6144,0.013175999952687157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,6144,0.013089777694808112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,8192,0.02920711040496826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,5120,0.02162844439347585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,7168,0.027049778236283198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,5120,0.011919111013412476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,5120,0.012034667034943899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,6144,0.010058666268984476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,5120,0.009545777406957414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,4096,0.010431999961535135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,4096,0.009641778137948778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,2560,0.016600888636377122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,4096,0.008795555267069075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,3584,0.018195556269751657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,3584,0.009475555684831407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,3584,0.008606221940782335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,3072,0.00865955568022198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,3072,0.008392888638708327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,2048,0.016487111647923786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,2560,0.016504888733228047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,2560,0.007968000239796108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,2560,0.007659555309348636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,3584,0.008290666672918532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,2048,0.015967110792795818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,2048,0.006251555350091722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,3072,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,2048,0.007271111011505127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,2560,0.00738666703303655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,1536,0.00553955551650789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,1536,0.0069822221994400024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,2048,0.006979555719428592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,1024,0.014634667171372307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,1024,0.004577777865860197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,1024,0.006646222124497096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,1536,0.006575111299753189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,768,0.01682844426896837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,768,0.0042204442951414315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,768,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,1024,0.006031111296680238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,512,0.013789333403110504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,512,0.0035546666218174826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,3072,768,0.017156443662113614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,512,0.005923555543025334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,768,0.005906666732496685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,256,0.003504000190231535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,256,0.006223111102978389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,512,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,128,0.014148443937301636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,128,0.0032044444233179092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,128,0.005965333431959152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,256,0.005559999909665849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,64,0.0028684443483750024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,64,0.006231110956933763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2560,32,0.0028453332682450614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2560,32,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2560,128,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,65536,0.10098489125569661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,65536,0.05658311314053006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,65536,0.06857066684299044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,16384,0.03636088967323303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,16384,0.02018311123053233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,16384,0.022265776991844177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,65536,0.06603644291559856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,12288,0.031120889716678198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,12288,0.01647644407219357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,12288,0.018801778554916382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,16384,0.021316443880399067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,7168,0.0266275554895401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,12288,0.017215111189418368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,10240,0.01588800052801768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,10240,0.01681155628628201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,8192,0.025425778494940862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,8192,0.01514933341079288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,8192,0.014656000667148165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,10240,0.01481244464715322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,7168,0.023231110639042322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,7168,0.013436444103717804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,6144,0.02478133307562934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,7168,0.01366666704416275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,8192,0.012488889197508493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,6144,0.021370665894614324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,6144,0.012625777886973487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,6144,0.012670222255918713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,7168,0.010562666588359408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,5120,0.010479110810491772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,5120,0.011014222270912595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,6144,0.009939555492666032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,5120,0.008976000050703684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,4096,0.00924711094962226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,4096,0.008988444175985124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,4096,0.019311999281247456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,3584,0.016882666283183627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,3584,0.008652444514963362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,4096,0.008597333398130205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,3584,0.008336000144481659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,3584,0.007990222010347579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,3072,0.007691555553012424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,3072,0.016875555117925007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,3072,0.007981333467695449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,2560,0.015792888071801927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,2560,0.006537777682145436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,2560,0.007649777664078607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,3072,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,2048,0.015126221709781222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,2048,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,2048,0.006974221931563483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,2560,0.007297777467303806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,1536,0.01513066722287072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,1536,0.014780445231331719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,1536,0.004875555634498596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,1536,0.0069422221018208405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,2048,0.006908444480763541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,1536,0.006573333508438534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,1024,0.0042088888585567474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,1024,0.0064222220745351576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,1024,0.006166222194830577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,768,0.01705511079894172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,768,0.0038773334688610504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,768,0.006261333409282897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,512,0.01310222182008955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,512,0.003555555724435382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,512,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2560,256,0.01521511044767168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,256,0.015232000086042615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,256,0.0032311110860771606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,768,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,256,0.005960000058015187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,128,0.014633778068754407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,128,0.002862222285734283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,128,0.005962666538026597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,512,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,64,0.0028168888141711554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,64,0.005667555663320754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,2048,32,0.003177777760558658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,2048,32,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,256,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,65536,0.0459555553065406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,65536,0.0992257793744405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,2048,128,0.005570666657553778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,65536,0.06677333513895671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,16384,0.01695644524362352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,16384,0.021528000632921856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,65536,0.06478933493296306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,12288,0.030491554074817236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,12288,0.014459555347760519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,16384,0.02087466584311591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,12288,0.018173333671357896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,10240,0.027094221777386133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,10240,0.015113777584499784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,10240,0.016400888562202454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,10240,0.028121779362360638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,8192,0.024718221690919664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,12288,0.01643288963370853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,8192,0.015127110812399121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,8192,0.014083556003040738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,10240,0.014119999276267158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,7168,0.01272266689274046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,7168,0.01293333371480306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,8192,0.01148088855875863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,6144,0.020255110330051847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,6144,0.011351111034552256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,6144,0.011371555427710215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,7168,0.010390222072601318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,6144,0.009660444325870937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,5120,0.010086221827401055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,5120,0.019202666150199044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,5120,0.009992000129487779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,4096,0.017504889104101393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,4096,0.008802666432327693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,4096,0.008671999805503422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,5120,0.008986666798591614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,3584,0.016689777374267578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,3584,0.008032889001899296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,3584,0.00831288927131229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,4096,0.008623110751310984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,3584,0.008288000192907121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,3072,0.006969778074158563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,3072,0.008060444560315874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,4096,0.01789688898457421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,3072,0.007726222276687622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,2560,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,2560,0.0076044441925154785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,2560,0.007280888656775157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,3072,0.01650311052799225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,2048,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,2048,0.006737777756320105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,2048,0.006907555378145642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,1536,0.014106666048367819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,1536,0.004909333255555895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,1536,0.00665155549844106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,1536,0.006501333167155583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,1024,0.01351377781894472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,1024,0.004243555582231945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,1024,0.006266666783226862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,2048,1024,0.013471999930010902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,768,0.01613333324591319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,768,0.0038355555799272326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,768,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,1024,0.005960888746711943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,512,0.013122667041089801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,512,0.003536888708670934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,512,0.0063564446237352155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,768,0.005623999983072281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,512,0.005568888866239124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,256,0.0032026666320032547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,256,0.00563466673096021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,256,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,128,0.014142221874660917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,128,0.0032071111102898917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,128,0.005919111271699269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,16384,0.03580355644226074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,64,0.0028897778441508612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,64,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1536,32,0.0028648889727062653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1536,32,0.005666666560702854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1536,128,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,65536,0.09240177604887222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,65536,0.034239110019471906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,65536,0.06495733393563165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,16384,0.034303999609417386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,16384,0.013969777358902825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,16384,0.02107111116250356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,7168,0.02199466692076789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,65536,0.06416089004940458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,12288,0.015295111470752291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,12288,0.029122667180167303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,12288,0.0177182224061754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,10240,0.025851555996470984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,16384,0.019648889700571697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,10240,0.014096889230940076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,10240,0.016163556112183463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,12288,0.014711111783981323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,8192,0.02216177847650316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,5120,0.018519111805491976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,8192,0.01235466698805491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,8192,0.013105777402718862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,7168,0.011763555308183035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,7168,0.01146577795346578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,10240,0.012443555725945367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,8192,0.011342222491900126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,6144,0.010999999940395355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,6144,0.01036977767944336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,7168,0.010438222024175854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,6144,0.009683555199040307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,5120,0.010305778019958073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,5120,0.009639110830095079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,3072,0.016492444607946608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,5120,0.008950221869680617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,4096,0.00792711145348019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,4096,0.008621333373917473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,4096,0.00832177781396442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,3584,0.01649777756796943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,3584,0.011301333705584208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,3584,0.008202667037645975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,3584,0.007939555578761632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,3072,0.01566488875283135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,2560,0.01551555593808492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,3072,0.01032977799574534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,3072,0.007668444679843054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,2048,0.015153777268197803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,2560,0.008936000367005667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,2560,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,3072,0.007600000335110559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,2560,0.007293333609898885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,2048,0.007643555601437886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,2048,0.00703022215101454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,2048,0.006918222126033571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,1536,0.014133334159851074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,1536,0.004824000100294749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,1536,0.006735999965005451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,1536,0.006607111129495833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1536,256,0.014766222900814481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,1024,0.00424177779091729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,1024,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,1024,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,7168,0.020625778370433383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,768,0.003858666867017746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,768,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,768,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,6144,0.01960000064637926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,512,0.0035457776652442086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,512,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,4096,0.017179555363125272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,5120,0.018242667118708294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,256,0.015208888385030957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,256,0.0033777778347333274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,256,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,128,0.014947555131382413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,512,0.005472888963090048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,128,0.0031662223239739737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,128,0.005758222192525864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,64,0.0028684443483750024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,256,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,64,0.005966222120655908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,1024,32,0.0028817777832349143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,1024,32,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,65536,0.02681066592534383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,1024,128,0.005573333137565189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,65536,0.06409688790639241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,16384,0.03368799885114034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,16384,0.014905777242448596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,65536,0.06407822502983941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,16384,0.020755555894639757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,16384,0.01813155578242408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,12288,0.013040000365840064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,12288,0.017151110702090792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,2048,0.014764444695578681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,12288,0.014440000057220459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,10240,0.024001777172088623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,10240,0.01311288856797748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,2560,0.015503111812803479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,10240,0.0147706667582194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,8192,0.021983999345037673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,8192,0.011523555550310345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,8192,0.011953777737087674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,10240,0.012425777812798819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,8192,0.011337777806652917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,7168,0.011177777416176267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,1024,0.013732444081041547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,7168,0.010993777877754636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,6144,0.01955111159218682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,6144,0.01033155537313885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,7168,0.01033333357837465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,6144,0.01017066670788659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,5120,0.018180444836616516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,5120,0.008959111240175035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,5120,0.009686222506894005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,6144,0.009644444617960189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,4096,0.007941332956155142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,4096,0.008623110751310984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,5120,0.008902221918106079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,3584,0.01647733317481147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,3584,0.007635555333561367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,4096,0.008640888664457533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,3584,0.008210666477680206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,3072,0.015792000624868605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,3072,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,3584,0.008220444122950235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,3072,0.00794577764140235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,2560,0.015132443772421943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,2560,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,2560,0.007310222420427535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,3072,0.007828444242477417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,2048,0.015096889601813423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,2048,0.005210666606823604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,2048,0.006952889263629913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,2560,0.007253333098358578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,1536,0.013809777796268463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,1536,0.004560889055331548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,1536,0.0069146665434042616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,2048,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,1024,0.01346933344999949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,1024,0.004208000169859992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,1024,0.0063182223174307085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,1536,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,768,0.015846222639083862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,768,0.015796444482273526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,768,0.0038488888078265717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,768,0.005960888746711943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,1024,0.006142222219043308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,512,0.012776888906955719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,512,0.0031911111954185697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,512,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,768,0.005561777700980504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,256,0.0031742221779293488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,256,0.005624889085690181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,512,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,128,0.01440622243616316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,128,0.0028648889727062653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,128,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,256,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,64,0.002858666703104973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,64,0.005623111294375525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,768,32,0.002826666666401757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,768,32,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,768,128,0.005537777725193236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,65536,0.028976887464523315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,65536,0.06361244122187297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,65536,0.06412622001436022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,16384,0.031567109955681696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,16384,0.017090666625234816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,16384,0.020438222421540152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,16384,0.017501332693629794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,12288,0.026512889398468867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,12288,0.011513777905040316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,12288,0.015455999308162265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,12288,0.01443733274936676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,10240,0.02365422248840332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,10240,0.01071111112833023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,10240,0.013172444370057849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,10240,0.012430222498046027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,1024,512,0.013399110900031196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,8192,0.009099555512269339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,8192,0.011692444483439127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,8192,0.011388444238238864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,7168,0.019882667395803664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,7168,0.008907555705971187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,7168,0.010812444819344414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,7168,0.010341333018408881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,6144,0.018571555614471436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,6144,0.007979555262459649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,6144,0.010383110907342698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,65536,0.09162044525146484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,5120,0.01756266587310367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,6144,0.009646221995353699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,5120,0.007663110891977946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,5120,0.009651555783218807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,12288,0.028464890188641016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,4096,0.01644000079896715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,5120,0.008964444200197855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,4096,0.007671111159854465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,4096,0.008632889224423302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,3584,0.015471110741297403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,3584,0.007623111208279927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,4096,0.008352888955010308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,3584,0.00795733349190818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,3072,0.015459555718633862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,3072,0.006578666468461354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,3584,0.008005333443482717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,3072,0.00795022232664956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,2560,0.014785778191354541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,2560,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,3072,0.0075333333677715724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,2560,0.00759733302725686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,2560,0.0069546666410234236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,2048,0.005247999810510212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,2048,0.006970666348934174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,2048,0.006601777755551868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,1536,0.013457777599493662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,1536,0.0047333331571684945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,1536,0.006569777925809224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,7168,0.020977778567208186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,1024,0.013105777402718862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,4096,0.016879111528396606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,1024,0.003859555555714501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,1024,0.006284444282452266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,1536,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,768,0.014483556151390076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,768,0.003572444534964032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,768,0.005940444353553984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,1024,0.006057777752478917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,512,0.003510222252872255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,512,0.0058702222175068325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,768,0.005622222191757626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,256,0.0032026666320032547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,256,0.014530667000346713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,256,0.005690666536490123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,512,0.005462222215202119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,128,0.012099555797047086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,128,0.0028417776856157514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,128,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,256,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,64,0.0025395556456512874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,64,0.0059262220230367445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,512,32,0.00281866660548581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,512,128,0.005254222287072076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,512,32,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,65536,0.019368888603316415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,65536,0.09035555521647136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,65536,0.06313155757056342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,16384,0.029841777351167467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,16384,0.01036088913679123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,65536,0.06266844272613525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,16384,0.01754844519827101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,12288,0.025437333517604407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,16384,0.017487110363112558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,12288,0.00908622228437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,12288,0.014449778530332776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,10240,0.02273066673013899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,10240,0.01240888900227017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,12288,0.01440355512830946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,10240,0.012792888614866467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,8192,0.020265777905782063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,8192,0.01107022249036365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,10240,0.012431110772821637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,8192,0.011714666253990598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,8192,0.011239111423492432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,7168,0.009697777529557547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,7168,0.010799999866220685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,7168,0.01033155537313885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,6144,0.007657777931955125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,768,256,0.014156444205178155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,6144,0.01034311122364468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,5120,0.01681688924630483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,6144,0.009534222384293875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,5120,0.00829777783817715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,5120,0.009341333475377824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,4096,0.015820443630218506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,5120,0.008949333594905006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,4096,0.007282666862010956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,4096,0.008663111262851292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,3584,0.015159999330838522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,4096,0.008282666405042013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,3584,0.007349333001507654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,3584,0.007939555578761632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,3072,0.014415999253590902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,3584,0.008223111430803934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,3072,0.0069395556218094295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,3072,0.007977777885066139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,2560,0.014116444521480136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,3072,0.007594666547245449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,2560,0.0058702222175068325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,2560,0.007301333049933116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,65536,0.0905839999516805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,2048,0.013172444370057849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,2560,0.007302222152551015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,2048,0.005223999834722943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,2048,0.006738666858938005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,1536,0.01238666640387641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,2048,0.006585777633719974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,1536,0.004303110970391167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,1536,0.006586666736337874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,1536,0.006274666637182236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,1024,0.0038328886859946777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,1024,0.006280888699822956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,1024,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,768,0.013455111119482251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,768,0.003495111233658261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,768,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,768,0.005668444352017508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,512,0.011381333072980246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,512,0.0034977777136696708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,512,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,512,0.005530666559934616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,256,0.012094222009181976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,256,0.002867555452717675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,256,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,8192,0.02161600026819441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,128,0.012096000214417776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,256,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,128,0.00282577777074443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,128,0.005623999983072281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,64,0.002522666628162066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,256,128,0.005554666535721884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,64,0.0057795556883017225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,256,32,0.002830222249031067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,256,32,0.005583111196756363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,65536,0.015400000744395785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,65536,0.06161777840720283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,65536,0.060897780789269336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,16384,0.029789331886503432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,16384,0.00959822204377916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,16384,0.01739911072784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,12288,0.025059555967648823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,16384,0.017179555363125272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,12288,0.008385777473449707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,12288,0.014167999227841696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,12288,0.014178666803571912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,10240,0.021925333473417494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,10240,0.009317333499590555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,10240,0.012757333616415659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,10240,0.012381333443853589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,8192,0.008292444050312042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,8192,0.011624000138706632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,8192,0.01145866678820716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,7168,0.018896889355447557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,7168,0.00797333319981893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,7168,0.01110400011142095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,7168,0.010194666683673859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,6144,0.018170666363504197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,6144,0.007283555964628856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,6144,0.0103502223889033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,6144,0.009345778160625035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,5120,0.016521778371598985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,5120,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,5120,0.009645333720578088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,5120,0.008946666287051307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,2048,0.014141332772043018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,4096,0.006581333362393909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,4096,0.00865955568022198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,4096,0.008416000339719985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,3584,0.01477688882086012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,3584,0.006270222365856171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,3584,0.008336889247099558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,3584,0.008055999875068665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,512,512,0.012094222009181976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,3072,0.006285333385070165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,3072,0.007965332931942409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,3072,0.007655999726719326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,2560,0.0141004439857271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,2560,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,2560,0.007301333049933116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,2560,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,2048,0.013456888496875763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,2048,0.006292444550328785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,2048,0.006962666908899943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,2048,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,7168,0.01881511178281572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,1536,0.004579555657174853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,1536,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,1536,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,6144,0.017907554904619854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,1024,0.003909333298603694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,1024,0.006204444501135085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,1024,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,768,0.01277955538696713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,768,0.0035333335399627686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,768,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,768,0.005625777774386936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,256,1024,0.011768889096048145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,512,0.0032293332947625052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,512,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,512,0.0052639999323421055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,256,0.010679110884666443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,256,0.0030906666070222855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,256,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,65536,0.08978933095932007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,256,0.005583999885453119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,128,0.011696889168686338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,128,0.002814222127199173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,128,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,160,128,128,0.005571555760171678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,64,0.002642666714058982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,64,0.005680000202523337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,128,32,0.002817777709828483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,128,32,0.005943999936183293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,65536,0.018573333819707234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,65536,0.06086577971776327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,8192,0.019820445113711886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,16384,0.009032888544930352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,16384,0.016869333055284288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,12288,0.007965332931942409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,12288,0.014191110928853353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,10240,0.007612444460391998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,10240,0.012726222475369772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,8192,0.007261333366235097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,8192,0.011679111255539788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,7168,0.0069208890199661255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,7168,0.010668444136778513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,6144,0.006579555571079254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,6144,0.009707555174827576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,5120,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,5120,0.009309333231714036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,4096,0.00830666638082928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,4096,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,3584,0.006249777972698212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,3584,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,3072,0.006219555520349079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,3072,0.00793511089351442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,2560,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,2560,0.007300444775157505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,2048,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,2048,0.0069288888739215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,1536,0.00425155543618732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,1536,0.006599111275540457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,1024,0.003863111138343811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,1024,0.005980444451173146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,768,0.003551111039188173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,768,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,512,0.00315022220214208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,512,0.005932444499598608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,256,0.0029031110720502008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,4096,0.015485333071814643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,128,0.005587555468082428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,256,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,128,0.002529777793420686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,64,0.002520888836847411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,64,0.005561777700980504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,64,32,0.0028124445428450904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,64,32,0.005921777751710679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,65536,0.015839111473825242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,65536,0.05946755409240723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,12288,0.007646222081449296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,16384,0.008278222547637092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,16384,0.017216889394654166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,12288,0.014485332700941296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,10240,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,10240,0.01293333371480306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,8192,0.0069573331210348345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,8192,0.011384888655609555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,7168,0.006562666760550604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,7168,0.010679999987284342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,6144,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,6144,0.00997688869635264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,5120,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,5120,0.00922400007645289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,4096,0.006235555642180973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,3584,0.007689778175618913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,4096,0.008289777570300633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,3584,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,3072,0.005954666684071223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,3072,0.007631999750932057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,2048,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,2560,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,2560,0.007305777735180325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,2048,0.006575111299753189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,1536,0.004501333253251182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,1536,0.006272888845867581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,1024,0.0038497779104444715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,1024,0.00600888869828648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,768,0.00351555562681622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,768,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,512,0.0032551110618644287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,512,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,256,0.0029235554652081597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,128,0.005592888842026393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,256,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,128,0.0028568889117903183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,64,0.002581333327624533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,64,0.0052684446175893145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,160,32,32,0.0028106667515304354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,160,32,32,0.005551110953092575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,3072,0.014504000544548035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,16384,0.37787289089626735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,16384,0.18256355656517875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,12288,0.19237332873874238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,12288,0.2814719941880968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,12288,0.1367502212524414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,10240,0.1662720044453939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,1536,0.012758221891191272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,10240,0.24756177266438803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,10240,0.11477600203620063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,1024,0.011713777979214987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,8192,0.19889244768354628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,8192,0.1491155491934882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,8192,0.09580710861417983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,160,128,512,0.010729778144094678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,7168,0.17509599526723227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,7168,0.13191733095380995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,7168,0.08204355504777697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,6144,0.12264622582329644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,6144,0.14983999729156494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,6144,0.07172889179653592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,5120,0.10883644554350112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,5120,0.1267804437213474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,5120,0.06047733624776205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,16384,0.23698133892483184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,4096,0.09850400023990208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,4096,0.1018106672498915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,4096,0.049032890134387545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,16384,0.1361617777082655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,3584,0.09198666943444146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,3584,0.09068444702360366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,3584,0.044450667169359
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,12288,0.10238755411571926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,3072,0.08558133575651382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,3072,0.07851022481918335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,3072,0.0388062232070499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,10240,0.0861555536588033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,2560,0.07928533024258085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,8192,0.0701999995443556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,2560,0.06561866733762953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,2560,0.03369777732425266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,2048,0.07521866427527533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,2048,0.052892443206575185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,7168,0.06169066826502482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,2048,0.02941333254178365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,6144,0.05471377902560764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,1536,0.04125510983996921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,1536,0.023988443944189284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,5120,0.046331554651260376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,4096,0.03933599922392104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,1024,0.02978666623433431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,1024,0.0196426659822464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,1536,0.0688248872756958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,1024,0.06263911061816745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,768,0.02272266646226247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,768,0.017208000024159748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,3584,0.034953776333067156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,512,0.05719021956125895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,512,0.015818667080667283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,512,0.01566755606068505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,3072,0.031382222970326744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,256,0.02551199992497762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,256,0.010346666806273991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,256,0.014162666267818876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,2560,0.027514666318893433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,2048,0.023923556009928387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,128,0.022154667311244543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,128,0.008323555191357931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,128,0.011044444309340583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,64,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,64,0.011361777782440186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,65536,32,0.006252444452709622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,65536,32,0.012500444220172035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,1536,0.02012888921631707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,65536,0.33805155754089355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,65536,0.1908337805006239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,65536,768,0.030729777283138696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,16384,0.07694933149549696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,16384,0.08857155508465236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,16384,0.052687999274995595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,65536,0.14866133530934653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,16384,0.04177777634726631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,12288,0.06712889009051852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,12288,0.04105510976579454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,1024,0.01651733285850949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,65536,0.2389226754506429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,10240,0.05720800161361694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,10240,0.03557777735922072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,768,0.014655111564530266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,12288,0.03299199872546726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,8192,0.04664977722697788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,8192,0.030199110507965088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,512,0.013734222286277346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,256,0.01254488858911726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,7168,0.04091644287109375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,65536,128,0.012440889245933957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,7168,0.02699022160636054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,10240,0.028509332074059382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,6144,0.03619644376966689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,6144,0.024620445238219366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,12288,0.06300266583760579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,5120,0.038380444049835205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,5120,0.031153778235117596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,5120,0.02130399975511763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,8192,0.024344000551435683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,7168,0.021634666456116572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,4096,0.0358497765329149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,4096,0.02588355541229248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,4096,0.017828444639841717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,10240,0.05504444572660658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,3584,0.023018666439586218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,6144,0.019663110375404358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,3584,0.016744888491100736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,3072,0.03230488962597317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,3072,0.020437333318922255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,3072,0.015116444892353482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,8192,0.0506879985332489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,5120,0.01758044461409251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,2560,0.01757688820362091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,2560,0.013631111217869652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,4096,0.015800888339678448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,2048,0.029657777812745836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,7168,0.04547199938032362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,2048,0.014486221803559197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,3584,0.014668444792429606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,2048,0.012340444657537671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,1536,0.027690667245123122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,1536,0.011715555356608497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,1536,0.011352889239788055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,3072,0.013730666703648038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,1024,0.009574222067991892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,1024,0.010028444230556488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,6144,0.04236977630191379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,768,0.020604444874657523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,768,0.00793866647614373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,768,0.008037333687146505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,2560,0.012690666649076672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,512,0.02199466692076789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,512,0.006504000061088138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,512,0.007257777783605788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,2048,0.011580444044537015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,256,0.01719466679626041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,256,0.0042204442951414315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,256,0.006738666858938005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,1536,0.010471111370457543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,128,0.015508444772826301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,128,0.003899555653333664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,128,0.006559111177921295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,1024,0.00868266655339135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,64,0.0037297776175869834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,64,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,16384,32,0.0038817777401871155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,16384,32,0.007320000065697565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,3584,0.03393244412210252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,768,0.007408888803588019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,65536,0.1808177762561374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,65536,0.2521608935462104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,65536,0.16832266913519967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,16384,0.06241599718729655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,16384,0.0664551125632392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,16384,0.0469448897573683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,12288,0.05101422137684292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,12288,0.051575111018286816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,12288,0.03661422265900506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,512,0.0069475554757648045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,10240,0.04520622226927015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,10240,0.04309066798951891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,10240,0.032093332873450384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,256,0.006601777755551868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,8192,0.041542222102483116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,8192,0.03661155700683594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,2560,0.030829333596759375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,8192,0.027560000618298847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,16384,128,0.006612444503439798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,7168,0.031303998496797346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,65536,0.13486754894256592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,7168,0.024704888463020325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,16384,0.03916533456908332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,6144,0.027727999620967444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,6144,0.022052443689770166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,12288,0.03140711122088962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,16384,1024,0.026026666164398193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,5120,0.03239288926124573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,5120,0.02398666739463806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,5120,0.01960088809331258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,4096,0.03048355711830987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,4096,0.020671111014154222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,4096,0.016741333736313712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,10240,0.02641955514748891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,3584,0.028982222080230713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,3584,0.01777688827779558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,3584,0.015436444017622205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,8192,0.022671111755900916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,3072,0.02805244591501024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,3072,0.01566133399804433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,3072,0.014512000812424554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,7168,0.02032444377740224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,2560,0.02661955522166358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,2560,0.01350488927629259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,2560,0.013111111190583972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,6144,0.018782221608691745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,5120,0.01662400033738878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,2048,0.011672000090281168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,2048,0.012069332930776807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,4096,0.0151582227812873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,3584,0.013835555149449242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,1536,0.0240320000383589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,1536,0.009747555686367882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,1536,0.010984888507260216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,1024,0.02181688944498698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,1024,0.00832177781396442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,3072,0.012977778083748288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,1024,0.00830666638082928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,768,0.017493334081437852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,768,0.006904888898134232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,768,0.007263110743628607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,2560,0.01219377749496036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,512,0.018906666172875296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,512,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,512,0.006955555743641323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,2048,0.011063999599880643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,256,0.01682577696111467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,256,0.0041982221106688184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,256,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,1536,0.009017777939637503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,128,0.01482311056719886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,128,0.00388355553150177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,128,0.00666933341158761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,1024,0.0074035558435651995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,64,0.003549333247873518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,6144,0.0351200004418691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,64,0.006686222222116258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,12288,32,0.0038488888078265717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,768,0.007085333267847697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,12288,32,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,7168,0.037434667348861694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,512,0.006969778074158563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,65536,0.21706222163306343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,65536,0.1617866622077094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,16384,0.05726310941908094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,16384,0.05554933349291483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,65536,0.15688089529673258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,16384,0.04420266548792521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,12288,0.043972445858849414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,12288,0.03512444429927402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,10240,0.041259557008743286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,10240,0.03670488794644674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,10240,0.03089244498146905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,256,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,8192,0.03832622369130453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,8192,0.03045511245727539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,8192,0.026085333691702947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,12288,128,0.006599999964237213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,7168,0.03432710965474447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,7168,0.026679999298519556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,7168,0.023214222656355962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,65536,0.13133955001831055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,16384,0.037921776374181114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,6144,0.023396443989541795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,12288,0.029731555117501154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,6144,0.021292444732454088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,5120,0.030216889248953924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,5120,0.02033422225051456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,12288,2048,0.025664889150195654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,5120,0.018888889087571036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,4096,0.0284586681260003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,4096,0.01756622228357527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,10240,0.025763556361198425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,4096,0.016159999701711867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,8192,0.022299556268586054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,3584,0.014767110347747803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,3584,0.015159999330838522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,7168,0.019869332512219746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,3072,0.02583822276857164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,3072,0.013539555172125498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,3072,0.014018666413095264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,6144,0.017961778574519686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,2560,0.025047111842367385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,2560,0.011782222323947482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,2560,0.01274400038851632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,5120,0.016306666864289176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,2048,0.023919999599456787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,2048,0.010298666854699453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,2048,0.011390222443474663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,4096,0.014808000789748298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,1536,0.022630222969584998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,1536,0.008775999976529015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,1536,0.01000088867213991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,3584,0.013721778160995908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,3072,0.01256977766752243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,1024,0.007321777443091075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,1024,0.007620444728268518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,2560,0.011723555624485016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,768,0.01722666621208191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,768,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,768,0.007326222128338284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,2048,0.009591110878520543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,512,0.017831110292010836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,512,0.004505777938498391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,512,0.0069644442862934535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,1536,0.008255999949243333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,256,0.01679999960793389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,256,0.003839111162556542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,256,0.006565333240562015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,12288,0.04689866635534498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,1024,0.007143999967310164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,128,0.014814222852389017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,128,0.0035537779331207275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,128,0.006593777901596493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,768,0.006980444822046492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,64,0.003563555578390757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,64,0.006302222195598815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,10240,32,0.0038515557017591265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,10240,32,0.006635555376609166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,512,0.006572444405820634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,65536,0.19168710708618164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,65536,0.10468266407648723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,16384,0.051368001434538096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,16384,0.046479112572140165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,6144,0.032856888241238065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,256,0.006247111078765657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,16384,0.031018667750888403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,12288,0.03602044449912177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,12288,0.04247466723124186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,12288,0.025210666987631056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,10240,0.037575112448798284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,10240,0.03030222323205736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,10240,0.022664000590642292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,10240,128,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,8192,0.03432622220781114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,8192,0.025343111819691126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,8192,0.01917155583699544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,3584,0.02682400080892775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,65536,0.08018310864766438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,7168,0.022427555587556627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,7168,0.017644445101420086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,16384,0.025637333591779072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,6144,0.030227555168999568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,12288,0.020415110720528495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,6144,0.019720888800091214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,6144,0.016321778297424316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,5120,0.027786665492587622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,5120,0.01700622174474928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,5120,0.014750222365061441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,10240,0.018174222773975797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,4096,0.026182222697469924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,4096,0.014532445205582513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,4096,0.012455999851226807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,8192,0.015832889411184523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,3584,0.024978665841950312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,3584,0.012865777644846173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,3584,0.01202666676706738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,10240,1024,0.01959822244114346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,3072,0.024488000406159297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,7168,0.014460444450378418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,3072,0.011796444654464722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,3072,0.011054221954610614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,2560,0.023174222972657945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,2560,0.010481778118345471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,2560,0.010426667001512317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,6144,0.01346666696998808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,2048,0.022115555074479844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,2048,0.009184000392754873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,2048,0.009535111486911774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,5120,0.012434666355450949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,4096,0.01127911110719045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,1536,0.007851555943489075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,1536,0.007346666521496243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,3584,0.01036977767944336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,1024,0.01744888888465034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,1024,0.006545777950021956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,3072,0.009501333038012186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,1024,0.006618666566080517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,768,0.016865778300497267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,768,0.00490399988161193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,768,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,2560,0.008350222474998897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,512,0.01686844395266639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,512,0.00424177779091729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,512,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,2048,0.007601777712504069
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,256,0.01650311052799225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,256,0.0038177776667806837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,256,0.006299555715587404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,1536,0.006986666884687211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,128,0.014753777119848462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,128,0.0038480001191298165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,128,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,1024,0.006588444527652528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,64,0.003504000190231535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,64,0.006252444452709622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,8192,32,0.00369777778784434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,8192,32,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,768,0.00591111100382275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,512,0.005920889063013925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,65536,0.16565599706437853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,65536,0.141884446144104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,65536,0.14613155523935953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,16384,0.04891555507977804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,65536,0.09706133604049683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,16384,0.04551466637187534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,16384,0.029189334975348577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,12288,0.04074577821625604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,12288,0.035537779331207275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,256,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,12288,0.0236151119073232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,10240,0.03638489047686259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,10240,0.02976622184117635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,10240,0.020899555749363367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,8192,128,0.0058702222175068325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,8192,0.03329955538113912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,8192,0.025568889247046575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,8192,0.018435556027624343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,65536,0.07771555582682292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,7168,0.030243555704752605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,7168,0.02251200046804216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,16384,0.024901333782407973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,7168,0.016851555969980028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,6144,0.028743111424975928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,12288,0.020103999310069613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,6144,0.019863110449579027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,6144,0.015617777903874716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,5120,0.01697688963678148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,5120,0.01408088869518704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,10240,0.017649778061442904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,7168,0.031607998741997614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,4096,0.025457777910762366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,4096,0.015185778339703878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,4096,0.012339555554919772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,8192,0.0158915552828047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,3584,0.012349333200189801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,3584,0.011239111423492432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,7168,0.01404888927936554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,3072,0.023563555545277063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,3072,0.011047999891969891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,6144,0.01314755529165268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,3072,0.01070577816830741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,2560,0.02237155536810557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,2560,0.009717333647939894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,5120,0.01203733351495531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,2560,0.010243555737866296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,2048,0.02083822256988949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,2048,0.0087333329849773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,2048,0.008332444561852349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,4096,0.01094133324093289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,1536,0.017906667457686532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,1536,0.007606222397751278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,1536,0.007277333074145847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,3584,0.01020088874631458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,1024,0.016897777716318767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,1024,0.005761777775155173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,1024,0.006555555595291986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,8192,1536,0.02018044392267863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,768,0.016892444756295946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,768,0.004857777721352047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,768,0.0062977779242727495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,3072,0.008719999757077958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,512,0.016174222032229107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,512,0.004226666771703296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,2560,0.007939555578761632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,512,0.006263999889294307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,256,0.01622222198380364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,256,0.00406577769252989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,2048,0.006980444822046492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,256,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,128,0.014825777875052558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,128,0.0038924444880750445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,128,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,1536,0.006744888921578725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,64,0.003522666792074839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,64,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,7168,32,0.00388355553150177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,7168,32,0.005998222364319696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,1024,0.006298666612969504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,65536,0.1340764496061537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,768,0.00591111100382275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,65536,0.1484017769495646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,65536,0.08762222528457642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,16384,0.04677155613899231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,16384,0.036658667855792575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,16384,0.0277164445983039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,512,0.005895111295912001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,12288,0.038662221696641706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,12288,0.02927288744184706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,12288,0.02200711104604933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,256,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,10240,0.03460266523891025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,10240,0.024519999821980793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,10240,0.019683554768562317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,7168,128,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,8192,0.02124799953566657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,8192,0.03182844320933024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,8192,0.01735733283890618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,65536,0.07443733347786798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,7168,0.018512000640233357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,7168,0.015787555111779105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,16384,0.023914666639433965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,6144,0.027513777216275532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,12288,0.01958488921324412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,6144,0.0162124451663759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,6144,0.015049777097172208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,5120,0.025944888591766357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,5120,0.013775111072593264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,10240,0.01705422169632382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,5120,0.013491555220550962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,8192,0.015275556180212231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,4096,0.024264888630972967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,4096,0.012737777498033313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,4096,0.011696000066068439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,7168,0.013726222018400827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,3584,0.022963555322753057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,3584,0.011177777416176267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,3584,0.01129155523247189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,6144,0.012879999975363413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,3072,0.022317333353890315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,3072,0.010080888867378235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,3072,0.010374222364690568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,5120,0.011717333561844297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,5120,0.027061333258946735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,2560,0.009131555755933126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,2560,0.009303111169073317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,4096,0.010326222413116032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,7168,3584,0.024033778243594702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,2048,0.007914666500356462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,2048,0.007916444705592262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,3584,0.00923999978436364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,1536,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,1536,0.006629333313968446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,3072,0.008230222596062554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,1024,0.016393777396943834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,1024,0.005184000151024925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,2560,0.007301333049933116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,1024,0.006539555473460092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,768,0.01626755628320906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,768,0.004575110971927643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,768,0.00600000015563435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,2048,0.0070879997478591064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,1536,0.006568000134494569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,512,0.0038817777401871155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,512,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,1024,0.006072000082996156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,256,0.016531555189026725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,256,0.0035626664757728577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,256,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,768,0.005993777679072485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,128,0.014467555615637036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,128,0.003261333331465721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,128,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,512,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,64,0.0032257777121331957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,64,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,6144,32,0.0031902222997612427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,6144,32,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,256,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,65536,0.12409778436024983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,6144,128,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,65536,0.07996088928646512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,65536,0.07253155443403456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,16384,0.039397332403394915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,7168,0.029077334536446467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,16384,0.025569778349664476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,12288,0.03743288914362589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,12288,0.031167109807332356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,12288,0.020842666427294414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,16384,0.022999111149046157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,10240,0.033174223370022245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,10240,0.022375111778577168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,10240,0.018737778067588806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,12288,0.018937778141763475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,2048,0.0199626667631997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,2560,0.020968889196713764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,8192,0.019197333190176222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,8192,0.016688000824716356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,7168,0.028367112080256145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,1536,0.017495999733606975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,7168,0.01681866745154063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,7168,0.015055111712879606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,10240,0.016513778103722464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,6144,0.026701332794295415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,6144,0.014881778094503613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,6144,0.014230221509933472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,8192,0.01461955573823717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,5120,0.013193777865833707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,5120,0.013046222428480784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,7168,0.013242666920026144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,4096,0.0233608881632487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,6144,512,0.015438222222858004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,6144,0.012233778006500669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,4096,0.011733333269755045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,4096,0.011233777635627322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,3584,0.02232711182700263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,3584,0.01036622209681405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,3584,0.010598222414652506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,3072,0.02142488956451416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,3072,0.009189333352777693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,3072,0.009655999640623728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,5120,0.011026666396194033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,2560,0.01919911139541202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,2560,0.00832622249921163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,4096,0.009083555804358589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,2560,0.008455110920800103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,2048,0.01749955614407857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,3584,0.008279110822412703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,2048,0.0074746666683091064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,2048,0.007128000259399414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,3072,0.007613333563009898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,1536,0.01681600014368693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,1536,0.005910222315125995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,1536,0.007264888948864407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,2560,0.007327111230956183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,1024,0.004567999806669024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,1024,0.006584888945023219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,2048,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,65536,0.12883822123209634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,768,0.004204444587230682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,1536,0.006506666541099548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,768,0.006285333385070165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,512,0.015117333994971381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,512,0.0041928887367248535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,512,0.005946666830115848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,1024,0.006265777680608962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,256,0.016470222009552848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,256,0.003528000166018804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,16384,0.044453332821528115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,256,0.006221333311663733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,768,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,128,0.003504000190231535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,128,0.005910222315125995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,512,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,64,0.003061333257291052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,64,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,5120,32,0.003173333282272021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,5120,32,0.005910222315125995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,65536,0.10481244325637817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,65536,0.09866399897469415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,256,0.005616000129116907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,65536,0.07365866502126057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,5120,128,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,16384,0.031467556953430176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,16384,0.023753777146339417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,65536,0.06985333230760363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,12288,0.03257155418395996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,12288,0.025703110628657874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,12288,0.01981244484583537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,8192,0.030639999442630347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,16384,0.022223111655977037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,10240,0.01924622224436866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,10240,0.02915466825167338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,10240,0.017959111266665988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,8192,0.027120888233184814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,8192,0.016488000750541687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,8192,0.016028443972269695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,5120,0.024999111890792847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,7168,0.025171554750866357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,7168,0.014394667413499622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,7168,0.014800000521871777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,12288,0.018056000272432964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,6144,0.024243555135197107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,6144,0.013054221868515015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,6144,0.013580444786283704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,10240,0.016069332758585613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,8192,0.014024888475735983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,5120,0.011405333048767514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,5120,0.012396444049146442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,7168,0.012883555557992725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,4096,0.021578666236665513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,4096,0.010496888723638324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,4096,0.010728889041476779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,6144,0.011381333072980246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,3584,0.019479110836982727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,3584,0.009371555513805812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,3584,0.009340444372759925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,5120,0.010000000397364298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,3072,0.018197332819302876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,3072,0.008615111311276754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,3072,0.008282666405042013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,4096,0.008387555678685507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,3584,0.00794044468137953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,2560,0.007997333175606197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,2560,0.007831110722488826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,3072,0.007616889145639207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,2560,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,768,0.01646933290693495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,2048,0.006783111227883234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,2048,0.0069057775868309875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,1024,0.015807999504937064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,1536,0.005469333380460739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,1536,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,2048,0.006713777780532837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,1024,0.015166223049163818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,1024,0.004239999999602635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,1024,0.006286222073766921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,5120,128,0.014818666709793938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,768,0.016191111670600045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,1536,0.006325333482689328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,768,0.004181333300140169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,768,0.005929777605666055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,1024,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,512,0.003859555555714501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,512,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,768,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,16384,0.03762222329775492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,256,0.003504000190231535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,256,0.0058853332367208265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,512,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,256,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,128,0.003531555748648114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,128,0.005879999862776862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,5120,0.022460444105996027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,64,0.0031600000543726813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,64,0.005953777581453323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,4096,128,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,4096,32,0.003294222263826264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,4096,32,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,65536,0.08643200000127156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,65536,0.10414399703343709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,65536,0.07162844472461276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,16384,0.037059555451075234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,16384,0.029076443778143987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,16384,0.023152889476882085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,2560,0.017429333594110277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,12288,0.032093332873450384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,12288,0.02460177739461263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,65536,0.06854133473502265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,12288,0.019215111931165058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,10240,0.028811554114023846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,10240,0.021702221698231165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,10240,0.017800889081425138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,16384,0.021712000171343487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,8192,0.026331555512216356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,8192,0.015828443898095023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,8192,0.01536266671286689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,2048,0.01637066735161675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,12288,0.017921777235137094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,7168,0.0247013337082333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,7168,0.014070222775141398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,7168,0.014191110928853353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,6144,0.023603555228975084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,10240,0.015824889143308003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,6144,0.012404444317022959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,6144,0.013383111192120446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,1536,0.015453333655993143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,5120,0.0221368885702557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,5120,0.011033777561452655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,5120,0.012415111064910889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,4096,0.020149333609475028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,4096,0.010307555397351583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,8192,0.013791110780504016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,4096,0.010041777458455827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,3584,0.01887999971707662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,3584,0.009000889129108852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,3584,0.008999110923873054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,7168,0.012717333104875354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,3072,0.01756266587310367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,6144,0.011033777561452655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,3072,0.00831733312871721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,3072,0.008427555362383524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,2560,0.01680355601840549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,5120,0.010045333041085137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,2560,0.0074240002367231585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,2560,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,4096,0.008300444318188561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,2048,0.00601333338353369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,2048,0.00693866651919153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,3584,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,3072,0.007644444704055786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,1536,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,1536,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,2560,0.007261333366235097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,512,0.014163555370436775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,1024,0.004279110994603899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,1024,0.01482488877243466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,1024,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,768,0.016099555624855887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,2048,0.0069066666894488865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,768,0.003916444463862313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,768,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,512,0.013783999615245394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,512,0.0038488888078265717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,256,0.015117333994971381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,512,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,256,0.01477688882086012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,256,0.0034871109657817413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,256,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,1536,0.006616000086069107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,128,0.014162666267818876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,1024,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,128,0.003546666767862108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,128,0.005962666538026597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,64,0.0029111111329661477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,768,0.005709333138333426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,64,0.005922666854328579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3584,32,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3584,32,0.0031946665710873077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,512,0.005547555370463266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,65536,0.07806844181484647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,65536,0.10130400127834743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,65536,0.06956622335645887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,16384,0.036835554573271014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,256,0.005557333429654439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,16384,0.02550044490231408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,16384,0.022572444544898138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,12288,0.021607110897699993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,12288,0.030985779232449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,12288,0.01886755559179518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3584,128,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,10240,0.028142223755518597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,10240,0.018582221534517076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,10240,0.017162667380438913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,65536,0.06749333275689019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,8192,0.025731555289692346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,16384,0.021597334080272253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,8192,0.016880000631014507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,8192,0.014988443917698331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,7168,0.024200000696712073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,7168,0.01332711097266939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,12288,0.017514665921529133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,7168,0.013789333403110504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,6144,0.022967111733224656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,6144,0.011747555600272285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,6144,0.01298577752378252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,10240,0.015488889482286243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,4096,128,0.014814222852389017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,5120,0.010599111517270407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,8192,0.01293155550956726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,5120,0.011696000066068439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,4096,0.018592889110247295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,4096,0.009668444593747457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,4096,0.009297777381208207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,7168,0.011747555600272285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,3584,0.008603555460770925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,3584,0.017664889494578045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,3584,0.008488888541857401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,3072,0.01682400041156345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,6144,0.010064000056849586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,3072,0.007963555554548899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,3072,0.007682667010360294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,2560,0.016140444411171805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,2560,0.006969778074158563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,2560,0.007611555357774098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,5120,0.009292444421185387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,4096,0.008336000144481659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,2048,0.006180444525347815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,2048,0.006999111009968652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,3584,0.008001777860853408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,1536,0.014808000789748298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,1536,0.0052177777720822235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,1536,0.006637333167923822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,3072,0.007615110940403408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,2560,0.007250666618347168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,1024,0.00452977791428566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,1024,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,2048,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,1536,0.00627377794848548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,768,0.0041804446114434134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,768,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,1024,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,512,0.013408888545301227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,512,0.0038399998512532976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,512,0.005953777581453323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,768,0.0058755555914507965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,256,0.015106666419241162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,256,0.0035111109415690103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,256,0.0058782220714622075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,512,0.005547555370463266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,256,0.00555288874440723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,128,0.0031973332580592898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,128,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,2048,0.015463999576038785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,64,0.0032151111712058387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,64,0.005940444353553984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,3072,32,0.003216000066863166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,3072,128,0.005543111099137201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,3072,32,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3584,1536,0.015169777803950839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,65536,0.06532533301247491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,65536,0.06737866666581896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,16384,0.03594400154219733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,16384,0.022996443841192458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,16384,0.021914665897687275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,65536,0.06645066870583428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,16384,0.021290666527218286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,12288,0.01958755486541324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,12288,0.018601778480741713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,12288,0.017352888981501263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,10240,0.027730666928821143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,10240,0.017265778448846605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,10240,0.016731555263201397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,10240,0.015132443772421943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,5120,0.02068622244728936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,8192,0.015156444576051501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,8192,0.01481599940194024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,8192,0.012276444170210095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,7168,0.023476445012622412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,2048,0.0155395550860299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,7168,0.014141332772043018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,7168,0.013457777599493662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,6144,0.02223288847340478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,1024,0.014448000325096978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,6144,0.013390222357379066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,6144,0.012714666624863943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,5120,0.019833778341611225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,5120,0.010535111029942831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,5120,0.010688888529936472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,7168,0.011130666567219628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,6144,0.010038221875826517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,4096,0.01811022228664822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,4096,0.009338666995366415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,768,0.016507556041081745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,4096,0.00850844466023975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,5120,0.009141333401203156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,3584,0.008316444853941599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,3584,0.008579555484983657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,4096,0.00815288888083564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,3072,0.01663733356528812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,3072,0.007522666619883643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,3072,0.007666666474607255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,3072,128,0.014454222387737699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,3584,0.007986666427718269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,2560,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,2560,0.007325333025720384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,2048,0.015180443723996481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,2048,0.00591111100382275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,2048,0.006966222491529252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,3072,0.007320889168315464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,2560,0.007327111230956183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,1536,0.005234666582610872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,1536,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,2048,0.006591111007663939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,65536,0.09923377964231704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,1024,0.00453244439429707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,1024,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,1536,0.006414222220579783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,768,0.015524443652894763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,768,0.003887999802827835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,768,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,12288,0.03046400017208523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,1024,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,512,0.01313244468635983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,512,0.003859555555714501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,512,0.005965333431959152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,256,0.014416888356208801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,256,0.003206222214632564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,768,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,256,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,512,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,128,0.0031644445326593188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,128,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,256,0.005466666486528184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,64,0.0031999999450312722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,64,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2560,32,0.0032177778581778207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2560,32,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,65536,0.09164355860816108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,65536,0.05526311198870341
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2560,128,0.005562666803598404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,65536,0.06490133206049602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,16384,0.034007999632093645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,16384,0.019889778561062284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,16384,0.021280889709790547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,8192,0.02533066603872511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,12288,0.029081778393851385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,65536,0.06491200129191081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,12288,0.015847999188635085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,12288,0.017997332745128207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,10240,0.02643377747800615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,10240,0.01479288935661316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,10240,0.016466667254765827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,16384,0.020865778128306072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,8192,0.024166221419970196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,8192,0.014177777700954013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,8192,0.01407911048995124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,12288,0.016895999511082966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,7168,0.02264533274703556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,7168,0.01274400038851632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,7168,0.013168888787428537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,10240,0.014132445057233175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,6144,0.020921778347757127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,6144,0.011343999869293638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,6144,0.011695110963450538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,8192,0.01146399974822998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,5120,0.018574222922325134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,5120,0.010391111175219217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,5120,0.010372444159454769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,7168,0.010374222364690568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,4096,0.01735111077626546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,4096,0.008931555681758458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,4096,0.00832711077398724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,6144,0.009655999640623728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,3584,0.01680888897842831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,3584,0.008216888540320927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,3584,0.007954667011896769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,5120,0.009001777403884465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,3072,0.016136889656384785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,3072,0.007308444215191736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,3072,0.007660444411966536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,4096,0.008275555239783393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,3584,0.007670222057236566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,2560,0.0063662222690052455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,2560,0.007612444460391998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,3072,0.007646222081449296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,2048,0.014498665928840637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,2048,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,2048,0.006676444576846228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,2560,0.00703111125363244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,2048,0.006647111227114995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,1536,0.01441511180665758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,1536,0.004868444469239977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,1536,0.006911110960774952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,3584,0.01686755485004849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,1024,0.013436444103717804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,1024,0.0042408886882993905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,1024,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,1536,0.006361777583758037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,768,0.0038933331767717996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,768,0.006296000132958095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,2560,0.015805333852767944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,1024,0.005953777581453323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,512,0.0035502223504914176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,512,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,768,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,512,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,256,0.0031928889867332247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,256,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,1536,0.014502222339312235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,128,0.014811555544535318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,128,0.002986666642957263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,128,0.005940444353553984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,1024,0.014351111319330005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,64,0.002833777831660377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,64,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,2048,32,0.002847111059559716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,256,0.0053546664615472155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,2048,32,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,65536,0.04351199997795952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,65536,0.08993066681755914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,2048,128,0.0052595556610160405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,65536,0.06314755810631646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,16384,0.033589333295822144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,16384,0.01574222246805827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,16384,0.02091377807988061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,12288,0.02910311023394267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,12288,0.014255111416180929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,65536,0.06387288702858819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,12288,0.017873777283562552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,16384,0.02033866610791948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,10240,0.013047999805874295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,10240,0.015804444750150044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2560,128,0.014084445105658637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,12288,0.015215999550289579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,8192,0.022498667240142822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,8192,0.011711999773979187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,8192,0.013209777573744455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,7168,0.02087466584311591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,7168,0.01127022256453832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,7168,0.011679111255539788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,10240,0.012693333129088083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,6144,0.01923199991385142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,6144,0.010451555252075195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,8192,0.011048888994587792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,6144,0.010039110978444418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,5120,0.018205333087179396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,5120,0.009657777845859528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,5120,0.009641778137948778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,7168,0.010253333383136326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,4096,0.01682400041156345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,4096,0.008658666577604082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,6144,0.009350222018029954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,4096,0.008329778081840938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,5120,0.008641777767075432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,3584,0.0074515557951397365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,3584,0.008286221987671321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,4096,0.008086221913496653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,3072,0.015793777174419828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,3072,0.0069395556218094295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,3072,0.007635555333561367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,3584,0.00795377790927887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,3072,0.00739555557568868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,2560,0.005919111271699269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,2560,0.007306666837798224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,2560,0.007251555720965068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,2048,0.014471999473041959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,2048,0.005261333452330695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,2048,0.0069422221018208405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,2048,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,1536,0.013870221873124441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,1536,0.004928888960017098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,1536,0.00627377794848548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,1536,0.006340444501903322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,1024,0.013099555340078143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,1024,0.004267555558019214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,1024,0.006284444282452266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,2560,0.015563555889659457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,768,0.01550044450494978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,768,0.0038693332009845306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,768,0.006221333311663733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,1024,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,512,0.012787555654843649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,512,0.003539555602603488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,512,0.005959110955397288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,768,0.0058559998869895935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,512,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,256,0.0032248888164758682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,256,0.005693333430422678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,768,0.015802666544914246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,512,0.013071999781661563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,256,0.00554933316177792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,128,0.0031831111345026228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,128,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,64,0.0028577778074476454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,64,0.005924444645643234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1536,32,0.0028568889117903183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1536,128,0.005390222287840313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1536,32,0.005687110953860813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,65536,0.03219822380277846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,65536,0.08885688914193048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,65536,0.06148711178037855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,16384,0.032792889409595065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,2048,256,0.014464888307783337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,16384,0.013445333474212222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,16384,0.02015111181471083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,65536,0.0633573333422343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,12288,0.02756533357832167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,12288,0.013958222336239286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,12288,0.016731555263201397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,16384,0.01850133306450314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,10240,0.012768000364303589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,10240,0.014463999205165438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,12288,0.013791110780504016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,8192,0.02160088883505927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,8192,0.011391110718250275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,8192,0.012080888781282636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,10240,0.012073777616024017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,8192,0.011008000208271874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,7168,0.010447111394670276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,7168,0.01072355525361167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,7168,0.009708444277445475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,10240,0.02591200007332696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,6144,0.019093333019150626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,6144,0.010032888915803697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,6144,0.009813333551088968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,5120,0.017827555537223816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,5120,0.00868088917599784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,5120,0.009299555586444007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,6144,0.009321778184837764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,4096,0.01686755485004849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,4096,0.007666666474607255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,4096,0.008310221963458592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,5120,0.008660444782839881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,3584,0.015834665960735746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,3584,0.016149333781666227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,3584,0.007246221933099959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,3584,0.00776444458299213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,4096,0.008002666963471307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,3072,0.006634666687912411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,3072,0.007670222057236566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,3584,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,2560,0.014767999450365702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,2560,0.0059368887709246735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,2560,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,3072,0.007328000333574083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,2048,0.014495111174053617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,2048,0.005274666680230035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,2048,0.006955555743641323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,2560,0.0069617778062820435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,2560,0.015511110424995422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,1536,0.013456888496875763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,1536,0.004567111117972268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,1536,0.006342222293217977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,2048,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,1024,0.004008000095685323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,1024,0.006274666637182236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,1536,0.006229333165619109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,768,0.015459555718633862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,768,0.003572444534964032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,768,0.006247111078765657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,1024,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,768,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,512,0.0035137778355015647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,512,0.005919111271699269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,512,0.00553688903649648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,256,0.01477599971824222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,256,0.003132444495956103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,256,0.005736888696750005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,256,0.005281777845488654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,128,0.014436443646748861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,128,0.0028906667398081887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,128,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,1024,128,0.00527555536892679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,64,0.0028159999185138275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,64,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,1024,32,0.0028248888750871024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,1024,32,0.005910222315125995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,256,0.015075556106037565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,65536,0.088082664542728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,65536,0.02760355505678389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,65536,0.06083910995059543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,65536,0.06293155749638875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,16384,0.013654222091039022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,16384,0.02020711037847731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,16384,0.01682044400109185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1536,128,0.014104000396198697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,12288,0.012076444096035428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,12288,0.015792000624868605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,10240,0.024079999989933435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,10240,0.023016000787417095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,10240,0.011305777562989129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,12288,0.013800000150998434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,10240,0.013451555536852943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,8192,0.021207999851968553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,8192,0.010212444596820408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,8192,0.011053333679835001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,10240,0.012092444631788464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,7168,0.01995111174053616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,7168,0.009696000152164036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,7168,0.01036000003417333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,8192,0.011024889018800525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,7168,0.009711999860074785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,6144,0.008999110923873054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,6144,0.009777777724795872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,6144,0.009022222624884712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,5120,0.01759377784199185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,3072,0.015818667080667283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,5120,0.008029333419269985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,5120,0.009303111169073317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,4096,0.016876444220542908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,4096,0.007649777664078607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,4096,0.008259555531872643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,5120,0.008570666942331526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,3584,0.01553600033124288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,3584,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,3584,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,4096,0.008010666403505538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,3584,0.007643555601437886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,3072,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,3072,0.007613333563009898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,1024,0.012755555411179861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,2560,0.01480977733929952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,2560,0.005758222192525864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,2560,0.007248000138335758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,3072,0.007301333049933116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,2048,0.014144889182514615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,2048,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,2048,0.006897777732875612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,2560,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,1536,0.013456000222100152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,1536,0.00453599997692638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,2048,0.006594666590293248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,1536,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,1024,0.01311377767059538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,1024,0.0038764443662431505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,1024,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,1536,0.006263111200597551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,1024,0.005881777654091517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,768,0.003842666745185852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,768,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,1024,512,0.012784889174832238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,512,0.01276000009642707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,512,0.003568888952334722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,512,0.005903111149867375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,768,0.00555288874440723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,256,0.01408622165520986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,256,0.0031555555760860443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,256,0.005621333503060871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,512,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,128,0.014149333039919535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,128,0.0028088889602157804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,128,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,256,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,64,0.0028213332924577924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,64,0.005554666535721884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,768,32,0.002826666666401757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,768,32,0.0058577776783042485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,768,128,0.0052986666560173035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,65536,0.021912889348136053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,65536,0.08756355444590251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,65536,0.0596746669875251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,16384,0.030092444684770372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,16384,0.011411555939250521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,65536,0.06271377536985609
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,16384,0.01772622267405192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,12288,0.025760889053344727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,12288,0.00981866651111179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,16384,0.016585777203241985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,12288,0.01496088835928175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,12288,0.013796444568369122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,10240,0.010103999740547603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,10240,0.012432888978057437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,16384,0.031721777386135526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,8192,0.020584000481499564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,8192,0.008641777767075432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,10240,0.012075555821259817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,12288,0.02674399978584713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,8192,0.011521778172916837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,7168,0.00813066628244188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,7168,0.010671111444632212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,8192,0.010759111079904767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,7168,0.009826666778988307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,6144,0.007616000043021307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,6144,0.009704888694816166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,6144,0.00906844437122345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,7168,0.01956355571746826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,5120,0.008007110820876228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,5120,0.009153777526484596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,6144,0.018588445252842374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,5120,0.008456000023418004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,4096,0.007602666815121968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,4096,0.008018666671382057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,3072,0.015750222735934787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,4096,0.007962666451931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,3584,0.007263999846246507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,3584,0.0076657773719893555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,3584,0.007619555625650618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,3072,0.014882667197121514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,3072,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,3072,0.007589333587222629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,3072,0.007267555428875818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,2560,0.014098667436175875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,2560,0.0058853332367208265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,2560,0.007305777735180325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,2560,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,2048,0.013774221969975365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,2048,0.005220444252093633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,2048,0.006693333387374878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,2048,0.006468444648716185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,1536,0.012832889126406776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,768,768,0.0148044443792767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,1536,0.004573333180612988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,1536,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,1024,0.0124195557501581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,1024,0.0038746665749284956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,1536,0.006244444598754247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,1024,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,768,0.014472888575659858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,768,0.0035386664999855887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,1024,0.005720888988839255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,768,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,512,0.012098666694429187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,512,0.003181333343187968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,768,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,512,0.005623111294375525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,256,0.013807999591032663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,256,0.003175111073586676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,512,0.0052391112678580815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,256,0.006063111126422882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,256,0.0052586665583981415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,128,0.0028595555987623003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,128,0.005472888963090048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,512,128,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,64,0.00253866674999396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,64,0.005906666732496685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,512,32,0.002591111179855135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,512,32,0.0055440002017551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,65536,0.08681955602433945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,65536,0.017086222767829895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,7168,0.019384000036451552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,10240,0.022660444180170696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,65536,0.05910044246249729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,6144,0.017895110779338412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,16384,0.00998933364947637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,16384,0.016176000237464905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,65536,0.06172177526685926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,16384,0.016505777835845947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,12288,0.00832622249921163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,12288,0.013799111048380533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,10240,0.021957332889238994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,12288,0.013743110828929476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,10240,0.008972444468074376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,10240,0.012693333129088083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,8192,0.020266667008399963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,10240,0.011857777833938599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,8192,0.00793688909875022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,8192,0.011032889286677042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,7168,0.018435556027624343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,8192,0.011006222003036074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,7168,0.007679111427730984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,7168,0.01056177748574151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,6144,0.017877333694034152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,5120,0.017183999220530193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,7168,0.009664889011118147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,6144,0.007293333609898885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,6144,0.010006222460005019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,5120,0.016813332835833233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,6144,0.008980444735950893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,5120,0.00793422261873881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,5120,0.008972444468074376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,4096,0.007290666302045186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,5120,0.008615111311276754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,4096,0.00830844458606508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,3584,0.015213333898120456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,4096,0.00793866647614373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,3584,0.006960888703664144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,3584,0.007640000018808577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,3072,0.014485332700941296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,3584,0.007641777396202087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,3072,0.006640000061856375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,3072,0.00739733295308219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,2560,0.014415999253590902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,3072,0.007266666326257918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,2560,0.006339555399285422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,2560,0.007261333366235097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,2048,0.013082666529549493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,2560,0.0069368887278768755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,2048,0.006232000059551663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,2048,0.0069057775868309875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,2048,0.006616888774765863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,1536,0.004503111044565837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,1536,0.006239999913507038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,1536,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,1024,0.012051555845472546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,1024,0.0038533334930737815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,1024,0.006231110956933763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,1024,0.005704000178310607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,768,0.014087110757827759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,768,0.0035626664757728577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,768,0.00566133318675889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,768,0.005567111074924469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,3584,0.01552177800072564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,512,0.003208000005947219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,512,0.005913777897755305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,4096,0.015488000379668342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,512,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,256,0.002830222249031067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,256,0.005960000058015187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,256,0.005475555443101459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,128,0.011698666546079846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,128,0.0028071111689011254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,128,0.005782222168313132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,256,128,0.0055440002017551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,64,0.002583111118939188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,64,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,256,32,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,256,32,0.0057795556883017225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,65536,0.0868871079550849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,65536,0.015775110986497667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,65536,0.05875911315282186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,65536,0.05997066365347969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,16384,0.02882222334543864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,16384,0.009383111364311641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,16384,0.01648888819747501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,16384,0.016600888636377122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,12288,0.024341333243581984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,12288,0.008175111479229396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,12288,0.013796444568369122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,12288,0.013750221994188098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,10240,0.02160444524553087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,10240,0.007639110916190677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,10240,0.012720888687504662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,10240,0.012071111136012606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,8192,0.020060444871584576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,8192,0.006979555719428592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,8192,0.011235555840863122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,8192,0.011056889262464313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,7168,0.018794667389657762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,512,128,0.012420444852775998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,7168,0.006615110983451207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,7168,0.010382222632567087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,7168,0.009646221995353699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,6144,0.006590222318967183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,6144,0.009993777506881291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,6144,0.008997333546479544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,5120,0.016141333513789706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,5120,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,12288,0.024666666984558105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,5120,0.009006222089131674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,5120,0.008328888979223039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,4096,0.006903999795516332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,4096,0.00794488853878445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,4096,0.007924444145626491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,3584,0.014124444789356656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,16384,0.02916533417171902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,3584,0.006599999964237213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,3584,0.00794488853878445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,3584,0.007596444752481248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,3072,0.014490667316648694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,3072,0.006567111031876669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,3072,0.007660444411966536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,2560,0.013787555197874704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,3072,0.007250666618347168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,2560,0.006629333313968446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,2560,0.006949333681000604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,2048,0.0123831108212471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,2560,0.006991110742092133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,2048,0.006226666685607698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,2048,0.0069360000391801195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,1536,0.012768000364303589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,2048,0.006592888798978593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,1536,0.004373333520359463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,1536,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,4096,0.015804444750150044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,1024,0.011403555671374003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,1536,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,1024,0.0038604442444112566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,1024,0.005946666830115848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,768,0.0035377778112888336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,1024,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,768,0.0058711109062035876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,768,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,512,0.0031866667171319327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,512,0.005943999936183293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,512,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,256,0.011740444435013665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,1536,0.012792000340090858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,256,0.002883555574549569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,256,0.005877333382765452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,256,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,128,0.011056889262464313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,128,0.0027946666296985415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,128,0.005960000058015187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,128,128,128,0.005469333380460739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,64,0.002536888958679305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,64,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,128,32,0.0028151110228565005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,128,32,0.005883555445406172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,65536,0.014537778165605334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,65536,0.057745774586995445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,16384,0.008043555749787224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,16384,0.01625333395269182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,12288,0.007600888609886169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,12288,0.014074666632546319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,10240,0.00721777809990777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,10240,0.012713777522246042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,8192,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,8192,0.0111191107167138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,7168,0.006611555400821898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,7168,0.010653333531485664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,6144,0.006575999988449945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,6144,0.009728000395827824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,5120,0.006208000083764394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,5120,0.009303111169073317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,4096,0.005647999958859549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,4096,0.008263111114501953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,3584,0.006330666856633292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,3584,0.007930667036109501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,3072,0.005943999936183293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,3072,0.007627555893527136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,2560,0.00629066675901413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,512,0.011075555450386472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,2560,0.007287999822033777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,2048,0.005962666538026597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,2048,0.006908444480763541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,1536,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,1536,0.004607111215591431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,1024,0.003846222327815162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,1024,0.00629333323902554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,768,0.0034968890249729156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,768,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,512,0.0031991110493739447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,256,0.002879111096262932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,512,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,128,0.0028239999794297745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,256,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,128,0.005600000007285013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,64,0.002491555487116178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,64,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,64,32,0.0028106667515304354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,64,32,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,65536,0.0147386673423979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,65536,0.0565262238184611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,16384,0.00794311116139094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,16384,0.016184000505341422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,12288,0.007701333198282454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,12288,0.013735111388895245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,10240,0.007370666497283512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,10240,0.012455110748608908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,8192,0.006577777779764599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,8192,0.011056889262464313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,7168,0.0068986668354935115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,7168,0.010446222292052375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,6144,0.006604444649484422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,6144,0.009638222555319468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,5120,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,5120,0.009019555317031013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,4096,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,4096,0.008018666671382057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,3584,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,3584,0.00794222205877304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,3072,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,3072,0.007592889169851939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,2560,0.006572444405820634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,2560,0.007258666886223688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,2048,0.005953777581453323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,2048,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,1536,0.004256000121434529
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,1536,0.006611555400821898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,1024,0.0038622220357259116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,1024,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,768,0.0037537775933742523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,768,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,256,256,0.011752000285519494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,512,0.003346666693687439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,512,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,256,0.002814222127199173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,256,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,128,0.0028391112056043413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,128,0.005616000129116907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,64,0.002528000002106031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,64,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,128,32,32,0.002536888958679305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,128,32,32,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,16384,0.3727733294169108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,16384,0.23716889487372506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,16384,0.17926132678985596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,12288,0.19047378169165718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,12288,0.2788453367021349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,12288,0.1339084439807468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,10240,0.16489688555399576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,10240,0.21437244945102266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,4096,0.015542222393883599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,6144,0.016926222377353244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,10240,0.11367021666632758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,8192,0.14555022451612684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,8192,0.17387378215789795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,8192,0.09482400284873115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,7168,0.15324532985687256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,512,0.011013333168294696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,7168,0.08060889111624824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,128,128,768,0.01348355578051673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,6144,0.1213102208243476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,6144,0.13026844130622015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,6144,0.07102400064468384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,5120,0.10801866981718276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,5120,0.11020177602767944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,5120,0.059677335951063365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,7168,0.13115289476182726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,4096,0.09728178050782944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,4096,0.09193066755930583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,4096,0.04857688811090258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,16384,0.13696355289883083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,3584,0.09136888715955947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,3584,0.07963111003239949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,3584,0.043324444029066295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,12288,0.10212177700466579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,3072,0.08465155627992417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,3072,0.06921066840489705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,3072,0.03825866513782077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,10240,0.0856284432941013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,2560,0.07864177889294095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,2560,0.05810933642917209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,2560,0.03292266527811686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,8192,0.07227200269699097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,2048,0.07426400317086114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,2048,0.04792266752984789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,2048,0.028813332319259644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,6144,0.057499554422166615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,7168,0.060752001073625356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,1536,0.06805511315663655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,1536,0.03739022215207418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,1536,0.02310577697224087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,5120,0.04852799905671013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,1024,0.028520888752407495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,1024,0.019704000817404855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,1024,0.062077330218421094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,768,0.03016888764169481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,4096,0.04059022333886888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,768,0.021260443660948012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,768,0.017204445269372728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,512,0.05688799752129448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,512,0.015127999915017022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,512,0.015507555670208402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,3584,0.03587199913130866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,256,0.02514133354028066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,256,0.009347555538018545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,256,0.013717333475748697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,3072,0.030897776285807293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,65536,128,0.0226275556617313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,128,0.008009778128729926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,128,0.010813333094120026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,2560,0.026995556222067937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,64,0.0069368887278768755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,64,0.011372444530328115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,65536,32,0.007186666958861881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,65536,32,0.012090666426552666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,2048,0.02336799932850732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,65536,0.3507244322035048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,65536,0.2342533270517985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,1536,0.0195795562532213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,65536,0.19263377454545763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,16384,0.08978755606545342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,16384,0.051671998368369214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,1024,0.016234666109085083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,12288,0.06167199876573351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,12288,0.06961777475145128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,12288,0.04047822289996677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,65536,0.15029422442118326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,10240,0.05470755696296692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,10240,0.05793155564202202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,10240,0.03534577621353997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,16384,0.042917334371142916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,8192,0.048320889472961426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,8192,0.04853689008288913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,8192,0.02996266550487942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,12288,0.03420889046457078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,7168,0.044509334696663745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,7168,0.04233422213130527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,7168,0.026517333255873785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,10240,0.029153777493370905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,6144,0.04184977875815498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,6144,0.03662755423121982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,6144,0.024422221713595923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,8192,0.025048888391918604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,16384,0.07607111003663805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,5120,0.03784888982772827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,5120,0.03178222311867608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,5120,0.020917332834667627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,768,0.014461333552996317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,4096,0.035587555832333036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,7168,0.02197333342499203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,4096,0.02681066592534383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,4096,0.01754844519827101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,3584,0.033534222178988986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,3584,0.023687111006842718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,3584,0.016303999556435477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,256,0.012051555845472546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,3072,0.03205688794453939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,3072,0.020796444680955675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,512,0.013108444710572561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,3072,0.014984000060293408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,2560,0.030545777744717065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,2560,0.01754755609565311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,2560,0.013776000175211163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,6144,0.020522667302025687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,2048,0.014831110835075378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,2048,0.012207110722859701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,5120,0.01761155492729611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,1536,0.027339556150966223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,1536,0.012191111014948951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,1536,0.011360888679822287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,4096,0.015790222419632804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,1024,0.025752888785468206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,65536,128,0.012143111063374413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,1024,0.009633777870072259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,1024,0.01017955525053872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,768,0.01984355515903897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,3584,0.014506666196717156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,768,0.008353778057628209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,768,0.008340444829728868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,512,0.021962665849261816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,512,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,512,0.007295110987292395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,3072,0.013556444810496436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,256,0.01722666621208191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,256,0.004541333350870344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,2560,0.01221244451072481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,256,0.006631111105283101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,128,0.015328889091809591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,128,0.003935111065705617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,128,0.006606222026877933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,2048,0.011564444336626263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,64,0.003560888684458203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,64,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,16384,32,0.0035546666218174826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,16384,32,0.007048000064161088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,1536,0.009759110709031424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,65536,0.17752355999416777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,65536,0.2770826551649305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,1024,0.008240000241332585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,768,0.007284444239404466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,65536,0.16421511438157824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,16384,0.0716257757610745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,512,0.0069644442862934535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,16384,0.04604533314704895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,12288,0.056335111459096275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,12288,0.03620355659061008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,256,0.006412444429265128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,10240,0.04484088884459602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,10240,0.046443555090162486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,16384,128,0.006640000061856375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,10240,0.031650665733549334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,8192,0.04122577773200141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,8192,0.03899111019240485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,8192,0.027489778068330552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,16384,2048,0.029286222325430974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,7168,0.03688177797529433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,7168,0.03427733315361871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,7168,0.02420444455411699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,65536,0.13646933767530653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,6144,0.034507556094063654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,6144,0.029768887493345473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,6144,0.02188355558448368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,16384,0.04040177663167318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,5120,0.031876444816589355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,5120,0.02565866708755493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,12288,0.03222844335767958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,5120,0.019465777609083388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,4096,0.030104888810051814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,4096,0.022100443641344707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,4096,0.016615110966894362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,10240,0.026641777820057336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,3584,0.028825776444541082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,3584,0.01921244462331136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,3584,0.015232000086042615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,8192,0.022915555371178523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,3072,0.02755644420782725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,3072,0.01719288859102461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,3072,0.014021333720948962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,7168,0.0206675562593672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,2560,0.02645422187116411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,2560,0.014411555396185981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,2560,0.013155555559529198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,6144,0.01902666687965393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,2048,0.02533333334657881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,2048,0.012607999973826937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,2048,0.011979555918110741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,5120,0.016691555579503376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,4096,0.015180443723996481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,1536,0.024325332707828943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,1536,0.01036800030204985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,1536,0.01054933336046007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,3584,0.014129777749379476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,3072,0.012938666674825879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,1024,0.008685333033402761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,1024,0.008741333252853816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,768,0.018228444788191054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,768,0.007264888948864407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,768,0.007908444437715743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,2560,0.0120328888297081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,512,0.019544000426928203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,512,0.005245333330498801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,512,0.007131555842028723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,16384,0.06079644627041287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,256,0.016855110724767048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,256,0.004259555704063839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,2048,0.010683555569913654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,256,0.006578666468461354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,12288,0.04991199904017978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,128,0.0038195554580953387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,128,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,1536,0.008462222086058723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,64,0.0032106666929192017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,64,0.006611555400821898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,12288,32,0.0035111109415690103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,12288,32,0.0069413334131240845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,1024,0.007613333563009898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,65536,0.1579128901163737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,65536,0.2131964498096042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,768,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,65536,0.15406489372253418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,16384,0.0555973317888048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,16384,0.043876445955700345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,512,0.00692622239391009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,12288,0.04554933309555054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,12288,0.043656001488367714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,256,0.006596444381607904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,12288,0.03478577733039856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,10240,0.04065422217051188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,10240,0.03654399845335219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,10240,0.030212445391549006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,12288,128,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,8192,0.03738311264250014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,8192,0.030605332718955144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,8192,0.02572000026702881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,65536,0.1313733392291599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,7168,0.03412266572316488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,7168,0.02695466743575202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,7168,0.023386667172114056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,16384,0.03886577818128798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,6144,0.03223377797338698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,6144,0.023736889163653057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,6144,0.020959110723601446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,12288,0.030347555875778198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,5120,0.029742221037546795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,5120,0.020795555578337777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,5120,0.018793778287039865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,10240,0.025799110531806946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,4096,0.02801777919133504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,4096,0.018044445249769423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,4096,0.016010666886965435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,8192,0.022316444251272414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,3584,0.026465778549512226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,3584,0.01574488812022739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,3584,0.014896889527638754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,7168,0.02035466664367252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,3072,0.02603911028967963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,3072,0.013789333403110504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,3072,0.013861333330472311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,6144,0.01793866687350803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,1024,0.02109866672092014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,2560,0.012494222157531314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,2560,0.012999999854299756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,5120,0.015999111864301894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,2048,0.02422222163942125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,2048,0.011010666688283285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,2048,0.011663999822404651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,4096,0.014690667390823364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,1536,0.02233244478702545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,1536,0.009296000003814697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,1536,0.00928177767329746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,3584,0.013427555561065674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,12288,128,0.015507555670208402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,1024,0.019588443968031142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,1024,0.007594666547245449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,1024,0.007985778152942657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,768,0.017886221408843994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,768,0.005963555640644497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,768,0.007669332954618666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,3072,0.012467555701732635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,512,0.01852888862291972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,512,0.004912888838185204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,512,0.006981333096822103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,2560,0.011459555890825061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,256,0.016883555385801528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,256,0.003920888735188378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,256,0.006616888774765863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,2048,0.009654222263230218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,16384,0.05630044142405192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,128,0.014838222000333997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,128,0.0038355555799272326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,128,0.00657066661450598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,1536,0.00795022232664956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,64,0.0035013332962989807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,64,0.006559111177921295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,10240,32,0.0035013332962989807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,10240,32,0.006626666833957036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,1024,0.007189333438873291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,65536,0.18180267016092935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,768,0.006622222148709827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,65536,0.14256710476345485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,65536,0.09876710838741726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,16384,0.049555556641684644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,16384,0.04774844315316942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,512,0.006616888774765863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,16384,0.03067644437154134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,12288,0.04166133205095927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,12288,0.037808888488345675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,12288,0.0247706671555837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,10240,0.03118488854832119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,10240,0.037414222955703735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,10240,0.022273777259720698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,8192,0.03456266721089681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,8192,0.026592888765864905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,8192,0.019201777047581144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,256,0.006244444598754247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,7168,0.03127377894189622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,7168,0.02347555591000451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,7168,0.017285333739386666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,10240,128,0.006254222244024277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,6144,0.029551111989551123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,6144,0.020419556233617995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,6144,0.016182222300105624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,65536,0.08265778091218737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,16384,0.025495110286606684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,5120,0.027463111612531874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,5120,0.017800889081425138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,5120,0.014627556006113688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,12288,0.020782222350438435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,4096,0.02604622145493825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,4096,0.015091554986106025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,4096,0.012659555508030785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,10240,0.01829155617290073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,3584,0.013468444347381592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,3584,0.024712888730896845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,3584,0.011720889144473605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,3072,0.024102222588327196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,3072,0.0120666664507654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,3072,0.010778666370444827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,8192,0.016208888755904306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,2560,0.022794667217466567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,2560,0.010793777803579966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,2560,0.010315555665228102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,7168,0.014759999182489185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,2048,0.02199555602338579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,2048,0.009263111485375298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,2048,0.009315555294354757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,6144,0.01348622226052814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,1536,0.0192266669538286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,1536,0.007982222570313348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,1536,0.0074888889988263445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,5120,0.012646222280131446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,1024,0.017906667457686532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,1024,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,4096,0.011556444068749746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,1024,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,768,0.017519111434618633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,768,0.004913777940803104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,3584,0.010501333408885531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,768,0.006629333313968446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,10240,2560,0.024711110525661047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,512,0.004219555606444676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,512,0.006265777680608962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,3072,0.009356444080670675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,2560,0.007972444097201029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,256,0.003549333247873518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,256,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,2048,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,128,0.01516711049609714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,128,0.0031893334041039148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,128,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,1536,0.006628444625271692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,64,0.0031662223239739737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,64,0.005966222120655908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,8192,32,0.003177777760558658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,1024,0.006285333385070165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,8192,32,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,768,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,65536,0.162754668129815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,512,0.0059075554211934405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,16384,0.048319998714658946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,65536,0.09432711203893025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,16384,0.043735110097461276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,16384,0.029493331909179688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,256,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,12288,0.039303110705481634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,12288,0.03456444541613261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,12288,0.023339556323157415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,8192,128,0.005903111149867375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,10240,0.03585955500602722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,10240,0.029093331760830347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,10240,0.020622221959961783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,65536,0.07738133271535237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,16384,0.025224000215530396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,8192,0.02457511093881395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,8192,0.017873777283562552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,12288,0.020611556039916146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,7168,0.02196088929971059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,7168,0.016528889536857605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,10240,0.017661333084106445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,6144,0.028571556011835735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,6144,0.019111999207072787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,6144,0.015433778365453085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,8192,0.015786666009161208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,7168,0.01407999959256914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,5120,0.01647733317481147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,5120,0.014030221435758801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,6144,0.013393777940008374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,5120,0.01198933356338077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,4096,0.014492443866199918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,4096,0.011977777712874942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,512,0.016905777984195285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,8192,256,0.01685688893000285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,3584,0.0127368892232577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,3584,0.011505777637163797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,4096,0.010588444769382477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,3072,0.023183110687467787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,3072,0.011367999845080905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,3072,0.010924444430404238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,65536,0.1362933317820231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,2560,0.021713778376579285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,2560,0.0103493332862854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,2560,0.009974222216341231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,3584,0.009724444813198512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,2048,0.020624889267815482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,2048,0.008753777378135258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,2048,0.008385777473449707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,3072,0.008463111188676622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,2560,0.007652444144090016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,8192,0.032808889945348106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,1536,0.007816889219813878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,1536,0.007259555160999298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,1024,0.018008000320858426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,1024,0.005838222387764189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,1024,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,2048,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,768,0.017380444539917838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,768,0.004576888763242298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,768,0.0063946665161185795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,7168,0.030305776331159804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,1536,0.006618666566080517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,512,0.004211555752489301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,5120,0.02693866689999898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,512,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,1024,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,256,0.016188444362746347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,256,0.003549333247873518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,256,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,128,0.014486221803559197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,128,0.0035235554807715942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,128,0.0058764442801475525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,768,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,64,0.0029297777348094513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,64,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,7168,32,0.0031519999934567344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,7168,32,0.005948444621430502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,65536,0.12915733125474718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,65536,0.13378577762179905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,4096,0.02517866591612498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,512,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,65536,0.08591644631491767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,16384,0.045484443505605064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,16384,0.04149688945876227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,256,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,16384,0.027036445008383855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,12288,0.03723466727468703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,12288,0.034095999267366194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,7168,128,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,12288,0.02179466684659322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,10240,0.028957333829667833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,10240,0.03381422162055969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,10240,0.01960177719593048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,8192,0.031022220849990845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,8192,0.02294844388961792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,8192,0.01685599982738495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,65536,0.07527644766701592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,7168,0.02901955445607503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,7168,0.01979377700222863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,7168,0.015784889459609985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,16384,0.02437422176202138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,6144,0.027463111612531874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,12288,0.01980977753798167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,6144,0.017859554953045316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,6144,0.01475822263293796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,5120,0.025419554776615564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,5120,0.015419556034935845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,10240,0.0170817772547404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,5120,0.013447110851605734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,3584,0.023688889212078516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,4096,0.02403555479314592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,8192,0.014839111102951897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,4096,0.013648889131016202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,4096,0.011729777687125735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,3584,0.0229093333085378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,3584,0.012248888611793518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,3584,0.011005333728260465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,7168,0.013850666582584381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,3072,0.021990221407678392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,3072,0.010993777877754636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,3072,0.010654222634103564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,6144,0.012631999949614206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,2560,0.021187555458810594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,2560,0.009716444545321995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,5120,0.011976000335481433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,2560,0.009300444689061906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,2048,0.01926666663752662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,2048,0.00903911143541336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,2048,0.007643555601437886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,4096,0.009424000150627559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,1536,0.007631999750932057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,1536,0.006626666833957036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,3584,0.008889777792824639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,1024,0.01688088807794783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,1024,0.005592888842026393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,1024,0.006873777757088344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,3072,0.00795288880666097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,768,0.01725955473052131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,1536,0.018206222189797294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,768,0.004872000051869286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,768,0.00629688882165485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,2560,0.00758133331934611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,512,0.015800888339678448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,512,0.003842666745185852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,512,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,256,0.016712889075279236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,256,0.0035111109415690103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,256,0.005906666732496685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,2048,0.0069297779765393995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,128,0.015104888214005364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,128,0.0032462223122517266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,128,0.00591111100382275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,1536,0.006569777925809224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,64,0.0028728888266616394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,64,0.005947555518812603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,6144,32,0.0029128889242808023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,6144,32,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,1024,0.006116444451941385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,65536,0.11555643876393636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,65536,0.12585333983103433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,65536,0.07977688974804349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,768,0.005903111149867375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,16384,0.03610488772392273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,512,0.005867555737495422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,16384,0.02520977788501316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,256,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,12288,0.030138668086793687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,7168,512,0.01590133375591702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,12288,0.02067377832200792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,6144,128,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,10240,0.03280977739228143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,10240,0.027271999253167048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,10240,0.01868799991077847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,8192,0.0302471121152242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,8192,0.018904889623324077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,65536,0.07298666901058622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,8192,0.016226665841208566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,7168,0.028090665737787884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,7168,0.01651022169325087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,7168,0.015223999818166098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,16384,0.02312177750799391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,6144,0.026486222942670185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,6144,0.01482311056719886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,6144,0.013922666509946188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,12288,0.019116444720162284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,5120,0.02460799945725335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,10240,0.016522667474216886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,5120,0.013341333303186627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,5120,0.013099555340078143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,4096,0.02323377794689602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,4096,0.011779555843936073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,4096,0.011050666371981302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,8192,0.014463111758232117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,3584,0.021985777550273474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,3584,0.01199288914601008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,3584,0.010392000277837118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,7168,0.013289777768982781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,3072,0.020594666401545208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,3072,0.010676444404655032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,3072,0.00923822240697013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,6144,0.012143111063374413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,2560,0.018567999203999836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,2560,0.00963288876745436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,2560,0.008436444732877944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,5120,0.01072355525361167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,2048,0.017200888858901132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,2048,0.00871022211180793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,4096,0.00866844422287411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,2048,0.006962666908899943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,3584,0.007989333735571967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,1536,0.016509332590632968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,1536,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,1536,0.006609777609507243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,3072,0.007652444144090016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,1024,0.015778667396969266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,1024,0.005401777724424998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,1024,0.006583111153708563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,2560,0.007269333634111617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,2048,0.006984000404675801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,768,0.004918222212129169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,768,0.006308444258239534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,1536,0.006581333362393909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,512,0.015153777268197803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,512,0.004087999877002505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,512,0.006139555739031897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,256,0.014873777826627096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,1024,0.0059075554211934405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,256,0.0038684445122877755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,256,0.005917333480384614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,128,0.015096889601813423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,128,0.0031795555518733132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,128,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,768,0.0059573331640826324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,64,0.0032142222755485107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,6144,1536,0.01686044368478987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,64,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,5120,32,0.003145777723855443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,5120,32,0.005927111125654644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,65536,0.0917359987894694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,512,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,65536,0.10168355703353882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,65536,0.07371911075380114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,256,0.005597333527273602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,16384,0.030881779061423406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,16384,0.023571555813153584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,12288,0.03128444486194187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,5120,128,0.00554933316177792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,12288,0.025777777036031086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,12288,0.019544000426928203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,10240,0.028833776712417603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,10240,0.022583999567561682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,10240,0.01808177762561374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,65536,0.06953244739108615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,16384,0.02204177776972453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,8192,0.016137777103318107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,8192,0.015495111544926962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,12288,0.01808444493346744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,7168,0.02451466686195797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,7168,0.014689778288205465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,7168,0.014631110760900708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,10240,0.01613066593805949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,16384,0.04327911138534546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,8192,0.01406044430202908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,6144,0.013176889055305056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,6144,0.013487999637921652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,12288,0.036580443382263184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,5120,0.011408000356621213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,5120,0.02235377828280131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,5120,0.012484444512261285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,4096,0.020953777763578627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,7168,0.012655110822783576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,4096,0.010437332921557956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,4096,0.010508444574144153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,6144,0.011676444775528379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,3584,0.00923999978436364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,3584,0.009361777868535783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,5120,0.009764444496896531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,3072,0.018132444885041978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,3072,0.008512000242869059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,4096,0.008286221987671321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,3072,0.008650666309727563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,3584,0.00796088907453749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,2560,0.007927999728255803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,2560,0.007658667034573025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,3072,0.0075911109646161394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,2048,0.016140444411171805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,2048,0.006640000061856375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,2048,0.007017778025733099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,2560,0.007268444531493717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,2048,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,1536,0.005222222043408289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,1536,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,5120,768,0.01685599982738495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,1536,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,1024,0.004222222086456087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,1024,0.006400888992680444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,1024,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,768,0.017099555995729234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,768,0.0038764443662431505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,768,0.006261333409282897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,16384,0.03681866659058465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,768,0.0058853332367208265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,512,0.0038382220599386427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,512,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,8192,0.026820444398456152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,512,0.005542222410440445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,256,0.015527110960748462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,256,0.00349866681628757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,256,0.005945777727497949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,6144,0.023400889502631292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,128,0.014876445134480795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,128,0.0031635556370019913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,128,0.005867555737495422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,256,0.005445333404673471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,64,0.002887111157178879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,64,0.0058408888677755995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,4096,32,0.0029075555503368378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,4096,32,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,4096,128,0.0052382221652401825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,65536,0.099181334177653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,3584,0.01923644377125634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,65536,0.08268533150355022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,65536,0.07117066780726115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,16384,0.03615555498335097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,16384,0.02861777941385905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,16384,0.023054222265879314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,12288,0.023278221487998962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,12288,0.019134221805466544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,65536,0.06815022230148315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,10240,0.02793688906563653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,2560,0.01686311099264357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,10240,0.021464889248212177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,10240,0.01737155516942342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,16384,0.021700445148679946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,8192,0.02608977754910787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,8192,0.018838221828142803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,8192,0.0151555554734336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,12288,0.017664889494578045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,7168,0.024319110645188227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,7168,0.013978666729397245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,7168,0.014115555418862237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,10240,0.015785778562227886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,6144,0.012235555383894177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,8192,0.013613333304723104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,6144,0.01330488920211792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,5120,0.021506667137145996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,7168,0.012139555480745105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,5120,0.011032000184059143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,5120,0.011701333026091257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,4096,0.0199635558658176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,1536,0.015152888165579902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,4096,0.010289777484205034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,4096,0.009721777505344814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,3584,0.01852888862291972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,3584,0.008960000342792934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,3584,0.00903555585278405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,6144,0.010854221880435944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,3072,0.017276444368892245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,1024,0.015447111593352424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,3072,0.008207110895050896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,5120,0.009564444422721863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,3072,0.00796888851457172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,2560,0.007252444823582967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,2560,0.01649688846535153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,2560,0.007620444728268518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,4096,0.008294222255547842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,2048,0.01578044394652049
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,2048,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,2048,0.006966222491529252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,1536,0.015494222442309061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,1536,0.00499199993080563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,3584,0.007664889097213745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,1536,0.006622222148709827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,1024,0.015153777268197803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,1024,0.004184000194072723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,1024,0.006609777609507243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,3072,0.007595555649863348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,768,0.01680266691578759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,2560,0.007152000235186682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,768,0.0038764443662431505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,768,0.006258666515350342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,2048,0.006640888750553131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,512,0.003522666792074839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,512,0.006061333335108227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,1536,0.006224888894293044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,256,0.01499466598033905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,256,0.0031679999083280563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,256,0.0058711109062035876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,1024,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,128,0.015105777316623263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,128,0.0028640000770489373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,128,0.005618666609128316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,768,0.005798222290145026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,64,0.0029928889125585556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,64,0.005600000007285013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3584,32,0.003165333428316646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,4096,512,0.014132445057233175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3584,32,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,512,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,65536,0.07271022266811795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,65536,0.09775022003385757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,65536,0.06912355290518866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,256,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,16384,0.0225591113169988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,16384,0.022317333353890315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3584,128,0.005534222142563925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,12288,0.030105776256985132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,12288,0.01928444372283088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,12288,0.018567111757066514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,65536,0.06693866517808703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,16384,0.021427555216683283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,10240,0.027826666831970215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,10240,0.016607999801635742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,10240,0.01680355601840549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,12288,0.017401778035693698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,8192,0.0255724440018336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,8192,0.01483022173245748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,8192,0.014814222852389017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,10240,0.015405333704418607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,7168,0.01311822235584259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,7168,0.013914666242069669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,8192,0.01311288856797748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,6144,0.022341334157519873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,6144,0.011979555918110741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,6144,0.01296533313062456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,5120,0.021174222230911255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,7168,0.011309333145618439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,5120,0.010642666783597736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,5120,0.011695110963450538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,6144,0.010228444304731157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,4096,0.009703111317422655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,4096,0.009347555538018545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,5120,0.009424000150627559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,4096,0.00814844419558843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,3584,0.008619555168681675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,3584,0.008651555412345463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,12288,0.030928889910380047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,3072,0.01722400055991279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,3072,0.008073777788215214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,3072,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,3584,0.007634667058785756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,2560,0.015849777393870883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,2560,0.007314666277832455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,2560,0.007274666594134436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,3072,0.007342221836249034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,2048,0.01551999979548984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,2048,0.006056889063782162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,6144,0.023030221462249756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,2048,0.0069715554515520734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,1536,0.015328889091809591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,1536,0.0052684446175893145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,1536,0.006640888750553131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,2560,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,1024,0.01443733274936676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,1024,0.004211555752489301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,1024,0.006272888845867581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,2048,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,768,0.016303999556435477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,768,0.003960000144110786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,1536,0.006274666637182236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,768,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,512,0.013820444544156393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,512,0.003632889025741153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,512,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,1024,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,256,0.014885332849290637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,256,0.0031582222630580268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,256,0.0058853332367208265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,768,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,128,0.01480977733929952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,128,0.0031848889258172777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,128,0.005584888988071018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,512,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,64,0.0028719999310043124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,64,0.005640888793600931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,3072,32,0.0028728888266616394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,256,0.0052586665583981415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,3072,32,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,65536,0.0634017785390218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,65536,0.09541244639290704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,65536,0.06640177965164185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3584,512,0.013814222481515674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,16384,0.03467999895413717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,16384,0.0221377776728736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,3072,128,0.005249777601824866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,16384,0.021341333786646526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,12288,0.019262222780121695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,12288,0.029663112428453233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,12288,0.01833511061138577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,65536,0.06571911440955268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,10240,0.016585777203241985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,10240,0.016144000821643405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,16384,0.02084622283776601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,16384,0.03554755449295044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,8192,0.024705777565638225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,8192,0.01475822263293796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,12288,0.0170471105310652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,8192,0.014389332797792224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,7168,0.023331556055280898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,7168,0.012894222305880653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,7168,0.013462222284740873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,6144,0.02197333342499203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,6144,0.011384888655609555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,6144,0.0127688886390792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,10240,0.014753777119848462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,5120,0.01924977699915568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,5120,0.010335110955768162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,8192,0.012038222617573209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,5120,0.010496888723638324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,7168,0.010760000182522668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,4096,0.017855111095640395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,4096,0.009312888814343346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,7168,0.023846222294701472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,4096,0.00833333366447025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,6144,0.010036444498433007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,3584,0.008287111090289222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,3584,0.01682577696111467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,3584,0.007976888782448238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,3072,0.016315556234783597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,5120,0.009307555854320526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,3072,0.007641777396202087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,3072,0.007624000310897827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,2560,0.015559110376569959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,2560,0.006917333437336816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,2560,0.007330666813585494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,4096,0.008078222473462423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,3584,0.007620444728268518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,2048,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,2048,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,3072,0.007328888608349695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,1536,0.01481599940194024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,1536,0.004935111022657818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,1536,0.006602666858169768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,2560,0.006960888703664144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,1024,0.01401511165830824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,1024,0.0041759999261962045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,1024,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,768,0.01646933290693495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,768,0.0038497779104444715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,2048,0.006602666858169768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,768,0.006218666831652324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,512,0.013405332962671915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,512,0.0035617777870761026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,4096,0.018428444862365723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,512,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,1536,0.006284444282452266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,256,0.014422222971916199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,256,0.0032026666320032547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,256,0.005882666756709416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,128,0.014137778017255994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,1024,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,128,0.0028568889117903183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,128,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,64,0.0028124445428450904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,768,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,64,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2560,32,0.002826666666401757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2560,32,0.005591111050711737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,3072,3584,0.017557332913080852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,512,0.005564444594913059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,65536,0.05224533213509453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,65536,0.08790755271911621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,65536,0.0641804469956292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,256,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,16384,0.018789332773950364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,16384,0.03256977929009332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,16384,0.020948444803555805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,12288,0.015511110424995422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,12288,0.02847911251915826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,12288,0.017903111047214933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2560,128,0.005247999810510212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,10240,0.025607999828126695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,10240,0.015122666954994202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,10240,0.01628177695804172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,8192,0.023725334140989516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,8192,0.013796444568369122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,8192,0.01399466726515028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,65536,0.06459999746746488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,7168,0.021983999345037673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,16384,0.020652444826232064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,7168,0.012324444121784635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,7168,0.012887111140622033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,12288,0.016349333855840895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,6144,0.01977244516213735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,6144,0.011424000064531961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,6144,0.011520889070298938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,10240,0.013454222016864352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,5120,0.018385777870814007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,5120,0.012808889150619507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,5120,0.009679999616410997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,4096,0.011083555718262991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,8192,0.011351111034552256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,4096,0.01688088807794783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,4096,0.00833422193924586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,3584,0.009689778089523315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,3584,0.01614755557643043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,7168,0.01037688884470198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,3584,0.007970666719807519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,3072,0.0158897770775689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,3072,0.008656000097592672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,6144,0.00960177762640847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,3072,0.007692444655630324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,5120,0.008665777742862701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,2560,0.007333333293596904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,2560,0.007303111255168915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,4096,0.008032889001899296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,2048,0.0067528887755341
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,2048,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,3584,0.007611555357774098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,3072,0.007291555404663086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,2560,0.0069511110583941145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,1536,0.004870222260554631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,1536,0.006640888750553131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,2048,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,1024,0.014145778285132514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,1024,0.0041911109454101985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,1024,0.006280888699822956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,1536,0.0062871111763848206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,768,0.016180444094869826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,768,0.003903999924659729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,768,0.0063164445261160536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,512,0.013470222552617392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,1024,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,512,0.003559111307064692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,512,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,256,0.014659555421935188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,256,0.0031644445326593188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,256,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,768,0.005884444548024072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,128,0.013792888985739814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,128,0.003160888950030009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,512,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,128,0.005675555517276128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,64,0.0028080000645584534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,64,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,2048,32,0.002826666666401757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,256,0.005285333428117964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,2048,32,0.005665777872006099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,65536,0.03994399971432156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,2048,128,0.00526311124364535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,65536,0.06256533331341214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,65536,0.06402044163809882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,16384,0.01789866718981001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,16384,0.020589333441522386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,16384,0.019893333315849304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,10240,0.026777777406904433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,12288,0.028175999720891316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,12288,0.015446222490734525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,12288,0.017551110850440133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,10240,0.02550044490231408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,10240,0.013790222505728403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,10240,0.015783111254374187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,12288,0.014684443672498068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,8192,0.022260444031821355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,8192,0.012728888955381183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,8192,0.012846222354306115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,10240,0.012628444366984896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,7168,0.020296888218985665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,7168,0.011584888729784222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,7168,0.01106844428512785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,8192,0.01108444482088089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,6144,0.018981332580248516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,6144,0.011062222222487131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,6144,0.010027555955780877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,7168,0.010055111514197456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,5120,0.01791733337773217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,5120,0.0103502223889033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2560,2048,0.015288000305493673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,5120,0.009334222310119206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,4096,0.01719555589887831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,4096,0.009025777379671732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,6144,0.009376000199053023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,4096,0.008341333104504479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,5120,0.008683555656009251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,3584,0.007483555210961237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,3584,0.007992888490358988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,4096,0.008002666963471307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,3072,0.015497777197096082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,3072,0.006874666859706243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,3072,0.007634667058785756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,3584,0.007648888561460707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,2560,0.015112888481881885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,2560,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,2560,0.007313778003056844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,3072,0.007312888900438945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,2560,0.006989333364698622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,2048,0.006124444305896759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,2048,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,2048,0.006625777731339137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,1536,0.014091556270917257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,1536,0.004855999930037392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,1536,0.0063528890411059065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,1536,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,2048,0.01480888823668162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,1024,0.0041555555330382455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,1024,0.006267555471923616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,2560,0.015159999330838522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,768,0.016144000821643405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,768,0.0038382220599386427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,768,0.006021333237489064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,1024,0.005959110955397288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,768,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,512,0.0031946665710873077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,512,0.0059262220230367445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,2048,1536,0.014073777529928418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,256,0.014826666977670459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,256,0.003212444484233856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,256,0.005600000007285013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,512,0.005569777968857024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,256,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,65536,0.08681688706080119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,128,0.002846222163902389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,16384,0.03222577770551046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,128,0.005883555445406172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,64,0.00282577777074443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,64,0.005888888819350137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1536,32,0.002838222309947014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1536,32,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,65536,0.029434667693244085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,65536,0.0852960016992357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1536,128,0.00554044461912579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,65536,0.060580445660485156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,16384,0.014537778165605334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,16384,0.01996977792845832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,65536,0.06328711244795057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,16384,0.017824000782436795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,12288,0.01274222218328052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,12288,0.01589777734544542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,12288,0.013843555417325763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,10240,0.022864888111750286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,10240,0.011760888828171624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,10240,0.014120888378885059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,10240,0.012110222544935016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,3584,0.016314667132165697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,8192,0.013754666679435305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,8192,0.011408888631396823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,8192,0.010781333678298526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,7168,0.0199644449684355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,7168,0.012455999851226807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,7168,0.01070844464831882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,7168,0.009940444595283931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,2048,0.014487110906177096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,6144,0.018908444378111098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,6144,0.011404444773991903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,6144,0.009672889278994666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,5120,0.01791733337773217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,5120,0.009692444569534725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,5120,0.009343110852771336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,6144,0.009361777868535783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,4096,0.01626399987273746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,4096,0.00904799997806549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,4096,0.007983110845088959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,5120,0.00868088917599784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,4096,0.007990222010347579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,1024,0.013535999589496188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,3584,0.007086222370465596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,3584,0.007983110845088959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,3072,0.015153777268197803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,3072,0.006606222026877933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,3072,0.007599111232492659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,3584,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,2560,0.014519110321998596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,2560,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,3072,0.007322666545708974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,2560,0.007299555672539606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,2048,0.014142221874660917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,2048,0.00629333323902554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,2048,0.006643555644485686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,2560,0.006959999601046245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,1536,0.013757333159446716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,512,0.01312888910373052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,1536,0.005428444594144821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,1536,0.006252444452709622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,2048,0.006421333385838403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,1024,0.013745778136783175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,1024,0.004567111117972268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,1024,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,768,0.01579466627703773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,1536,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,768,0.0038355555799272326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,768,0.005966222120655908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,1024,0.005723555468850666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,512,0.003560888684458203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,512,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,768,0.005782222168313132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,256,0.014577777849303352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,256,0.0032035555276605818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,256,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,512,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,256,0.005320888840489917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,128,0.003167111012670729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,128,0.0058764442801475525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1536,128,0.01478666729397244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,64,0.0028133332315418455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,1024,128,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,64,0.005956444475385878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,1024,32,0.0028168888141711554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,1024,32,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,65536,0.03093866507212321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,65536,0.08494666549894546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,65536,0.06012711260053846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,16384,0.031609776947233416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,16384,0.012048000262843238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,65536,0.06291555696063571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,16384,0.01961777773168352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,12288,0.025429333249727886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,12288,0.010670222342014313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,12288,0.01571555601225959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,16384,0.016738666428460013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,12288,0.013769778112570444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,16384,0.0318195554945204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,10240,0.022709333234363135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,10240,0.010038221875826517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,10240,0.013041777743233575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,8192,0.020964443683624268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,8192,0.009649777577983009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,10240,0.012079999678664737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,8192,0.011402666568756104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,12288,0.025771554973390367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,7168,0.008965333302815756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,7168,0.01071999967098236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,8192,0.010782221953074137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,6144,0.018211555149820115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,6144,0.008976889153321585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,6144,0.009707555174827576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,7168,0.009768000079525841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,5120,0.01715555621518029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,5120,0.00795911086930169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,6144,0.009297777381208207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,5120,0.0090506664580769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,4096,0.016472889317406546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,4096,0.007611555357774098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,4096,0.008027555214034187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,5120,0.00868000007337994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,3584,0.015447111593352424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,3584,0.007679999702506595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,3584,0.007643555601437886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,4096,0.007728888756699032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,3072,0.0151591118839052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,3072,0.006579555571079254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,3072,0.007296000089910295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,2560,0.014602666099866232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,3584,0.007602666815121968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,2560,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,2560,0.007229333122571309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,3072,0.007257777783605788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,8192,0.020970667401949566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,2048,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,2048,0.006771555377377405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,2560,0.006832888970772426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,1536,0.013790222505728403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,1536,0.004556444370084339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,1536,0.006584888945023219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,2048,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,1024,0.013449778159459433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,1024,0.0041306668685542215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,1024,0.006266666783226862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,768,0.015136000182893542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,1536,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,768,0.0035955554081334006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,768,0.0059075554211934405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,1024,0.005740444279379315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,512,0.012411555482281579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,512,0.0033004445334275565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,512,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,768,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,256,0.014143110977278816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,256,0.0031511110977994073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,256,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,512,0.005224888937340842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,128,0.0028124445428450904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,128,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,256,0.005244444641802046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,64,0.002516444358560774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,64,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,768,32,0.0028773333049482773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,768,128,0.0052062223354975385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,768,32,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,65536,0.023353776997990076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,65536,0.0844871136877272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,65536,0.05923022164238823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,16384,0.029622223642137315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,65536,0.06299999687406752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,16384,0.013429332938459186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,16384,0.017314665847354464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,12288,0.02457777824666765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,12288,0.011115555961926779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,16384,0.016275554895401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,12288,0.014298665854665967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,12288,0.013447110851605734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,10240,0.010093332992659675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,10240,0.01276088919904497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,10240,0.012110222544935016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,3584,0.015860444969601102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,8192,0.009702222214804756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,8192,0.011071110765139261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,8192,0.01071466671095954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,512,0.01276533305644989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,7168,0.008957333034939235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,7168,0.010673777924643623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,7168,0.009715555442704095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,6144,0.01794755624400245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,6144,0.00758133331934611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,6144,0.010001777774757808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,6144,0.009016888837019602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,5120,0.016876444220542908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,5120,0.008046222229798635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,5120,0.009302222066455418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,1024,128,0.01346933344999949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,5120,0.00886044485701455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,4096,0.015665777855449252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,4096,0.007657777931955125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,4096,0.007891555627187094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,4096,0.007991111112965478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,7168,0.019555555449591745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,3584,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,3584,0.007818666597207388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,3584,0.007655111451943715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,3072,0.006285333385070165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,3072,0.007655999726719326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,3072,0.007308444215191736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,2560,0.014434667097197639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,2560,0.005619555711746216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,2560,0.007272889216740926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,2560,0.0069955554273393415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,2048,0.014512000812424554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,2048,0.005223999834722943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,2048,0.0069422221018208405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,2048,0.006574222197135289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,768,128,0.01311911145846049
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,1536,0.004512888689835866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,1536,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,1536,0.006162666612201267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,10240,0.022014222211307947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,1024,0.0038275557259718576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,1024,0.005969777703285217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,1024,0.00573511090543535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,8192,0.020253333780500624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,768,0.003552000141806073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,768,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,768,0.005589333259397083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,512,0.011753777662913004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,512,0.0035084444615576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,512,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,512,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,256,0.013820444544156393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,256,0.0032311110860771606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,256,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,256,0.0052284445199701525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,128,0.012450666891203986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,128,0.002887111157178879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,7168,0.01926488843229082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,128,0.005591111050711737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,64,0.0025555555605226094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,64,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,512,128,0.005245333330498801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,512,32,0.0028479999552170434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,512,32,0.005687110953860813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,65536,0.0170471105310652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,65536,0.058380444844563804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,65536,0.0616204473707411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,16384,0.028647998968760174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,16384,0.010044444766309526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,16384,0.016248888439602323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,16384,0.016532444291644625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,12288,0.02405333353413476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,12288,0.0083155557513237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,12288,0.013797333670987023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,12288,0.013739555246300168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,10240,0.021760889225535925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,10240,0.009293333523803288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,10240,0.012421333127551608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,10240,0.01202222208182017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,8192,0.01993955506218804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,8192,0.008456000023418004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,8192,0.011063111325105032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,8192,0.010791999598344168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,7168,0.01891288823551602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,7168,0.00794933322403166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,7168,0.010434666441546546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,7168,0.009758222434255812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,6144,0.01789688898457421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,3584,0.01520088811715444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,6144,0.0074879998962084455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,6144,0.009693333672152625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,5120,0.015855110353893705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,6144,0.009338666995366415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,5120,0.00831911133395301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,5120,0.008989333278603023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,3072,0.015115555789735583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,4096,0.015836444165971544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,4096,0.007305777735180325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,5120,0.00867377801073922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,4096,0.008001777860853408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,3584,0.007318221860461765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,3584,0.014563555518786112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,4096,0.007998222278224098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,3584,0.007837333612971837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,3072,0.014426666829321118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,3072,0.00673333348499404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,3584,0.007670222057236566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,3072,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,2560,0.01312355531586541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,2560,0.006267555471923616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,3072,0.007280000381999546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,2560,0.0069955554273393415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,2048,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,2048,0.01346577786737018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,2560,0.0069679998689227635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,2048,0.006630222416586346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,1536,0.012440889245933957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,2048,0.006630222416586346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,1536,0.004558222161398994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,1536,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,1024,0.012549333274364471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,1536,0.005946666830115848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,1024,0.003834666477309333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,2048,0.01408088869518704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,1024,0.005960000058015187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,768,0.014774221513006421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,1024,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,768,0.00351555562681622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,768,0.006232000059551663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,512,0.0032631111227803757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,512,0.011400889191362592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,512,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,768,0.00564444437623024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,256,0.013134222063753339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,512,0.005271111097600725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,256,0.0028204443968004654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,256,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,128,0.0028533333291610083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,256,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,128,0.0058853332367208265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,64,0.0025395556456512874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,256,128,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,64,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,256,32,0.00253866674999396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,256,32,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,65536,0.015767999821239047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,65536,0.0825893349117703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,65536,0.0581244428952535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,16384,0.027801778581407335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,65536,0.058838221761915416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,16384,0.009326222042242685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,16384,0.016179554992251925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,12288,0.022675555613305833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,16384,0.01647644407219357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,12288,0.008351999852392409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,12288,0.013757333159446716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,10240,0.021370665894614324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,12288,0.013463111387358772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,10240,0.0076044441925154785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,10240,0.012355555262830524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,10240,0.012074666718641916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,1536,0.013414222333166333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,8192,0.019270223047998216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,8192,0.007130666739410824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,8192,0.011137777732478248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,8192,0.01072533345884747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,7168,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,7168,0.010489777558379702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,7168,0.009825777676370408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,6144,0.017115554875797696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,6144,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,6144,0.009993777506881291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,1024,0.012420444852775998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,6144,0.00908088849650489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,5120,0.016472889317406546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,5120,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,5120,0.008957333034939235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,5120,0.008632889224423302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,4096,0.006603555546866522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,4096,0.00795288880666097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,4096,0.00792622235086229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,3584,0.014607111612955729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,3584,0.006263111200597551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,3584,0.00794488853878445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,3584,0.007621333003044128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,3072,0.013456888496875763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,3072,0.006212444355090459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,3072,0.007658667034573025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,2560,0.006249777972698212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,2560,0.013752889302041797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,3072,0.007248000138335758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,2560,0.0074035558435651995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,2560,0.006966222491529252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,2048,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,2048,0.012424888710180918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,2048,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,2048,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,1536,0.004535111288229625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,1536,0.011732444167137146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,1536,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,1536,0.006248000181383557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,1024,0.011404444773991903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,1024,0.0038897775941424896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,1024,0.0064088888466358185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,1024,0.005709333138333426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,768,0.013804444836245643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,768,0.0034560000316964257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,768,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,512,768,0.014127999544143677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,512,0.011363555159833698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,768,0.005555555638339784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,512,0.0032631111227803757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,512,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,512,0.00554666668176651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,256,0.0028702221396896574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,256,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,256,0.005227555417352253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,128,0.011033777561452655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,128,0.0026319999661710528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,128,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,64,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,96,128,128,0.005205333232879639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,64,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,128,32,0.0026391111314296722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,128,32,0.00555288874440723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,65536,0.014536889062987434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,16384,0.007984889050324758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,65536,0.056582219070858426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,16384,0.015769778026474845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,12288,0.007602666815121968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,12288,0.013627555635240344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,10240,0.007269333634111617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,10240,0.012056888805495368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,8192,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,8192,0.010710222025712332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,7168,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,7168,0.010010666317409938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,6144,0.006220444622966979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,6144,0.009473777479595607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,5120,0.006228444476922353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,5120,0.008621333373917473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,4096,0.005562666803598404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,4096,0.007914666500356462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,3584,0.005910222315125995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,3584,0.007630222373538547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,3072,0.005616000129116907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,3072,0.007280888656775157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,2560,0.006015111174848344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,2560,0.006909333169460297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,2048,0.005562666803598404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,2048,0.006907555378145642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,1536,0.004512888689835866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,1536,0.006566222343179915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,1024,0.0038942222793896994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,1024,0.0058871110280354815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,768,0.0034924443397257063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,768,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,512,0.0032151111712058387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,512,0.005620444400442972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,256,0.0028657778683635923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,256,0.0053280000057485365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,128,0.0025439999169773525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,128,0.005777777896987067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,64,0.002526222210791376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,64,0.005394666559166378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,64,32,0.0028213332924577924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,64,32,0.005615111026499007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,65536,0.012771555946932899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,65536,0.055952886740366616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,16384,0.006903111106819577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,12288,0.007281777759393056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,16384,0.01585155559910668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,12288,0.013439999686347114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,10240,0.007253333098358578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,10240,0.01239733315176434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,65536,0.0836942195892334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,8192,0.006911110960774952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,8192,0.011016888750924004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,7168,0.006736889067623351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,7168,0.010375111467308467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,6144,0.006401777681377199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,6144,0.009663999908500248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,5120,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,5120,0.009013333254390294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,4096,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,4096,0.008000888758235509
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,3584,0.005938666562239329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,3584,0.0074622225430276655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,3072,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,3072,0.007600000335110559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,2560,0.006259555617968242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,2560,0.007304888632562425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,2048,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,2048,0.006619555668698416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,1536,0.0042444442709287005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,1536,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,1024,0.003900444342030419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,768,0.005938666562239329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,1024,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,768,0.0035004446076022256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,512,0.003152888889114062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,512,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,256,0.0028684443483750024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,256,0.0053102220926019884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,128,0.002521777732504739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,128,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,64,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,64,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,96,32,32,0.0028133332315418455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,96,32,32,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,16384,0.36053866810268825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,16384,0.2345351113213433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,16384,0.1778951088587443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,12288,0.2772702111138238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,12288,0.13246933619181314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,256,128,0.011677333050303988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,10240,0.21634666124979654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,10240,0.11182222101423477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,7168,0.017854221993022494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,4096,0.015495111544926962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,8192,0.17407111326853433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,8192,0.09451466798782349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,96,128,256,0.011370666325092316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,7168,0.13009866078694662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,7168,0.15272266334957546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,7168,0.0796604421403673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,12288,0.18906311194101968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,10240,0.1638151142332289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,6144,0.12041600545247395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,6144,0.1313306623035007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,6144,0.07059644328223334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,5120,0.10707822110917832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,5120,0.11385599772135417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,5120,0.05918044514126248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,8192,0.14715377489725748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,4096,0.09701066546969944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,4096,0.09416711330413818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,4096,0.04810133245256212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,16384,0.1340106725692749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,12288,0.10211910804112752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,3584,0.08043644163343641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,3584,0.04308355516857571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,10240,0.08408355712890625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,8192,0.0686746637026469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,3072,0.07015022304322985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,3072,0.03750666644838121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,7168,0.06077510780758328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,3584,0.09081333213382298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,2560,0.05943644709057278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,2560,0.03284800052642822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,6144,0.058697779973347984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,2048,0.07430755429797702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,2048,0.04785511228773329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,2048,0.028551999065611098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,3072,0.08457155360115899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,5120,0.04553866717550489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,1536,0.037500444385740496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,1536,0.025107555919223364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,4096,0.03812177644835578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,2560,0.07859022087521024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,1024,0.027847111225128174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,1024,0.01962222158908844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,3584,0.03519555595186021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,768,0.030247999562157526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,768,0.021357332666714985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,768,0.01793155570824941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,3072,0.03007733490731981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,512,0.05676177475187513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,512,0.014680888917711047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,512,0.015658666690190632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,1536,0.06801866822772555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,256,0.02514133354028066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,256,0.009322666459613377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,256,0.013802666631009845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,1024,0.06223022275500827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,2560,0.02715822226471371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,128,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,128,0.01074044406414032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,2048,0.024241778585645888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,64,0.006619555668698416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,64,0.011051555474599203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,1536,0.019083556201722886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,65536,32,0.006802666518423293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,65536,32,0.012449777788586087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,65536,0.3268008761935764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,65536,0.22998044225904676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,65536,0.19142044915093315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,16384,0.07494755585988362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,16384,0.08395644691255356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,16384,0.051746666431427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,1024,0.015561777684423657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,12288,0.06100177764892578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,12288,0.0656133360332913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,12288,0.040298667218950056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,768,0.014316444595654806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,10240,0.05408622158898247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,10240,0.05392355720202128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,10240,0.03538489010598924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,65536,0.15140977170732287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,8192,0.04964266551865471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,8192,0.04654399885071648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,8192,0.03051911128891839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,65536,128,0.020966221888860066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,16384,0.044340444935692676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,7168,0.03923555546336704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,7168,0.026562665899594624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,512,0.01273155543539259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,6144,0.04158844550450643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,6144,0.034625778595606484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,6144,0.024282667371961806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,12288,0.03310488992267185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,10240,0.028775112496482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,5120,0.02948533495267232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,5120,0.020959999826219346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,8192,0.024526221884621516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,256,0.011796444654464722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,7168,0.02128000060717265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,4096,0.024671110841963027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,4096,0.017516444126764934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,6144,0.021095999413066443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,3584,0.021937777598698933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,3584,0.033302221033308245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,3584,0.016511110795868766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,3072,0.031713777118259005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,3072,0.01939733326435089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,3072,0.01498222185505761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,65536,128,0.011707555916574268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,2560,0.030174222257402208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,2560,0.016227554943826463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,2560,0.013774221969975365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,5120,0.017533333765135873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,2048,0.02951288885540432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,2048,0.013435555828942193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,2048,0.012336889074908363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,4096,0.016202666693263583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,1536,0.027962668074501887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,1536,0.011309333145618439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,1536,0.011913778053389655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,3584,0.01447111037042406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,1024,0.02591111097070906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,1024,0.008767999708652496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,1024,0.009975999593734741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,3072,0.01365244471364551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,768,0.0199955552816391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,768,0.007305777735180325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,768,0.008629333641793992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,7168,0.04395644532309639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,512,0.02200711104604933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,512,0.005568888866239124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,512,0.007319110963079665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,2560,0.012433778080675336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,5120,0.03794222076733907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,256,0.016878222425778706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,256,0.0041999999019834726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,256,0.006940444310506185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,128,0.01516355574131012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,128,0.0035439998739295532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,128,0.006626666833957036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,2048,0.011682666838169098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,64,0.0032017777363459268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,64,0.0069466663731469055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,16384,32,0.0035084444615576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,16384,32,0.007526222202512953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,1536,0.009639999932712978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,65536,0.25326400332980686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,16384,4096,0.035220444202423096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,1024,0.007639110916190677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,16384,0.06134577592213949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,16384,0.07047466437021892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,768,0.007296889192528195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,65536,0.16576622592078313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,16384,0.046111110183927745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,12288,0.04941599898868137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,12288,0.05605066484875149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,12288,0.036213331752353244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,512,0.0069857777820693115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,10240,0.04402755697568258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,10240,0.0458808872434828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,10240,0.03159733282195197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,256,0.0064035554726918536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,8192,0.03865777783923679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,8192,0.02708622150950962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,16384,128,0.006634666687912411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,7168,0.03663288884692722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,7168,0.034191999170515276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,7168,0.024194666081004675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,65536,0.13755555947621664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,6144,0.0343413319852617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,6144,0.029483556747436523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,6144,0.02204711072974735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,16384,0.039269334740108915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,5120,0.03166933192147149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,12288,0.03089600139194065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,5120,0.02589688864019182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,5120,0.019261333677503798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,4096,0.03017866611480713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,4096,0.02216000027126736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,10240,0.026699556244744196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,4096,0.016524444023768108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,3584,0.028363555669784546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,3584,0.019102222389645047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,3584,0.01536977787812551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,8192,0.02251822253068288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,7168,0.01998488936159346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,3072,0.016626665989557903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,3072,0.014185777968830533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,6144,0.018271999226676095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,2560,0.026322666141721938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,2560,0.014173333843549093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,2560,0.01311644415060679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,5120,0.01681155628628201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,2048,0.0255422227912479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,2048,0.012163555456532372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,2048,0.01178311142656538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,4096,0.014928888943460254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,3584,0.013564444250530668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,1536,0.025403555896547105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,1536,0.010096888575288985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,1536,0.011095999843544431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,3072,0.012815111213260226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,1024,0.02273155583275689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,1024,0.008635555704434713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,1024,0.00888888869020674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,768,0.01819022165404426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,768,0.007254222200976477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,768,0.00794577764140235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,65536,0.17491200235154894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,512,0.019574221637513902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,512,0.005205333232879639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,512,0.007297777467303806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,2560,0.011735111474990845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,256,0.017216889394654166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,256,0.003909333298603694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,256,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,2048,0.01096088853147295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,128,0.014840889308187695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,128,0.003540444291300244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,128,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,8192,0.04077688852945963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,64,0.0034497777620951333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,1536,0.0084906667470932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,64,0.006618666566080517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,12288,32,0.003516444315512975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,12288,32,0.007272000114123027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,65536,0.21936444441477457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,1024,0.007333333293596904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,768,0.007240888973077138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,65536,0.15632532702551946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,16384,0.0579893324110243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,16384,0.05512444509400261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,512,0.006615110983451207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,16384,0.0440577765305837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,12288,0.04466222061051262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,12288,0.03445955448680454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,256,0.006341333190600078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,12288,128,0.0063484443558586975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,10240,0.037965334124035306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,10240,0.030476444297366675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,12288,3072,0.02759644389152527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,65536,0.1314328908920288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,8192,0.03194844391610887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,8192,0.025860443711280823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,16384,0.03792000148031447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,7168,0.03381777803103129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,7168,0.028611554039849177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,7168,0.023050665855407715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,12288,0.029320889049106177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,6144,0.03199999862247043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,6144,0.02513422237502204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,6144,0.02106933295726776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,10240,0.025751110580232408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,5120,0.029600001043743555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,5120,0.02164977788925171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,5120,0.01887999971707662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,8192,0.023035556077957153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,4096,0.028014222780863445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,4096,0.018954666124449838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,4096,0.015815999772813585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,7168,0.020217777954207528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,3584,0.026605332891146343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,3584,0.016739555531077914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,3584,0.014928888943460254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,6144,0.017901332841979135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,5120,0.01629155543115404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,3072,0.014864888456132678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,3072,0.013722666435771518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,4096,0.01422933406300015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,3584,0.013554666605260638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,2560,0.013426666458447775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,2560,0.012715555727481842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,65536,0.15679911772410074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,12288,0.04543200135231018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,2048,0.011367110742463006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,2048,0.011606222225560082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,3072,0.012470222181744046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,1536,0.023132445083724126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,1536,0.00980355590581894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,1536,0.010029333333174387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,10240,0.04036000039842393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,1024,0.01993777851263682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,1024,0.007995555798212687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,1024,0.008341333104504479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,2560,0.011397333608733283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,768,0.01780177818404304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,768,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,768,0.007639110916190677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,8192,0.03738044367896186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,2048,0.009335999687512716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,512,0.0052720002002186244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,512,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,1536,0.008191111187140146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,256,0.01689511040846507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,256,0.004193777839342753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,256,0.006633777585294511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,1024,0.007295110987292395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,768,0.006684444430801604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,128,0.0148062225845125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,128,0.003568888952334722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,128,0.006488889041874144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,512,0.006568000134494569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,64,0.003519111209445529
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,64,0.006634666687912411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,10240,32,0.0031866667171319327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,10240,32,0.006618666566080517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,3072,0.02548800077703264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,65536,0.17553867234124076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,65536,0.1400444507598877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,16384,0.04996711015701294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,65536,0.10256177849239773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,256,0.006249777972698212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,16384,0.047934220896826855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,16384,0.03017066584693061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,12288,0.0413626664214664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,12288,0.03729155659675598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,12288,0.02494311167134179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,10240,0.036871112055248685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,10240,0.031249778138266668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,10240,0.022086222966512043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,10240,128,0.006238222122192383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,8192,0.0336444444126553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,8192,0.026532444689008925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,8192,0.01888088881969452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,65536,0.08584800031450059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,7168,0.0307644440068139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,2560,0.0243511117166943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,16384,0.02551288902759552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,7168,0.023271110322740342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,7168,0.017516444126764934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,6144,0.029504888587527808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,6144,0.020558221472634208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,12288,0.02053244411945343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,6144,0.01609688831700219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,5120,0.01761244402991401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,5120,0.027092445227834914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,5120,0.014474666780895658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,4096,0.025920000341203477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,4096,0.015029332704014249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,4096,0.01237777786122428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,10240,0.018228444788191054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,3584,0.024822221861945257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,3584,0.013198222551080914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,3584,0.011832888755533429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,8192,0.016132444143295288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,3072,0.02348800003528595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,3072,0.012087110843923358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,3072,0.011011555790901184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,2048,0.024567999773555334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,2560,0.022729777627521094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,2560,0.010447999669445885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,2560,0.010411555568377176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,7168,0.014627556006113688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,2048,0.02186133298608992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,6144,0.014057776994175382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,2048,0.009370666411187913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,2048,0.009352000223265754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,5120,0.012310222619109683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,1536,0.00792711145348019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,1536,0.008288000192907121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,4096,0.011330666641394297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,3584,0.010382222632567087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,1024,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,1024,0.006933333145247565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,10240,512,0.018580444984965853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,3072,0.009463110731707679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,768,0.017335999343130324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,768,0.004905777672926585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,768,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,512,0.017156443662113614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,512,0.0041857775714662345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,512,0.006578666468461354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,2560,0.008006222546100616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,256,0.016869333055284288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,256,0.003565333369705412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,256,0.005920889063013925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,2048,0.0075528886583116316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,128,0.01516177753607432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,128,0.0031617778456873367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,1536,0.006625777731339137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,128,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,64,0.003152888889114062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,64,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,8192,32,0.003160888950030009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,8192,32,0.006170666466156642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,65536,0.13560799757639566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,1024,0.006561777657932705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,65536,0.15338310930464003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,65536,0.0950195524427626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,768,0.006168889088763132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,16384,0.04741777645217048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,16384,0.04713066750102573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,16384,0.028925332758161757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,512,0.005910222315125995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,12288,0.039750221702787615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,12288,0.03461600012249417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,12288,0.023057777020666335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,256,0.005591111050711737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,10240,0.03531644410557217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,10240,0.02883111106024848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,8192,128,0.005571555760171678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,10240,0.020800888538360596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,8192,0.03264355659484863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,8192,0.02464977734618717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,8192,0.018171555466122098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,65536,0.0771626697646247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,7168,0.030072000291612413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,7168,0.02163022259871165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,16384,0.025398222936524287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,7168,0.016519111063745286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,6144,0.028416888581381902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,6144,0.0188426673412323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,6144,0.01516266663869222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,12288,0.020081778367360432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,5120,0.026403556267420452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,5120,0.01650044487582313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,5120,0.014021333720948962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,4096,0.02495466669400533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,4096,0.014754666222466363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,10240,0.017733333839310538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,4096,0.012070222033394707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,8192,0.015384889311260648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,3584,0.01275111072593265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,3584,0.011608888705571493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,7168,0.014354666074117025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,3072,0.022780444886949327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,6144,0.013400889105266996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,3072,0.011722666521867117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,3072,0.010761777559916178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,2560,0.021944888763957556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,5120,0.011794666449228922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,2560,0.010217777556843227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,2560,0.00998755544424057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,2048,0.02111022174358368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,2048,0.008945778012275696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,4096,0.010551999840471478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,2048,0.008656000097592672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,1536,0.0191928893327713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,1536,0.007668444679843054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,1536,0.007583111524581909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,3584,0.009099555512269339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,1024,0.017622222503026325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,1024,0.005718222094906702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,1024,0.006952889263629913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,3072,0.00868800034125646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,2560,0.007652444144090016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,768,0.004632888982693355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,768,0.006582222051090664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,2048,0.007095110913117726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,512,0.016528000434239704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,512,0.0038853333228164246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,1536,0.020460445019933913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,512,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,1536,0.006616000086069107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,256,0.003469333259595765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,256,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,8192,1024,0.018231110440360177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,128,0.01482399966981676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,1024,0.005990222096443176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,128,0.0032071111102898917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,128,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,64,0.0028799999919202593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,64,0.005695999910434087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,7168,32,0.0032213332338465583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,7168,32,0.005953777581453323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,65536,0.12747466564178467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,65536,0.12786044014824763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,768,0.005946666830115848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,65536,0.08535110950469971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,512,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,16384,0.04495999879307217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,16384,0.04063733418782552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,256,0.00556355549229516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,16384,0.02701955537001292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,12288,0.03337422344419692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,12288,0.021885333789719477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,7168,128,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,10240,0.03354488809903463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,10240,0.029291556941138372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,10240,0.019475556082195707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,65536,0.07404266463385688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,8192,0.03108533223470052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,8192,0.022520000735918682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,8192,0.01714933415253957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,16384,0.02409244411521488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,7168,0.028468443287743464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,7168,0.02102577851878272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,12288,0.019323555959595572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,7168,0.015669332610236276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,6144,0.02719733284579383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,6144,0.018367111682891846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,6144,0.014473777678277759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,10240,0.01742755538887448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,5120,0.025124443901909724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,5120,0.01584533353646596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,5120,0.01367377738157908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,8192,0.014827556080288358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,7168,0.01346666696998808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,4096,0.024022221565246582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,4096,0.013991999957296582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,4096,0.011394666300879585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,6144,0.013437333206335703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,3584,0.023659555448426142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,3584,0.01218933363755544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,3584,0.022618666291236877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,3584,0.011152889165613385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,3072,0.011003555523024665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,3072,0.010334221853150262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,5120,0.011388444238238864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,4096,0.009593778186374241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,2560,0.009878222313192155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,2560,0.008839111361238692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,3584,0.008735111190213097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,2048,0.018224888377719455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,2048,0.008747555315494537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,3072,0.00793422261873881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,2048,0.007984889050324758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,1536,0.018073777357737224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,1536,0.007988444632954067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,1536,0.007304888632562425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,768,0.01755288905567593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,1024,0.01681511104106903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,1024,0.005214222189452913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,1024,0.006866666591829724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,2560,0.007320889168315464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,768,0.017159110969967313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,768,0.004566222015354368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,2048,0.0069573331210348345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,768,0.006335999816656113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,512,0.015799999237060547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,512,0.003939555750952827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,512,0.006251555350091722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,1536,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,7168,256,0.016474665866957772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,256,0.003556444413132138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,256,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,1024,0.006228444476922353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,128,0.014761777387724983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,128,0.0031475555151700974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,128,0.005884444548024072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,768,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,64,0.0028275555620590844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,64,0.005913777897755305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,6144,32,0.0028808888875775864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,6144,32,0.006225777582989798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,512,0.005699555493063397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,256,0.00554933316177792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,65536,0.11878044075436062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,12288,0.037624888949924044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,65536,0.12310843997531468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,16384,0.04295911060439216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,16384,0.034274667501449585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,65536,0.07968444294399686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,16384,0.02535377773973677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,12288,0.029151999288135107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,12288,0.020597333709398907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,6144,128,0.005555555638339784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,10240,0.032560000816980995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,10240,0.025331555141343012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,10240,0.018566222654448617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,65536,0.0718017750316196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,8192,0.02981688910060459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,16384,0.022828444838523865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,8192,0.02258844508065118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,8192,0.016767111089494493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,7168,0.027447111076778833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,12288,0.018565333551830716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,7168,0.0196426659822464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,7168,0.015098666151364645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,6144,0.026377777258555096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,6144,0.015134221977657743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,6144,0.014159111513031853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,10240,0.01759022143152025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,5120,0.024352888266245525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,5120,0.013444444371594323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,5120,0.012870222330093384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,8192,0.014178666803571912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,4096,0.023005333211686876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,4096,0.011744889120260874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,7168,0.012992888689041138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,4096,0.010999110837777456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,3584,0.02198933396074507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,3584,0.010625777973069085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,3584,0.01036800030204985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,6144,0.01219555570019616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,3072,0.021191999316215515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,3072,0.010518222219414182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,3072,0.0096142225795322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,5120,0.011271110839313932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,3072,0.021853332718213398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,4096,0.008409777449236976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,2560,0.009699555734793345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,2560,0.02091022166940901
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,2560,0.008309333688682979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,2048,0.017316444052590262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,2048,0.008339555727110969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,2048,0.007183111376232571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,3584,0.007850666840871176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,3072,0.007908444437715743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,1536,0.007072889142566257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,1536,0.007254222200976477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,2560,0.00702400008837382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,1024,0.016235555211702984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,1024,0.005578666511509154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,1024,0.006619555668698416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,2048,0.006959111326270633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,768,0.016823111308945548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,768,0.004556444370084339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,768,0.006603555546866522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,512,0.015461333923869662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,512,0.004097777936193678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,512,0.006064888917737537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,1536,0.006548444430033366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,256,0.015435554915004306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,256,0.0038293335172865125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,256,0.005873777800136142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,1024,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,128,0.014490667316648694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,768,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,128,0.003208000005947219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,128,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,64,0.003198222153716617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,64,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,512,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,5120,32,0.0031662223239739737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,5120,32,0.006220444622966979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,65536,0.08733155330022176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,6144,256,0.014819555812411837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,65536,0.10002488560146755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,256,0.00554666668176651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,16384,0.03631466627120972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,16384,0.030051555898454454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,65536,0.07209866576724581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,16384,0.023337778117921617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,12288,0.03151377704408433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,12288,0.024671110841963027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,12288,0.019541333119074505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,10240,0.028419554233551025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,10240,0.021635555558734473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,10240,0.017828444639841717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,5120,128,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,8192,0.02642311155796051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,8192,0.016166221764352586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,8192,0.01586666703224182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,65536,0.07633599970075819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,7168,0.02421244482199351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,16384,0.021850667066044275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,7168,0.014666666587193808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,7168,0.014152889450391134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,12288,0.017981333865059745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,6144,0.02367288867632548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,6144,0.013042666845851474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,6144,0.013411555853154926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,12288,0.03611288799179925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,5120,0.021934222843911912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,5120,0.011691555380821228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,5120,0.012540444731712341
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,4096,0.021185777253574792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,10240,0.015838222371207345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,4096,0.010657777388890585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,4096,0.01032266683048672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,8192,0.013767999907334646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,3584,0.009122666385438707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,3584,0.009388444324334463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,7168,0.012512889173295764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,3072,0.017657778329319425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,3072,0.008547555241319869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,6144,0.011352000137170156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,3072,0.00853066643079122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,2560,0.01687022215790219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,5120,0.010680889089902243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,2560,0.007595555649863348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,2560,0.007985778152942657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,2048,0.015809777710172866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,4096,0.008293333152929941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,2048,0.006912888752089606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,2048,0.0069546666410234236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,1536,0.016507556041081745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,1536,0.005249777601824866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,1536,0.0070471109615431885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,3584,0.007632888853549957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,1024,0.015634665886561077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,1024,0.00453333349691497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,1024,0.00629333323902554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,3072,0.007594666547245449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,768,0.01682488951418135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,768,0.003999999827808804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,768,0.006339555399285422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,2560,0.007306666837798224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,512,0.014501333236694336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,2048,0.0069120000633928515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,512,0.003563555578390757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,512,0.005953777581453323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,256,0.014815110299322339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,256,0.0032026666320032547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,1536,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,256,0.005888888819350137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,1024,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,768,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,128,0.00315022220214208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,128,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,512,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,64,0.0028577778074476454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,2560,0.018565333551830716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,64,0.00566133318675889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,4096,32,0.002985777747299936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,4096,32,0.0059164443777667145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,256,0.005504888792832692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,65536,0.07818755838606092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,65536,0.09749333063761394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,65536,0.0706968903541565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,16384,0.03563466668128967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,16384,0.026916444301605225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,16384,0.022823111878501043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,12288,0.030626667870415583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,12288,0.02117066747612423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,4096,128,0.005375999957323074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,12288,0.019127110640207928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,10240,0.027274666561020747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,10240,0.01850755512714386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,10240,0.01735733283890618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,65536,0.06820978058709039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,5120,1536,0.01757066614098019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,8192,0.017902221944597032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,16384,0.021591110361946955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,8192,0.015501333607567681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,7168,0.02370044423474206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,7168,0.016021332807011075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,7168,0.014058666096793281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,6144,0.022775999373859827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,12288,0.017543110582563613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,6144,0.014719110396173267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,6144,0.013088000317414602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,5120,0.02169688873820835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,10240,0.015497777197096082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,5120,0.010990222295125326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,5120,0.012327999704413943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,4096,0.01960177719593048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,4096,0.01033155537313885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,4096,0.009735110733244155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,8192,0.013223110801643796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,3584,0.0181368887424469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,7168,0.011705777711338468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,3584,0.009004444711738164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,3584,0.008671999805503422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,3072,0.017497777938842773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,3072,0.008063111040327284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,6144,0.011063999599880643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,3072,0.008295999632941352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,2560,0.016515556308958266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,2560,0.007304000357786815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,5120,0.009130666653315226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,2560,0.007585778004593319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,2048,0.015848888291252982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,2048,0.006165333506133821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,2048,0.006940444310506185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,4096,0.008285333712895712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,1536,0.01682133310370975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,1536,0.005025777965784073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,1536,0.006903999795516332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,3584,0.007996444072988298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,1024,0.015171556009186639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,1024,0.00434933313065105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,3072,0.007621333003044128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,1024,0.00630666646692488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,768,0.01682488951418135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,768,0.0038968887594011095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,2560,0.007263999846246507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,768,0.006009777800904379
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,512,0.014156444205178155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,512,0.0035706667436493766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,2048,0.006615110983451207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,512,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,256,0.014856000741322836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,256,0.0031999999450312722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,256,0.005618666609128316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,1536,0.006270222365856171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,128,0.015088889333936902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,128,0.0031759999692440033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,1024,0.005903111149867375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,128,0.005635555419656966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,64,0.0028088889602157804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,64,0.005587555468082428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3584,32,0.0028124445428450904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,768,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3584,32,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,65536,0.06967377662658691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,512,0.005633777628342311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,65536,0.06838755475150214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,256,0.0052675555149714155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,16384,0.035191999541388616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,16384,0.022639999787012737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,16384,0.022135110365019903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3584,128,0.0053137776752312975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,12288,0.018810666269726224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,12288,0.018808888064490426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,65536,0.06701333655251397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,3584,0.01852888862291972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,10240,0.027056889401541814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,16384,0.02139555580086178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,10240,0.016561778055297006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,10240,0.016846223009957206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,8192,0.025309332542949255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,8192,0.014538667268223233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,8192,0.01480977733929952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,12288,0.017308443784713745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,7168,0.02344177828894721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,7168,0.013624000052611033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,7168,0.013768889009952545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,10240,0.015288000305493673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,6144,0.021928888228204515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,6144,0.012223111258612739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,6144,0.01309511148267322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,8192,0.01293777757220798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,5120,0.020626667473051283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,5120,0.010728889041476779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,5120,0.011715555356608497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,4096,0.018212444252438016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,4096,0.010018666585286459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,4096,0.009352888498041365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,7168,0.010852444503042432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,3584,0.01716711123784383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,3584,0.00886044485701455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,3584,0.00851288851764467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,6144,0.00999111102686988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,3072,0.01680355601840549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,3072,0.008204444415039485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,5120,0.009154666629102495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,3072,0.007990222010347579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,4096,0.008300444318188561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,2560,0.007601777712504069
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,2560,0.007341333561473423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,3584,0.007646222081449296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,2048,0.015504888362354703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,2048,0.006524444454246097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,2048,0.006969778074158563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,4096,128,0.01445777714252472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,1536,0.01588622232278188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,3072,0.007315555380450354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,1536,0.005201777650250329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,1536,0.006974221931563483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,1024,0.01479111115137736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,1024,0.0041946665280395085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,1024,0.00629066675901413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,2560,0.007086222370465596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,768,0.016480889585283067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,768,0.003879111260175705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,768,0.006022222340106964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,2048,0.006607111129495833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,1536,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,512,0.0038897775941424896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,512,0.005662222289376789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,1024,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3584,8192,0.02568000058333079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,256,0.0032133333798911837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,256,0.014835556348164877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,256,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,128,0.014160000615649752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,128,0.0028799999919202593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,128,0.0057751110030545135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,768,0.005800888770156437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,64,0.0028693332440323303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,64,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,3072,32,0.0028764444092909494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,3072,32,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,512,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,65536,0.06005599763658312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,65536,0.09392178058624268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,256,0.0052986666560173035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,16384,0.034049779176712036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,65536,0.06618577904171415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,16384,0.020510221521059673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,16384,0.02163911031352149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,3072,128,0.0052391112678580815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,12288,0.029304001066419814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,12288,0.018047110901938546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,12288,0.01828622155719333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,10240,0.02624977793958452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,10240,0.015815999772813585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,65536,0.06515200270546807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,10240,0.01629599928855896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,8192,0.0244195560614268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,16384,0.020840888222058613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,8192,0.014071111877759298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,8192,0.014705777168273926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,12288,0.0169804443915685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,7168,0.013399110900031196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,7168,0.013439999686347114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,10240,0.014529777897728814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,6144,0.021327111456129286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,6144,0.011508444117175208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,8192,0.01185155577129788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,6144,0.012270222107569376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,5120,0.020097777247428894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,5120,0.010341333018408881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,5120,0.010796444283591377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,7168,0.010657777388890585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,6144,0.009701333112186855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,4096,0.01792977750301361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,4096,0.009045333498054082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,4096,0.008302222523424361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,5120,0.008967111508051554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,3584,0.016511999898486666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,3584,0.008338666624493068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,3584,0.008050666915045844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,65536,0.09540533357196385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,3072,0.016189333465364244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,3072,0.00794400026400884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,3072,0.00792622235086229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,2560,0.015801777442296345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,4096,0.007987555530336168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,2560,0.006739555547634761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,2560,0.007294221884674496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,12288,0.030230220821168687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,2048,0.005863111052248213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,3584,0.007648888561460707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,2048,0.006950221955776215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,1536,0.015435554915004306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,3072,0.007320889168315464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,1536,0.004855111241340637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,1536,0.006601777755551868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,2560,0.007084444165229797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,1024,0.014500444134076437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,1024,0.004086222085687849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,1024,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,768,0.016185777054892648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,768,0.0038417776425679526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,2048,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,768,0.006292444550328785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,1536,0.0062871111763848206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,512,0.0035031110876136353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,512,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,1024,0.005877333382765452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,256,0.014832889040311178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,256,0.0032195556494924757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,256,0.005906666732496685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,768,0.005882666756709416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,128,0.014031110538376702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,128,0.003149333306484752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,128,0.005695111221737332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,512,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,64,0.002858666703104973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,64,0.005923555543025334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2560,32,0.002793777734041214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,256,0.005281777845488654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2560,32,0.005721777677536011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,65536,0.05041244294908312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,65536,0.08662755621804132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2560,128,0.0052693333062860705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,65536,0.06399644745720758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,16384,0.03267822331852383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,16384,0.018217777212460835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,16384,0.020945777495702107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,65536,0.0646231108241611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,12288,0.015117333994971381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,12288,0.01774488886197408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,16384,0.020569778150982328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,2560,0.016207110550668504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,10240,0.025482666161325242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,10240,0.014081777797804939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,10240,0.01630311045381758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,8192,0.023696889479955036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,8192,0.013148444394270578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,8192,0.013743110828929476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,12288,0.016195555528004963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,7168,0.02180533276663886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,7168,0.011903111305501727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,10240,0.013062222136391534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,7168,0.013014222184816996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,6144,0.020623111062579684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,6144,0.011160888605647616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,6144,0.011178666518794166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,8192,0.01109866632355584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,5120,0.018201778332392376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,5120,0.010344888601038191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,5120,0.00963466697269016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,7168,0.00997866690158844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,4096,0.017280000779363845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,4096,0.011404444773991903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,4096,0.008341333104504479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,6144,0.009320889082219865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,3584,0.016521778371598985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,3584,0.009660444325870937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,3584,0.008005333443482717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,3072,512,0.014010666145218743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,5120,0.008625778059164682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,3072,0.016199999385409884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,3072,0.008619555168681675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,3072,0.007710222568776872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,2560,0.015644444359673392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,4096,0.007987555530336168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,2560,0.0076248885856734375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,2560,0.007268444531493717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,3584,0.007656888829337225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,2048,0.005555555638339784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,2048,0.0069208890199661255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,3072,0.0074684446056683855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,2560,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,1536,0.015823110938072205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,1536,0.004917333523432414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,1536,0.006599999964237213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,2048,0.006612444503439798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,1024,0.014496889379289416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,1024,0.004211555752489301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,1024,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,1536,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,768,0.016543111867374845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,7168,0.022626666559113398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,768,0.0039022221333450745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,768,0.006226666685607698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,512,0.013111111190583972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,512,0.00351200004418691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,512,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,1024,0.006014222072230444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,256,0.0151582227812873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,256,0.003192000091075897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,256,0.005859555469618903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,768,0.005632888939645555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,128,0.014843554960356818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,128,0.003173333282272021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,128,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,512,0.005571555760171678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,64,0.0028657778683635923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,64,0.00591111100382275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,2048,32,0.0028702221396896574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,2048,32,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,65536,0.03932888971434699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,256,0.005355555564165115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,65536,0.08506666951709324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,65536,0.062136001057095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,16384,0.03206844462288751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,16384,0.017961778574519686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,2048,128,0.005247999810510212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,16384,0.02049422264099121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,12288,0.015290665957662793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,12288,0.027850667635599773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,12288,0.0173662222094006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,10240,0.02472355630662706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,65536,0.06371377574072944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,10240,0.01389955480893453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,10240,0.015554666519165039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,16384,0.019872888922691345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,8192,0.021793777743975323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,8192,0.01239377756913503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,8192,0.012712000144852532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,12288,0.014553777045673795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,7168,0.019924443629052904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,7168,0.011750222080283694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,7168,0.011001778145631155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,10240,0.012376888758606382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,6144,0.011055111057228513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,6144,0.010335110955768162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,8192,0.011024889018800525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,5120,0.017831999394628737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,7168,0.00999644481473499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,5120,0.009959110783206092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,5120,0.009448889229032729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,4096,0.01720711092154185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,4096,0.008633777499198914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,6144,0.009332444104883406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,2048,0.014929778046078153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,4096,0.00832088871134652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,3584,0.015828443898095023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,3584,0.007298666569921706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,3584,0.007975111405054728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,5120,0.008614222208658854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,3072,0.015504888362354703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,3072,0.006626666833957036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,3072,0.00759733302725686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,4096,0.00795111142926746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,2560,0.005963555640644497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,2560,0.007279111279381647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,3584,0.007626666790909237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,2048,0.014836443795098199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,2048,0.005261333452330695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,3072,0.007244444555706448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,2048,0.0069573331210348345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,1536,0.015233778291278414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,1536,0.005591111050711737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,2560,0.00665244460105896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,1536,0.006616000086069107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,1024,0.014129777749379476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2560,512,0.013472889032628802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,1024,0.0042168887125121225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,1024,0.00629688882165485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,2048,0.006593777901596493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,768,0.0038568890757030914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,768,0.015198222464985318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,768,0.006225777582989798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,512,0.013265777793195514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,1536,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,512,0.003499555504984326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,512,0.005919999960396025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,256,0.014484445254007975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,1024,0.005920889063013925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,256,0.0032115555885765287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,256,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,768,0.005772444523043103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,128,0.013807999591032663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,512,0.0055440002017551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,128,0.002850666642189026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,128,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,64,0.0028124445428450904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,64,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,256,0.0052426668504873914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1536,32,0.0028444443725877335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1536,32,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,12288,0.027878221538331773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,65536,0.028470221492979262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,65536,0.08414489030838013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1536,128,0.00553688903649648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,65536,0.06028088596132067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,16384,0.03156800071398417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,16384,0.013314666847387949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,16384,0.01959822244114346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,12288,0.02595377796226078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,12288,0.011058666639857821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,65536,0.06322844160927667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,12288,0.01607377827167511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,10240,0.011506666739781698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,16384,0.017383999294704862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,10240,0.014490667316648694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,8192,0.020633776982625324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,12288,0.013808888693650564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,8192,0.013783111340469785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,8192,0.011390222443474663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,10240,0.012054222325483957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,7168,0.019332443674405415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,7168,0.012098666694429187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,7168,0.010664888554149203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,6144,0.018624888526068795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,6144,0.011405333048767514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,6144,0.009868444667922126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,8192,0.010752889017264048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,5120,0.010209778116808997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,5120,0.01756088932355245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,5120,0.009180444810125563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,7168,0.009890666438473595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,4096,0.016514667206340365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,4096,0.007995555798212687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,4096,0.008106666306654612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,6144,0.009294222626421187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,3584,0.006950221955776215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,3584,0.007932444413503012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,5120,0.008619555168681675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,3072,0.015473778049151102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,4096,0.007737778127193451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,3072,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,3072,0.007616889145639207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,2560,0.014449778530332776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,3584,0.007484444313579136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,2560,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,2560,0.007247111035717859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,3072,0.007272000114123027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,2048,0.013756444056828817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,2048,0.006304889089531369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,2048,0.0069146665434042616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,1536,0.013811555173661975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,2560,0.006704888823959563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,1536,0.0059368887709246735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,1536,0.006619555668698416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,1024,0.013362666798962487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,2048,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,1024,0.004537777768241035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,1024,0.006245333287451003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,1536,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,768,0.015445333388116626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,768,0.0041555555330382455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,768,0.0059573331640826324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,512,0.012768000364303589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,1024,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,512,0.0035235554807715942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,512,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,2048,2048,0.014813333749771118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,256,0.014472888575659858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,768,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,256,0.003181333343187968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,256,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,128,0.014494222071435718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,128,0.0031902222997612427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,128,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,512,0.005554666535721884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,64,0.002826666666401757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,64,0.005615111026499007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,1024,32,0.0028088889602157804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,1024,32,0.0058604445722368025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,256,0.005212444398138259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,65536,0.023451555106374953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,65536,0.08348177538977729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,65536,0.059446222252315946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,1024,128,0.005197333378924264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,16384,0.03019022279315525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,16384,0.012525333298577202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,16384,0.018280888597170513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,12288,0.025047999289300706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,12288,0.011011555790901184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,65536,0.06282577912012736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,12288,0.01442311041884952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,16384,0.016525333126386006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,10240,0.022230222821235657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,10240,0.00981155534585317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,10240,0.012706666356987424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,12288,0.013731555806265937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,8192,0.020277332928445604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,8192,0.011064888702498542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,8192,0.011016888750924004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,10240,0.01204177737236023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,7168,0.019152889649073284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,7168,0.00999022192425198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,7168,0.01069155583779017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,6144,0.018168000711335074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,8192,0.010854221880435944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,6144,0.009650666680600908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,6144,0.01000355515215132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,7168,0.009664889011118147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,5120,0.017183999220530193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,5120,0.008635555704434713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,5120,0.009296000003814697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,4096,0.016700444949997794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,6144,0.009293333523803288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,4096,0.0076142218377855085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,4096,0.008079999850855933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,5120,0.008642666869693333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,3584,0.015485333071814643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,3584,0.006912888752089606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,3584,0.0079297779334916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,4096,0.008005333443482717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,3072,0.015112888481881885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,3072,0.006599111275540457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,3072,0.007630222373538547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,3584,0.007431999676757389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,2560,0.014892444014549255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,2560,0.005617777920431561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,2560,0.007249777515729268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,3072,0.007281777759393056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,2048,0.01443733274936676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,2048,0.005200888961553574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,2048,0.006575111299753189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,2560,0.0069448889957533936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,1536,0.013738666971524557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,1536,0.004522666749027041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,1536,0.006424888968467712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,1024,0.0131217779384719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,2048,0.006594666590293248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,1024,0.0038568890757030914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,1024,0.006229333165619109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,768,0.015231110983424716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,768,0.003580444388919406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,1536,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,768,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,1024,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,512,0.012621333201726278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,512,0.003488000068399641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,512,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,256,0.0031475555151700974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,256,0.014654222461912366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,768,0.005537777725193236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,256,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,512,0.0052471111218134565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,768,128,0.013438222308953604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,128,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,128,0.002789333255754577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,6144,0.018927110566033255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,64,0.002806222273243798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,256,0.0052533331844541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,64,0.0058355554938316345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,768,32,0.0025235555238193935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,768,32,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,768,128,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,65536,0.022864888111750286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,65536,0.0823999974462721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,65536,0.05961689021852282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,16384,0.02918133470747206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,16384,0.01308444473478529
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,16384,0.01721244388156467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,65536,0.0627982219060262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,12288,0.024332443873087566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,12288,0.011006222003036074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,12288,0.014430221584108142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,10240,0.021652445197105408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,16384,0.0166311115026474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,10240,0.010216000179449717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,10240,0.012758221891191272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,12288,0.013402666482660504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,8192,0.02012266715367635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,8192,0.008647999829716152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,8192,0.011137777732478248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,10240,0.01204355557759603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,7168,0.01886399918132358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,7168,0.008103110724025303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,7168,0.010699555277824402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,8192,0.010718222293588849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,6144,0.007622222105662028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,6144,0.018075555562973022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,6144,0.010010666317409938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,7168,0.009706666900051964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,5120,0.016951110627916124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,5120,0.008273777862389883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,6144,0.009299555586444007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,5120,0.009095999929640029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1536,2560,0.0148053334818946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,4096,0.007599111232492659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,5120,0.008635555704434713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,4096,0.00833333366447025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,3584,0.014811555544535318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,3584,0.0069857777820693115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,3584,0.00797155582242542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,4096,0.007661333514584436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,3072,0.015092444088723926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,3072,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,3072,0.0074035558435651995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,3584,0.00758755538198683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,2560,0.014454222387737699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,2560,0.005583111196756363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,2560,0.0069208890199661255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,3072,0.007272000114123027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,2048,0.013440888788965015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,2560,0.0069324444565508105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,2048,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,2048,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,1536,0.013704000247849358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,1536,0.006591111007663939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,1536,0.0046533333758513136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,1024,0.003966222206751506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,2048,0.0064151109092765385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,1024,0.01275199982855055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,1024,0.005895111295912001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,1536,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,768,0.014476444986131458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,768,0.003531555748648114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,768,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,1024,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,512,0.011753777662913004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,512,0.0035031110876136353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,768,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,512,0.0058853332367208265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,256,0.013747555514176687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,256,0.003162666741344664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,256,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,512,0.005562666803598404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,128,0.012423999607563019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,128,0.0028391112056043413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,64,0.0028035555862718155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,128,0.005542222410440445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,256,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,64,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,512,32,0.002817777709828483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,512,128,0.005387555393907759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,512,32,0.005568888866239124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,65536,0.01791733337773217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,65536,0.08211288849512736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,65536,0.05834488736258613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,16384,0.02779910961786906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,65536,0.06149333053165012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,16384,0.010335110955768162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,10240,0.022852443986468848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,16384,0.015799111790127225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,12288,0.023992000354660883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,16384,0.016186666157510545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,12288,0.009648000200589498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,10240,0.01238666640387641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,12288,0.013814222481515674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,10240,0.02091911103990343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,12288,0.01369244439734353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,10240,0.008969777988062965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,8192,0.019320888651741874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,8192,0.008291555775536431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,10240,0.01203733351495531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,8192,0.010657777388890585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,7168,0.017998221847746108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,8192,0.01075466639465756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,7168,0.007972444097201029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,7168,0.010382222632567087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,6144,0.016935999194780987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,7168,0.009706666900051964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,6144,0.008344888687133789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,6144,0.00977422214216656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,1024,3584,0.01554844445652432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,5120,0.016743110285864938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,6144,0.009335999687512716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,5120,0.0076142218377855085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,5120,0.009297777381208207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,4096,0.01569333341386583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,5120,0.00867911097076204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,4096,0.0069848886794514125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,4096,0.008120888637171851
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,3584,0.014469333820872836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,4096,0.007631999750932057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,3584,0.0069155556460221606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,3584,0.008031999899281396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,3584,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,3072,0.006646222124497096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,3072,0.007294221884674496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,3072,0.0070977773931291364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,2560,0.0064088888466358185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,2560,0.007263999846246507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,2048,0.0052435555391841466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,2048,0.0127697777416971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,2560,0.0069155556460221606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,2048,0.006731555693679386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,1536,0.012151999606026543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,2048,0.00638133328821924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,1536,0.00452622233165635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,1536,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,1024,0.011759999725553723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,1024,0.003859555555714501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,1536,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,1024,0.006263111200597551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,768,0.014094221923086377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,768,0.0059573331640826324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,1024,0.005591111050711737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,768,0.0034968890249729156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,512,0.01204355557759603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,512,0.0031866667171319327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,768,0.005562666803598404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,512,0.005592888842026393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,256,0.002826666666401757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,256,0.011715555356608497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,512,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,256,0.0058124446206622655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,128,0.0027742222365405825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,256,0.0052044445441828836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,128,0.011446221835083432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,128,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,64,0.005518222020732031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,64,0.0028133332315418455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,256,32,0.002887111157178879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,256,128,0.005280888742870755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,256,32,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,65536,0.08132888873418172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,65536,0.015522667103343539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,65536,0.05759466356701321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,16384,0.027034666803148057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,65536,0.05877777602937487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,16384,0.009322666459613377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,16384,0.016480889585283067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,16384,0.016176000237464905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,12288,0.00832711077398724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,12288,0.013810666898886362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,12288,0.01370044466522005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,10240,0.02026844521363576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,10240,0.009024889104896123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,10240,0.01239822225438224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,10240,0.011977777712874942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,8192,0.0079644446571668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,8192,0.010775110787815519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,8192,0.0107351111041175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,7168,0.00795199970404307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,7168,0.010524444282054901
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,6144,0.017215111189418368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,7168,0.009757333331637913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,6144,0.007263999846246507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,6144,0.00941777808798684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,5120,0.0158595558669832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,6144,0.009030222064918941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,5120,0.00692622239391009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,5120,0.008996444443861643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,5120,0.00866844422287411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,4096,0.015461333923869662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,4096,0.006595555692911148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,4096,0.008296888735559251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,3584,0.014501333236694336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,4096,0.007672888537247975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,3584,0.00657066661450598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,3584,0.007771555748250749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,3584,0.007632888853549957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,3072,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,3072,0.007631111476156447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,3072,0.007259555160999298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,2560,0.013409777647919126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,2560,0.006286222073766921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,2560,0.007259555160999298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,2560,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,2048,0.013061333033773633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,2048,0.006232000059551663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,2048,0.006746666712893381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,2048,0.006584888945023219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,1536,0.01274222218328052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,1536,0.004530666602982415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,1536,0.00666222224632899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,1536,0.006124444305896759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,1024,0.012058667010731168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,1024,0.0038453332251972626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,1024,0.006246222390068903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,1024,0.005883555445406172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,768,0.01273333364062839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,512,4096,0.01612977849112617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,768,0.0035146665241983202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,768,0.006167999986145232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,512,0.011702222128709158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,768,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,512,0.003167111012670729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,512,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,512,0.005581333405441708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,256,0.011419555379284753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,256,0.0027964444210131965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,256,0.0058853332367208265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,256,0.005227555417352253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,128,0.002571555475393931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,128,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,80,128,128,0.005234666582610872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,64,0.0028035555862718155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,64,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,128,32,0.002598222138153182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,128,32,0.0058808889653947614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,65536,0.015043555034531487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,65536,0.056792888376447886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,16384,0.008329778081840938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,16384,0.01646133263905843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,12288,0.0075804442167282104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,12288,0.014096000128322177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,10240,0.007239999870459239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,10240,0.012410666379663678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,8192,0.006622222148709827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,8192,0.011178666518794166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,7168,0.006291555447710886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,7168,0.010440000229411654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,6144,0.006321777900060018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,2560,0.013811555173661975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,6144,0.009692444569534725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,5120,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,5120,0.009104889300134447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,4096,0.005960000058015187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,4096,0.00832622249921163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,3584,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,3584,0.007932444413503012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,3072,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,3072,0.007423111134105259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,2560,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,2560,0.007303111255168915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,2048,0.005879111174080107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,2048,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,1536,0.004555555681387584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,1536,0.006265777680608962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,1024,0.0038453332251972626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,1024,0.006239999913507038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,256,3072,0.014096000128322177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,768,0.0034888887570963967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,768,0.006217777729034424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,512,0.003126222226354811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,512,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,256,0.002879111096262932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,256,0.005584888988071018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,128,0.005530666559934616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,128,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,64,0.002536888958679305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,64,0.005884444548024072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,64,32,0.002532444480392668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,64,32,0.005575110928879843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,65536,0.014800889624489678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,65536,0.05549155672391256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,16384,0.008004444340864817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,16384,0.01628977722591824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,12288,0.007612444460391998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,12288,0.013445333474212222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,10240,0.007241778075695038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,10240,0.012714666624863943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,8192,0.006797333144479328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,8192,0.011037333144081963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,7168,0.006605333338181178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,7168,0.0099982221921285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,6144,0.006611555400821898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,6144,0.009706666900051964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,5120,0.006051555689838197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,5120,0.009332444104883406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,12288,0.02299555473857456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,4096,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,4096,0.007820444802443186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,3584,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,3584,0.007606222397751278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,3072,0.006228444476922353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,2560,0.007072000039948358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,3072,0.007319110963079665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,2560,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,2048,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,2048,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,1536,0.00423200014564726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,1024,0.005931555396980709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,1536,0.006637333167923822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,1024,0.0038533334930737815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,768,0.003556444413132138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,768,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,256,0.002934222213096089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,512,0.0031519999934567344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,128,0.0029111111329661477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,512,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,256,0.005531555662552516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,64,0.005912888795137405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,128,0.005587555468082428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,64,0.0025253333151340485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,80,32,32,0.002806222273243798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,8192,0.018924444913864136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,80,32,32,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,7168,0.018000889155599806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,16384,0.36454399426778156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,3072,0.01351377781894472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,12288,0.273784875869751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,16384,0.23565954632229277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,16384,0.17099200354682076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,12288,0.18820977210998535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,12288,0.12780443827311197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,10240,0.23006754451327852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,10240,0.1630328893661499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,10240,0.10799200005001491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,8192,0.18323200278811982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,8192,0.0908471080991957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,7168,0.12935466236538357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,7168,0.16124178303612605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,7168,0.07747822337680392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,6144,0.11930755774180095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,80,128,128,0.011698666546079846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,6144,0.13981510533226862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,6144,0.06803733110427856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,5120,0.10696533653471206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,5120,0.11892355812920465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,5120,0.05759555763668484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,8192,0.14679466353522405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,4096,0.0963111122449239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,4096,0.09814044502046372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,4096,0.0470106667942471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,16384,0.147070222430759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,12288,0.1028791136211819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,3584,0.09049777852164374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,3584,0.08361422353320652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,3584,0.04227288895183139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,3072,0.08429688877529568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,3072,0.07455378108554415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,3072,0.037273777855767146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,8192,0.07077155510584514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,10240,0.0958862238460117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,2560,0.06273333231608073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,2560,0.07828622394137912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,2560,0.03241422110133701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,7168,0.06218044625388252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,2048,0.050845334927241005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,2048,0.027817779117160376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,6144,0.054025779167811074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,1536,0.0682871143023173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,1536,0.03980266716745164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,1536,0.022651554809676275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,5120,0.04639200038380093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,1024,0.06218577755822075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,1024,0.02811288833618164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,1024,0.018590221802393597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,2048,0.07390133539835612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,768,0.030138668086793687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,768,0.022819555468029443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,768,0.016515556308958266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,4096,0.03958400090535482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,3584,0.035453332795037165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,512,0.015770667129092746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,512,0.015271999769740634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,3072,0.030464887619018555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,256,0.024833778540293377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,256,0.009715555442704095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,256,0.013724444641007317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,2560,0.028699556986490887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,128,0.021930666433440313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,128,0.006686222222116258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,128,0.010388444695207808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,2048,0.02278933260175917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,64,0.00453955555955569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,64,0.010999999940395355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,65536,32,0.0046986668474144405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,65536,32,0.01400888959566752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,65536,0.23085776964823404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,1536,0.019075555933846366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,1024,0.015498666299713982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,65536,0.3493653403388129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,65536,0.18423467212253145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,16384,0.07443377706739637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,16384,0.08327555656433105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,16384,0.051092445850372314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,12288,0.060566220018598765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,12288,0.0644435551431444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,12288,0.03978577918476529
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,65536,512,0.05651377969317966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,10240,0.05362311005592346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,10240,0.053671999110115894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,10240,0.03474044468667772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,65536,0.1612186696794298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,8192,0.04820977648099264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,8192,0.045200887653562755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,16384,0.046104888121287026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,8192,0.029699554045995075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,7168,0.04389422138532003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,7168,0.03894933395915561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,7168,0.02608266638384925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,12288,0.03426933288574219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,6144,0.04144800040456984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,6144,0.03415022293726603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,6144,0.02383022175894843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,10240,0.029085334804322984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,5120,0.037640889485677086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,5120,0.029215110672844782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,5120,0.020558221472634208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,768,0.014055111342006259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,8192,0.024401777320437964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,4096,0.03479199939303928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,4096,0.025052444802390203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,4096,0.017323555217848886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,7168,0.023644444015291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,3584,0.03314044409328037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,3584,0.021652445197105408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,3584,0.015970667203267414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,6144,0.02049688829316033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,3072,0.031719999180899724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,3072,0.019170666734377544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,3072,0.01463822192615933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,2560,0.03000977635383606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,2560,0.016210666961140104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,2560,0.013081777426931592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,512,0.012659555508030785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,2048,0.029317332638634577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,2048,0.013253333667914072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,2048,0.01201777739657296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,5120,0.01757066614098019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,256,0.011607999602953592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,1536,0.01071911139620675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,1536,0.01094133324093289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,4096,0.015841777125994366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,1024,0.02609155575434367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,1024,0.008886222210195329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,1024,0.008820444345474243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,3584,0.014582221706708273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,65536,128,0.011049778097205691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,768,0.007280888656775157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,768,0.007080888582600488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,3072,0.013628444737858243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,2560,0.012870222330093384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,512,0.005486222190989389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,512,0.0069057775868309875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,2048,0.011147555377748279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,256,0.01755555636352963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,256,0.0039048890272776284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,256,0.006578666468461354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,1536,0.00979288915793101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,128,0.015150222513410779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,128,0.003671110918124517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,128,0.0063031112982167145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,1024,0.007601777712504069
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,64,0.0031831111345026228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,64,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,16384,32,0.003536888708670934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,16384,32,0.007257777783605788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,768,0.007231999602582719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,65536,0.1726977825164795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,65536,0.24971021546257866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,512,0.0069475554757648045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,16384,0.05958844555748833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,65536,0.1619271172417535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,16384,0.06828088892830743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,16384,0.04440444376733568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,12288,0.049294223388036094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,12288,0.053322666221194796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,12288,0.03538844320509169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,256,0.006295111030340195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,10240,0.04392888810899523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,10240,0.04400533437728882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,10240,0.030840890275107488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,16384,128,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,8192,0.0403439998626709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,8192,0.036488887336519026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,65536,0.1411235597398546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,8192,0.02687377731005351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,7168,0.036143110858069524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,7168,0.032250665956073336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,7168,0.023616888456874426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,16384,0.04110666778352525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,6144,0.034381333324644305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,1536,0.02791022260983785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,6144,0.0281644430425432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,6144,0.021312000022994146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,12288,0.03264355659484863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,5120,0.02432800001568264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,5120,0.018851555056042142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,512,0.022276444567574397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,16384,768,0.019317333896954853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,4096,0.029792000850041706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,4096,0.021143999364640977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,4096,0.016136889656384785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,3584,0.028593778610229492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,3584,0.017986666825082567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,3584,0.014912888407707214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,10240,0.026932444837358262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,3072,0.027457776996824477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,3072,0.016201777590645682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,3072,0.013782222237851886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,8192,0.023000000251664057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,2560,0.026320889592170715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,2560,0.013724444641007317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,2560,0.01274222218328052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,7168,0.020555555820465088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,6144,0.01984444426165687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,2048,0.011696000066068439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,2048,0.011683555940786997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,5120,0.0166311115026474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,4096,0.014820444915029736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,1536,0.0248346659872267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,1536,0.00980888886584176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,1536,0.010032888915803697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,3584,0.013789333403110504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,1024,0.02221155497762892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,1024,0.0081386665503184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,1024,0.00793422261873881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,3072,0.013428444663683573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,768,0.006609777609507243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,768,0.018065777089860704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,768,0.0069884442620807225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,512,0.01993333299954732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,512,0.004850666556093427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,512,0.006782222125265334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,2560,0.012062222593360476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,256,0.017156443662113614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,256,0.003847111016511917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,256,0.006259555617968242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,2048,0.01034222212102678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,128,0.01483022173245748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,128,0.0034968890249729156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,128,0.0062871111763848206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,1536,0.008591999610265097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,64,0.003228444399105178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,64,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,12288,32,0.003843555433882607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,12288,32,0.007266666326257918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,1024,0.007273777491516537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,65536,0.22137778335147432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,768,0.007103111180994246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,65536,0.1546293364630805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,512,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,16384,0.05417422122425503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,16384,0.0543759995036655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,16384,0.04289066791534424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,65536,0.15134755770365396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,12288,0.04240622123082479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,12288,0.04537333382500542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,12288,0.03381955623626709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,10240,0.03991644581158956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,10240,0.03522400061289469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,10240,0.029554665088653564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,256,0.006254222244024277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,8192,0.037343111303117536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,8192,0.029823111163245306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,8192,0.02528088953759935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,12288,128,0.00629333323902554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,7168,0.033482667472627424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,7168,0.025742222865422566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,5120,0.031665778822369046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,7168,0.022612444228596155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,6144,0.03200177682770623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,6144,0.022777777579095628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,6144,0.02056533263789283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,65536,0.13417066468132868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,16384,0.037788444095187716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,5120,0.019810666640599568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,5120,0.018391110830836825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,12288,0.029375112719006006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,10240,0.02530577778816223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,4096,0.017143110434214275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,4096,0.01556711064444648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,8192,0.02253688871860504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,3584,0.026591110560629103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,3584,0.014807111687130399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,3584,0.014511111709806653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,12288,2048,0.02563200063175625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,7168,0.019671999745898776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,3072,0.025433777107132807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,3072,0.01312711089849472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,3072,0.013551111022631327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,2560,0.024508444799317256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,6144,0.01789688898457421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,2560,0.011379555695586734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,2560,0.012093333734406365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,2048,0.023803555303149756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,2048,0.009638222555319468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,2048,0.011086222198274402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,5120,0.015855999456511605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,1536,0.023688889212078516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,1536,0.008150222400824228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,1536,0.009420444567998251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,4096,0.014490667316648694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,1024,0.019924443629052904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,1024,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,1024,0.00720088928937912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,3584,0.013384000294738345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,3072,0.012432888978057437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,768,0.005211555709441503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,768,0.006963555183675554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,2560,0.010965333216720156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,512,0.018244443668259513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,512,0.0041884444653987885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,512,0.006602666858169768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,2048,0.008730666504965888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,256,0.016511110795868766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,256,0.003875555677546395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,256,0.006254222244024277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,1536,0.007967111137178209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,1024,0.007251555720965068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,128,0.015453333655993143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,128,0.003489777859714296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,128,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,64,0.003177777760558658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,64,0.006579555571079254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,10240,32,0.00351200004418691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,768,0.006683555742104848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,10240,32,0.0069679998689227635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,512,0.006294222341643439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,65536,0.18290932973225912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,65536,0.14013422860039607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,65536,0.10055289003584121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,16384,0.04904177784919739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,256,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,16384,0.049308445718553334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,16384,0.031152000029881794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,12288,0.03414222266938951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,12288,0.024461333950360615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,5120,0.029491557015313044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,10240,128,0.006235555642180973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,10240,0.02827733423974779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,10240,0.03631733192337884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,10240,0.02186133298608992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,8192,0.03353600038422479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,8192,0.02403644389576382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,8192,0.018530666828155518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,65536,0.08306666877534655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,16384,0.026220444175932143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,7168,0.020956445071432326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,7168,0.017234666479958426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,4096,0.027951998843087092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,6144,0.029119110769695703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,12288,0.021054221524132624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,6144,0.018904889623324077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,6144,0.015662221444977652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,5120,0.027149332894219294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,5120,0.016139555308553908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,5120,0.01420177850458357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,10240,0.01849422189924452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,4096,0.025623111261261836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,4096,0.013847110999955071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,4096,0.012087999946541257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,8192,0.016235555211702984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,7168,0.014835556348164877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,3584,0.013068444199032254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,3584,0.011696889168686338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,6144,0.014953777194023132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,3072,0.02379911144574483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,3072,0.011918221910794577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,3072,0.010919111470381418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,5120,0.01253600004646513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,2560,0.022516444325447083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,2560,0.010771555205186209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,2560,0.01015644437736935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,10240,768,0.018409777018758986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,2048,0.02196888956758711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,2048,0.008997333546479544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,2048,0.009339555270142024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,4096,0.011121778024567498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,1536,0.020069332586394418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,1536,0.00794311116139094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,1536,0.007198221981525421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,3584,0.010201777848932479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,1024,0.018564444449212816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,1024,0.006065777606434292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,1024,0.006256888724035687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,3072,0.009140444298585257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,768,0.01756088932355245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,768,0.004703999807437261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,768,0.006008000009589725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,2560,0.008261333737108443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,512,0.017098666893111337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,512,0.0039013334446483185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,512,0.005621333503060871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,2048,0.007377777662542131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,256,0.01686755485004849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,256,0.0032142222755485107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,256,0.005696889013051987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,1536,0.006962666908899943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,128,0.014704000618722705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,128,0.0032115555885765287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,128,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,1024,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,64,0.003185777821474605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,64,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,8192,32,0.003530666646030214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,8192,32,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,768,0.005959110955397288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,65536,0.15637244118584528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,65536,0.133443554242452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,512,0.00592533333433999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,16384,0.04684355523851183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,65536,0.09219466977649265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,16384,0.0461733341217041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,256,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,12288,0.04078133238686456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,16384,0.028331554598278467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,12288,0.03903911179966397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,12288,0.03576533330811395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,12288,0.02295200030008952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,10240,0.035387555758158364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,10240,0.027122666438420612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,8192,128,0.005562666803598404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,10240,0.020498666498396132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,8192,0.031938665442996554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,8192,0.022934221559100684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,8192,0.017664000391960144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,7168,0.03057066599527995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,7168,0.02918222215440538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,65536,0.08364533053504096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,7168,0.020115555988417733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,7168,0.01611288885275523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,6144,0.028424001402325098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,16384,0.025120000044504803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,6144,0.01752622259987725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,6144,0.015035554766654968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,5120,0.02596355477968852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,12288,0.02047288914521535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,5120,0.01532444523440467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,5120,0.013553777502642738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,4096,0.013031110995345645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,10240,0.018048889107174344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,4096,0.011777777638700275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,3584,0.02367199957370758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,3584,0.011729777687125735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,3584,0.011237333218256632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,8192,0.015374221735530429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,3072,0.022999111149046157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,3072,0.011343110766675739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,3072,0.010797333386209277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,8192,3584,0.02437777817249298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,2560,0.02225333286656274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,2560,0.009929777847396003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,2560,0.009679111341635386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,7168,0.013864888913101621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,6144,0.014095111025704278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,2048,0.008630221916569604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,2048,0.008062221937709385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,5120,0.012101333174440594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,1536,0.019630221856964957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,1536,0.007432888779375289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,4096,0.010651555326249866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,1536,0.006640888750553131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,1024,0.017695110705163743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,1024,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,1024,0.006365333166387346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,3584,0.009310222334331935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,768,0.017479111750920612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,768,0.00434311106801033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,768,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,3072,0.008059555457697975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,512,0.016898666818936665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,512,0.0039022221333450745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,512,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,2560,0.007596444752481248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,256,0.01681866745154063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,256,0.0035662220584021676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,256,0.005569777968857024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,2048,0.0069351109365622205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,128,0.01512888901763492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,128,0.0034933334423436057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,128,0.005616888817813661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,1536,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,64,0.0032008888406885993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,64,0.0058559998869895935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,7168,32,0.0031902222997612427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,7168,32,0.005919999960396025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,1024,0.0063075555695427795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,65536,0.12519644366370306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,768,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,65536,0.12626400258806017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,512,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,16384,0.04421155651410421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,65536,0.08619822396172418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,16384,0.03953155544069078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,16384,0.02660977840423584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,12288,0.03695911169052124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,12288,0.032462222708596125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,12288,0.02163822286658817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,256,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,10240,0.02832177612516615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,10240,0.01953066719902886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,7168,128,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,8192,0.030416889323128596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,8192,0.02035466664367252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,8192,0.016347555650605094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,65536,0.07817688915464613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,7168,0.02846133377816942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,7168,0.02126400007141961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,16384,0.02373955481582218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,7168,0.015372445185979208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,6144,0.02718399961789449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,12288,0.019478221734364826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,6144,0.018790221876568265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,6144,0.014294221997261047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,5120,0.02497600018978119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,5120,0.016798221402698092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,10240,0.017147555947303772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,5120,0.013030221892727746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,8192,0.014715555641386243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,4096,0.02369155486424764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,4096,0.014420444766680399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,4096,0.011383111278216044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,7168,0.013714666995737286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,6144,0.013391111459996967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,3584,0.012690666649076672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,3584,0.010667555862002902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,5120,0.011748444702890186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,3072,0.02169955604606205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,3072,0.009677333136399588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,3072,0.01019377758105596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,2560,0.02068177858988444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,2560,0.008775110873911116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,2560,0.008964444200197855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,4096,0.009654222263230218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,2048,0.018552000323931377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,2048,0.007612444460391998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,2048,0.007630222373538547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,4096,0.024850666522979736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,3584,0.00869511150651508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,1536,0.018577777677112155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,1536,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,1536,0.006626666833957036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,1024,0.016881777180565726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,1024,0.004560889055331548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,1024,0.006253333141406377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,3072,0.00794311116139094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,768,0.017376888129446242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,768,0.0041724443435668945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,768,0.005850666513045629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,2560,0.007327111230956183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,512,0.016153777639071148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,7168,2048,0.02032888929049174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,512,0.003552000141806073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,512,0.005910222315125995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,256,0.014843554960356818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,2048,0.0069226668112807805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,256,0.0034924443397257063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,256,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,128,0.014833778142929077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,128,0.003165333428316646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,128,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,1536,0.006574222197135289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,64,0.0028782222006056043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,64,0.005927111125654644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,6144,32,0.002814222127199173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,1024,0.006040888941950268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,6144,32,0.006258666515350342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,65536,0.11751911375257705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,768,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,65536,0.12169510788387722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,16384,0.04220889012018839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,65536,0.07691288656658597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,512,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,16384,0.03367822368939718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,16384,0.02460266649723053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,12288,0.035777777433395386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,12288,0.028528001573350694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,256,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,12288,0.019926221834288705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,10240,0.03197333216667175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,10240,0.02475555572244856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,10240,0.018159111340840656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,6144,128,0.005568000177542369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,8192,0.02914311157332526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,8192,0.02068000038464864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,8192,0.015857777661747403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,7168,0.027230223019917805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,7168,0.01920000049802992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,7168,0.014528888795110913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,65536,0.07446666558583577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,6144,0.026169778572188482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,16384,0.022674666510687932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,6144,0.014911111858155994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,6144,0.01351999988158544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,5120,0.024442666106753882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,12288,0.0185199992524253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,5120,0.012893333203262754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,5120,0.012646222280131446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,10240,0.017989334132936265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,4096,0.022821333673265245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,4096,0.011739555332395764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,4096,0.010891555911964841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,8192,0.014254222313563028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,3584,0.01033688916100396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,3584,0.010088000032636855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,7168,0.013260444833172692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,3072,0.020005333754751418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,3072,0.009575111170609793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,3072,0.008968888885445064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,10240,0.03341777788268195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,2560,0.018567999203999836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,6144,0.013008889224794177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,2560,0.00830488900343577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,2560,0.007888000044557784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,2048,0.01683288812637329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,5120,0.011405333048767514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,2048,0.007274666594134436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,2048,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,4096,0.008563555777072906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,1536,0.017903111047214933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,1536,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,3584,0.008083555433485243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,1536,0.006607999818192587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,1024,0.01645955608950721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,1024,0.004571555389298333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,1024,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,3072,0.007598222129874759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,768,0.0173653331067827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,768,0.004132444245947732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,768,0.0059368887709246735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,2560,0.007300444775157505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,512,0.015511110424995422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,512,0.0035582222044467926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,512,0.005924444645643234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,2048,0.00692622239391009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,256,0.01648177703221639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,256,0.0031582222630580268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,256,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,1536,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,128,0.015107555521859063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,128,0.0031795555518733132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,128,0.005555555638339784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,1024,0.005950222412745158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,64,0.0029013332807355454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,64,0.005764444255166584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,5120,32,0.003220444545149803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,5120,32,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,65536,0.08916266759236653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,768,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,6144,3584,0.02271466619438595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,65536,0.07144711414972942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,512,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,16384,0.02866577770974901
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,16384,0.035748445325427584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,16384,0.02260444396071964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,12288,0.03086577852567037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,12288,0.023016000787417095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,256,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,12288,0.018922666708628338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,10240,0.018938667244381376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,10240,0.028201778729756672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,10240,0.017190222938855488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,5120,128,0.00546488869521353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,8192,0.016170667277442086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,8192,0.015453333655993143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,65536,0.0684337748421563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,7168,0.024324445260895625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,7168,0.01420177850458357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,16384,0.021936888496081035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,7168,0.013750221994188098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,12288,0.017788444956143696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,6144,0.023345778385798138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,6144,0.01275199982855055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,6144,0.0129466669427024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,5120,0.021819555097156104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,10240,0.015796444482273526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,5120,0.011232888533009423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,5120,0.01202400028705597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,4096,0.020498666498396132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,8192,0.013745778136783175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,4096,0.010320000350475311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,4096,0.010225777824719747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,3584,0.018505778577592637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,3584,0.00906844437122345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,3584,0.008968888885445064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,7168,0.012119111087587146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,3072,0.017864889568752713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,6144,0.010737777584128909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,3072,0.008267555799749162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,3072,0.007635555333561367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,2560,0.016923555069499545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,2560,0.007637333538797166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,5120,0.009755555954244401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,2560,0.007320889168315464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,2048,0.016163556112183463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,4096,0.008124444219801161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,2048,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,2048,0.006917333437336816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,1536,0.017214222086800467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,1536,0.004851555658711327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,3584,0.007662222617202335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,1536,0.006586666736337874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,1024,0.015817777978049386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,1024,0.004238222208287981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,1024,0.006262222097979651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,3072,0.007598222129874759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,768,0.016861332787407767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,768,0.0038337777886125776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,2560,0.007272889216740926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,768,0.006227555374304454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,512,0.014456889695591398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,512,0.0034933334423436057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,512,0.005615111026499007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,2048,0.006632888896597757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,256,0.015277332729763456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,256,0.003216000066863166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,1536,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,256,0.006067555397748947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,1024,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,128,0.014754666222466363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,128,0.0028844444702068963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,128,0.005559999909665849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,768,0.0057751110030545135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,64,0.0028293333533737394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,64,0.005617777920431561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,4096,32,0.0028782222006056043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,4096,32,0.0058871110280354815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,512,0.00554044461912579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,65536,0.07687911060121325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,256,0.005274666680230035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,65536,0.09501689010196263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,5120,3584,0.021876444419225056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,65536,0.06839733653598361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,16384,0.035084442959891424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,16384,0.02624799973434872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,16384,0.02187555531660716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,12288,0.03052266769938999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,12288,0.020768889122539096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,4096,128,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,12288,0.01867022282547421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,10240,0.027444443768925134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,10240,0.017827555537223816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,10240,0.01680266691578759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,8192,0.015359110302395292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,65536,0.06775733497407702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,8192,0.015034667319721647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,16384,0.021367111139827307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,7168,0.023920888702074688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,7168,0.013493333425786761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,7168,0.013682666752073499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,12288,0.01754044493039449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,6144,0.022971555590629578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,6144,0.012059555285506778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,6144,0.012667555775907306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,10240,0.015492444237073263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,5120,0.02050844497150845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,5120,0.010993777877754636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,5120,0.011346666349305047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,4096,0.01928444372283088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,4096,0.009975999593734741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,4096,0.009009777671760982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,8192,0.013384888569513956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,7168,0.011648000114493899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,3584,0.008625778059164682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,3584,0.008623999853928884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,6144,0.00980000032318963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,3072,0.016520889268981088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,3072,0.008077333370844523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,5120,0.008980444735950893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,3072,0.007878222399287755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,2560,0.016132444143295288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,2560,0.007249777515729268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,4096,0.008020444048775567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,2560,0.006948444578382704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,2048,0.015543111496501498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,2048,0.0057608890864584185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,2048,0.006582222051090664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,3584,0.00795111142926746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,1536,0.01647733317481147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,1536,0.004905777672926585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,1536,0.006247111078765657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,3072,0.007288888924651676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,1024,0.015815999772813585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,1024,0.0041706665522522396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,2560,0.0069306666652361555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,1024,0.006218666831652324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,2048,0.006912888752089606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,768,0.0035617777870761026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,768,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,65536,0.09717955854203965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,512,0.014110222458839417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,512,0.0035386664999855887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,512,0.0057706667317284485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,1536,0.006245333287451003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,256,0.014815110299322339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,256,0.003168888803985384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,256,0.005639111002286275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,1024,0.005882666756709416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,128,0.014765333798196582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,128,0.0028640000770489373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,128,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,768,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,64,0.002826666666401757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,4096,8192,0.025991110338105097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,64,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3584,32,0.0028435554769304064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3584,32,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,512,0.005626666463083691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,65536,0.06953866614235772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,65536,0.09383644660313924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,65536,0.06568355692757501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,256,0.0052355556852287715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,16384,0.022268444299697876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,16384,0.021273778544531927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,12288,0.029529776838090684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3584,128,0.005453333258628845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,12288,0.018401778406567044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,12288,0.01798577772246467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,65536,0.06662399901284112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,16384,0.02102222210831112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,10240,0.016131555040677387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,10240,0.01645155582163069
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,12288,0.017165333032608032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,8192,0.0251226673523585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,8192,0.01404355631934272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,8192,0.014500444134076437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,10240,0.015169777803950839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,7168,0.023196443915367126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,7168,0.01330488920211792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,7168,0.013417777915795645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,8192,0.012424888710180918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,6144,0.022356443934970435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,6144,0.01328000012371275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,6144,0.012318222059143914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,5120,0.020270221763186984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,5120,0.010407110883129967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,5120,0.010762666662534079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,7168,0.010750222537252637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,6144,0.009707555174827576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,4096,0.009595555563767752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,4096,0.008642666869693333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,5120,0.008953777452309927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,3584,0.016837333639462788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,3584,0.008594666918118795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,3584,0.00815555536084705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,4096,0.007977777885066139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,3072,0.01648355523745219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,3072,0.007628444168302748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,3072,0.007627555893527136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,8192,0.025408888856569927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,3584,0.007899555895063613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,2560,0.016158221496476066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,2560,0.006892444358931647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,2560,0.00721777809990777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,2048,0.015447111593352424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,2048,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,2048,0.00656088896923595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,3072,0.007263999846246507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,1536,0.01647911138004727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,1536,0.004864888886610667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,2560,0.0069715554515520734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,1536,0.006261333409282897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,1024,0.015142222245534262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,1024,0.0041706665522522396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,2048,0.006601777755551868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,1024,0.006235555642180973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,768,0.016929777132140267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,768,0.0038577777643998465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,768,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,1536,0.006232000059551663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,512,0.013718222578366598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,512,0.003519999898142285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,3584,0.017815111411942374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,512,0.005704888867007361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,1024,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,256,0.014783110883500842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,256,0.003152888889114062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,256,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,768,0.005576000031497743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,128,0.01443733274936676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,128,0.0028826666788922418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,128,0.005561777700980504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,64,0.0028133332315418455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,64,0.005575110928879843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,3072,32,0.002875555513633622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,3072,32,0.0058808889653947614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,512,0.005619555711746216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,65536,0.09178133143319024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,65536,0.059169775909847684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,256,0.005225777626037598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3584,768,0.01715022159947289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,65536,0.06353955798678927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,3072,128,0.0052284445199701525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,16384,0.0199306673473782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,16384,0.020863110820452373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,12288,0.029578665892283123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,12288,0.01664355562792884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,12288,0.01753777762254079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,10240,0.026478222674793665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,10240,0.014979556202888489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,10240,0.016132444143295288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,65536,0.06547911299599542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,16384,0.020799110333124798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,8192,0.013448889056841532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,8192,0.013929777675204806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,12288,0.016882666283183627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,7168,0.022659555077552795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,7168,0.012599111431174807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,7168,0.012999999854299756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,10240,0.014416888356208801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,6144,0.02186577849917942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,6144,0.011339555184046427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,6144,0.01218311074707243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,8192,0.011801777614487542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,5120,0.01920977731545766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,5120,0.010849778023031024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,5120,0.01037599974208408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,7168,0.010046222143703038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,6144,0.009350222018029954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,4096,0.01772711177666982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,4096,0.008952889177534316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,16384,0.0343137780825297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,10240,0.026649778087933857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,4096,0.008275555239783393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,5120,0.008692444198661381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,3584,0.01647644407219357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,3584,0.008271111382378472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,3584,0.007998222278224098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,3072,0.01611911091539595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,3072,0.007485333416197035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,3072,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,4096,0.007977777885066139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,2560,0.006340444501903322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,2560,0.006958222223652734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,3584,0.0075911109646161394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,2048,0.014925332532988654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,2048,0.005236444373925527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,3072,0.007373332977294922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,2048,0.006726222319735422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,1536,0.01566488875283135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,1536,0.004602666530344221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,1536,0.006959111326270633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,1024,0.01443377799457974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,1024,0.0038577777643998465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,1024,0.006113777971929974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,2560,0.0070062221752272705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,768,0.015561777684423657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,2048,0.006907555378145642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,768,0.003918222255176968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,768,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,1536,0.006244444598754247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,512,0.013756444056828817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,512,0.003499555504984326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,512,0.005771555420425203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,256,0.014759999182489185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,256,0.0032097777972618737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,1024,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,256,0.00555288874440723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,768,0.00562755556570159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,128,0.0029128889242808023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,128,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,512,0.005557333429654439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,64,0.0030053332448005676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,64,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2560,32,0.0028737777223189673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2560,32,0.00554044461912579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,256,0.005241777747869492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,3072,4096,0.017870222528775532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,65536,0.04894222153557671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,65536,0.08477333519193862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,65536,0.06083466609319051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,16384,0.018560889694425795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,16384,0.03246577911906772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,16384,0.020030222005314298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,12288,0.027802666028340656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2560,128,0.005321777943107817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,12288,0.015257777439223396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,12288,0.017161778277821012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,10240,0.025081778566042583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,65536,0.06425511174731784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,10240,0.013786666923099093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,10240,0.015442666080262927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,16384,0.02051111062367757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,8192,0.0229768885506524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,8192,0.013139555851618448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,8192,0.013415999710559845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,7168,0.020789333515697055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,12288,0.01606399979856279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,7168,0.011621333658695221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,7168,0.011887110769748688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,6144,0.019128888845443726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,6144,0.010832000109884473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,10240,0.012825777961148156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,6144,0.010655110908879174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,5120,0.017872888180944655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,8192,0.011019555230935415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,5120,0.009866666462686326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,5120,0.00960000024901496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,4096,0.016884444488419425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,7168,0.010021333065297868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,4096,0.008844444321261512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,4096,0.00795199970404307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,3584,0.01611199975013733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,3584,0.007631999750932057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,6144,0.009345778160625035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,3584,0.007632888853549957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,5120,0.008936888641781276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,3072,0.0069226668112807805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,3072,0.007618666523032718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,4096,0.00795199970404307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,2560,0.015105777316623263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,2560,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,3584,0.007638221813572778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,2560,0.007267555428875818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,2048,0.014785778191354541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,2048,0.00554666668176651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,2048,0.0069679998689227635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,3072,0.007296889192528195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,2560,0.00700444479783376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,2048,0.006635555376609166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,1536,0.004537777768241035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,1536,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,16384,0.033725334538353816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,1024,0.014815110299322339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,1024,0.005959110955397288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,1024,0.0038533334930737815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,1536,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,768,0.01647733317481147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,768,0.00360355567600992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,768,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,1024,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,512,0.0034755555291970572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,512,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,768,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,256,0.01481599940194024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,256,0.0031884445084465873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,256,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,8192,0.024255999260478552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,512,0.005497777627574072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,128,0.0028293333533737394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,128,0.01482222146458096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,128,0.005700444595681296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,64,0.002814222127199173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,64,0.005561777700980504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,256,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,2048,32,0.0028364445186323593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,2048,32,0.0052782222628593445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,65536,0.038712001509136625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,65536,0.059438221984439425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,2048,128,0.0052684446175893145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,16384,0.031670222679773964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,16384,0.01552088889810774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,16384,0.019926221834288705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,65536,0.0634746683968438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,12288,0.027489778068330552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,12288,0.013801777528391944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,12288,0.01680266691578759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,2560,0.015825778245925903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,16384,0.01938488913906945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,10240,0.024880889389250014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,10240,0.012412444584899478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,10240,0.015152000718646579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,8192,0.021827555365032617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,8192,0.011002666420406766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,12288,0.014306666122542487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,8192,0.012089778151777057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,7168,0.019891555110613506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,7168,0.010609777437316047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,10240,0.012088889049159156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,7168,0.01112711098459032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,6144,0.01889244384235806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,6144,0.009741333623727163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,6144,0.009977777798970541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,8192,0.011060444845093621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,7168,0.009697777529557547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,5120,0.009076444639099969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,5120,0.009343999955389235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,6144,0.00903555585278405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,4096,0.016587555408477783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,4096,0.008404444489214156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,4096,0.007986666427718269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,5120,0.008655110994974772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,3584,0.016193777322769165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,3584,0.007811555431948767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,3584,0.007782222496138678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,4096,0.007974222302436829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,3072,0.015784000356992085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2560,128,0.014434667097197639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,3072,0.006568888823191325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,3072,0.007248000138335758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,2560,0.014888889259762235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,3584,0.007675555845101674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,2560,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,2560,0.007311111523045435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,2048,0.01475911173555586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,2048,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,2048,0.0069155556460221606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,3072,0.007278222176763747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,1536,0.015446222490734525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,1536,0.004574222283230888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,2560,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,1536,0.006568888823191325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,2048,0.006567111031876669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,1024,0.003912000192536248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,1024,0.006278222219811545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,1536,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,768,0.016142222616407607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,768,0.00360355567600992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,768,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,1024,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,512,0.013442666994200813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,512,0.0035031110876136353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,512,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,768,0.005389333185222413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,256,0.014863111906581454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,256,0.0028862222615215513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,256,0.005561777700980504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,512,0.005395555661784277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,3072,0.015458666616015963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,256,0.0052693333062860705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,128,0.014140443669425117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,128,0.0028604444944196274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,128,0.005517333332035277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,64,0.0028115556471877625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,64,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1536,32,0.002520888836847411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1536,32,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,65536,0.08203111092249553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1536,128,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,65536,0.03159822358025445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,65536,0.05781066417694092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,16384,0.03091288937462701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,16384,0.012829333543777466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,16384,0.01925066610177358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,65536,0.06290133131874932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,12288,0.025972444150182936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,12288,0.010664000279373594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,12288,0.015091554986106025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,16384,0.016541333662139047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,10240,0.0222382214334276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,10240,0.01070666644308302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,10240,0.013092444174819522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,12288,0.013799111048380533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,8192,0.00924177798959944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,8192,0.01146133326821857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,10240,0.011935110721323224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,7168,0.01940088967482249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,1536,0.015471110741297403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,7168,0.008664888640244802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,7168,0.010399999717871347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,8192,0.01073066641887029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,6144,0.008106666306654612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,6144,0.009752000371615091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,7168,0.009715555442704095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,2048,512,0.013791110780504016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,5120,0.01757244434621599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,5120,0.007913777397738563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,5120,0.009017777939637503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,4096,0.016199111938476562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,6144,0.00903200027015474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,4096,0.007287110719415877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,4096,0.007939555578761632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,3584,0.015511999527613321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,5120,0.00851999968290329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,3584,0.0069208890199661255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,3584,0.007702222300900354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,3072,0.015487111277050443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,4096,0.008000888758235509
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,3072,0.006272000157170826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,3072,0.0073724447025193115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,2560,0.014966222974989148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,3584,0.007335999773608313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,2560,0.006086222413513396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,2560,0.007012444237867991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,2048,0.014099554883109199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,3072,0.007259555160999298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,2048,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,2048,0.00655733338660664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,1536,0.014427555931939019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,2560,0.006980444822046492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,1536,0.004537777768241035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,1536,0.006576889091067844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,1024,0.013798221945762634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,2048,0.0063591111037466265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,1024,0.0038622220357259116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,1024,0.006001777946949005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,768,0.01533511115445031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,1536,0.006230222268237009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,768,0.003524444583389494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,768,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,512,0.01274400038851632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,1024,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,65536,0.08320533567004733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,512,0.005600000007285013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,512,0.0035075553589397003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,768,0.005433777968088786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,256,0.002879111096262932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,256,0.005740444279379315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,128,0.014087999860445658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,128,0.0028115556471877625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,512,0.0052737775776121355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,128,0.005554666535721884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,64,0.002507555608948072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,256,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,64,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,1024,32,0.0028400001012616684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,1024,32,0.005616000129116907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,65536,0.02549955579969618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,1024,128,0.0052800000541739995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,65536,0.08171377579371135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,65536,0.05670311053593954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,16384,0.011326221956147088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,16384,0.029836446046829224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,16384,0.017453332742055256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,65536,0.06275910801357694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,12288,0.009854222337404886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,12288,0.014840000205569796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,16384,0.016491555505328707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,12288,0.013759999639458127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,10240,0.02195022172398037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,10240,0.01034044474363327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,10240,0.012440000143316058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,10240,0.01202133297920227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,8192,0.020226667324701946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,5120,0.017508443858888414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,8192,0.008961777720186446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,8192,0.010835555692513784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,7168,0.018929777873886954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,7168,0.00833777752187517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,8192,0.010784000158309937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,7168,0.010388444695207808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,6144,0.018183110488785636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,6144,0.00797333319981893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,7168,0.009703999592198266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,6144,0.009693333672152625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,5120,0.01720266706413693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,5120,0.007979555262459649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,6144,0.008982222113344405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,5120,0.008993777963850234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,4096,0.016135111451148987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,4096,0.007652444144090016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,4096,0.00830844458606508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,5120,0.008578666382365758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,3584,0.015822221835454304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,3584,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,3584,0.007610666255156199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,4096,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,3072,0.01517777807182736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1536,1024,0.014347554908858405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,3072,0.006949333681000604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,3072,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,3584,0.0075635554061995606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,2560,0.006628444625271692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,2560,0.007279111279381647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,2048,0.014431110686726041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,3072,0.007312888900438945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,2048,0.0052062223354975385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,2048,0.006794666664467917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,2560,0.0069217777086628815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,1536,0.014101333088344999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,1536,0.00453599997692638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,1536,0.0064239998658498125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,2048,0.006590222318967183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,1024,0.013442666994200813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,1024,0.0038364442686239877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,1024,0.005929777605666055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,1536,0.006255110932721033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,768,0.015895111693276297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,768,0.003568888952334722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,768,0.0058746664888328975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,1024,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,512,0.013120888835854001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,512,0.0034906665484110513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,512,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,768,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,256,0.0028880000528362063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,256,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,512,0.005378666851255629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,128,0.014099554883109199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,128,0.0028195555011431375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,128,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,256,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,64,0.0028133332315418455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,64,0.005566222386227713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,768,32,0.0024826667375034755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,768,128,0.0052151112920708126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,768,32,0.005542222410440445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,65536,0.01956622302532196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,65536,0.08102222283681233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,65536,0.05657066901524862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,16384,0.02775555517938402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,65536,0.06264533599217732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,16384,0.010859555668301053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,16384,0.016936888297398884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,16384,0.016512889001104567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,12288,0.009310222334331935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,12288,0.013737777868906656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,12288,0.013731555806265937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,10240,0.02185955478085412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,10240,0.009347555538018545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,10240,0.012496000362767113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,10240,0.01205244412024816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,8192,0.020444444484180875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,8192,0.019557333654827543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,8192,0.00814133303032981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,8192,0.011005333728260465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,7168,0.018732445107565988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,7168,0.00794400026400884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,8192,0.010713777608341642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,7168,0.010314666562610203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,6144,0.017887110511461895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,6144,0.007299555672539606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,7168,0.009699555734793345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,6144,0.009865778187910715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,5120,0.016809778081046212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,5120,0.00794933322403166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,6144,0.009003555609120263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,5120,0.008940444224410588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,5120,0.008593777815500895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,4096,0.015828443898095023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,4096,0.007342221836249034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,4096,0.007978666987684038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,6144,0.01824711097611321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,3584,0.015068444940778943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,4096,0.007991111112965478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,3584,0.0069875551594628235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,3584,0.00793333351612091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,3072,0.014811555544535318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,3072,0.006899555524190267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,3584,0.007612444460391998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,3072,0.007302222152551015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,2560,0.014516444669829475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,2560,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,3072,0.006953777538405524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,2560,0.007246221933099959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,2048,0.013757333159446716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,2048,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,2560,0.006974221931563483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,2048,0.006618666566080517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,1536,0.013403555585278405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,1536,0.00452977791428566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,2048,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,1536,0.006584888945023219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,1536,0.005973333285914526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,1024,0.0038684445122877755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,1024,0.005966222120655908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,1024,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,768,0.003559111307064692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,768,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,768,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,1024,256,0.014896889527638754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,512,0.003248888792263137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,512,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,512,0.005569777968857024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,12288,0.024693333440356787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,256,0.0028968888024489083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,256,0.005573333137565189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,256,0.005218666460778978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,2560,0.014774221513006421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,128,0.0026488889836602737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,128,0.00554577757914861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,512,128,0.005049777941571342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,768,256,0.014840889308187695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,64,0.0025253333151340485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,64,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,512,32,0.00282577777074443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,512,32,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,65536,0.015480000111791821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,65536,0.08049155606163873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,65536,0.05575022432539198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,16384,0.027756444282001917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,65536,0.05932355589336819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,16384,0.00962400022480223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,16384,0.016154666741689045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,12288,0.023589332898457844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,16384,0.01650399963061015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,12288,0.008336889247099558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,12288,0.013754666679435305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,10240,0.021249777740902368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,12288,0.013407111167907715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,10240,0.007610666255156199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,10240,0.012279111478063794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,8192,0.01919822229279412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,10240,0.012063999970753988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,8192,0.0069422221018208405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,8192,0.011045333411958484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,8192,0.01056177748574151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,7168,0.0069679998689227635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,7168,0.01019200020366245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,7168,0.009688888986905416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,12288,0.024167999625205994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,6144,0.017260443833139207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,6144,0.00664533343580034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,6144,0.009365333451165093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,5120,0.01614755557643043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,6144,0.00905866672595342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,5120,0.006270222365856171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,5120,0.009023110899660323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,4096,0.015470221638679504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,5120,0.00849866701496972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,4096,0.006682666639486949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,4096,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,4096,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,3584,0.006612444503439798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,3584,0.007606222397751278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,3584,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,3072,0.014466666513019137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,3072,0.00630044440428416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,3072,0.007370666497283512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,3072,0.007258666886223688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,2560,0.013772444592581855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,2560,0.006576889091067844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,2560,0.007293333609898885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,2560,0.0069155556460221606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,2048,0.013422222601042854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,2048,0.006252444452709622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,2048,0.006640000061856375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,1024,0.012741333080662621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,2048,0.00657066661450598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,1536,0.012729778057999082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,1536,0.0042453333735466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,768,0.014112000664075216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,1536,0.006576889091067844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,1024,0.012422222230169507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,1536,0.005962666538026597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,1024,0.0038568890757030914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,1024,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,768,0.00350222239891688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,768,0.014800000521871777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,1024,0.005879111174080107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,768,0.005919111271699269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,768,0.00562755556570159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,512,0.01163733336660597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,512,0.0031591111587153543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,512,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,256,0.011776888536082374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,256,0.00282577777074443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,512,0.005562666803598404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,256,0.005425777700212266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,128,0.012149333126015134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,256,0.005211555709441503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,128,0.0028168888141711554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,128,0.00556355549229516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,64,0.002542222125662698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,64,0.005543111099137201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,256,128,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,256,32,0.0025111111915773815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,256,32,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,65536,0.01479022204875946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,512,0.011767110890812345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,65536,0.055461333857642285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,16384,0.00793866647614373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,16384,0.026759111218982275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,65536,0.05332000056902567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,16384,0.016185777054892648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,16384,0.016200888488027785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,12288,0.023002665903833177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,12288,0.007647111184067196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,12288,0.013136000268989138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,10240,0.007343110938866933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,10240,0.020629333125220407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,12288,0.013439999686347114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,128,0.012715555727481842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,10240,0.012087999946541257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,512,256,0.014781332678265043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,10240,0.011959999799728394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,8192,0.01841600073708428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,8192,0.006911110960774952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,8192,0.01055466632048289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,8192,0.01073333372672399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,7168,0.01795644395881229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,7168,0.006641777853171031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,7168,0.010272000398900773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,7168,0.009709333380063375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,6144,0.01685244507259793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,6144,0.006404444575309753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,6144,0.009355555805895064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,6144,0.008960000342792934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,5120,0.015498666299713982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,5120,0.00629688882165485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,4096,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,5120,0.008974221845467886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,5120,0.008344000412358178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,4096,0.014488000008794995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,3584,0.006229333165619109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,4096,0.008008889026112026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,3584,0.0076044441925154785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,4096,0.007843555675612556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,3072,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,3584,0.007296000089910295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,3072,0.014135110709402295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,3072,0.0069919998447100324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,3072,0.007240888973077138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,2560,0.006325333482689328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,2560,0.006976000136799282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,2560,0.006588444527652528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,2048,0.01312800000111262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,2048,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,2048,0.00645955569214291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,1536,0.004229333251714706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,1536,0.011733333269755045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,2048,0.0064488889442549805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,1536,0.006235555642180973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,1536,0.005956444475385878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,1024,0.012429333395428128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,1024,0.003846222327815162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,1024,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,1024,0.005899555567238066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,768,0.012721777790122561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,768,0.003543111185232798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,768,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,768,0.00544799988468488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,512,0.01039644413524204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,512,0.003334222154484855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,512,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,256,0.002817777709828483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,512,0.005280888742870755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,256,0.011394666300879585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,256,0.005532444351249271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,256,0.005247999810510212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,128,0.011003555523024665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,128,0.0028604444944196274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,128,0.005369777894682354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,64,0.00254133323000537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,64,128,128,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,64,0.0052426668504873914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,128,32,0.002558222247494592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,128,32,0.0052266667286554975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,65536,0.011762667033407422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,65536,0.05336622065967984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,16384,0.007337777978844113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,16384,0.01601777805222405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,12288,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,12288,0.013442666994200813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,10240,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,10240,0.012375999655988483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,8192,0.006616888774765863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,8192,0.01072266697883606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,7168,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,7168,0.010183999935785929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,6144,0.006607999818192587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,6144,0.009535999761687385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,5120,0.006274666637182236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,5120,0.008631111019187504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,4096,0.00619911112719112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,4096,0.00793688909875022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,3584,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,3584,0.007648888561460707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,3072,0.005968889014588461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,3072,0.007258666886223688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,2560,0.0063164445261160536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,2560,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,2048,0.006294222341643439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,2048,0.006642666541867786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,1536,0.0042275554604000514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,1536,0.006285333385070165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,1024,0.0036106668412685394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,1024,0.006039999839332368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,768,0.003563555578390757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,768,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,512,0.003208000005947219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,512,0.005554666535721884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,256,0.0028568889117903183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,256,0.005246222019195557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,128,0.0028773333049482773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,7168,0.01854311095343696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,64,0.002536888958679305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,128,0.005391110976537068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,64,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,64,32,0.002532444480392668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,64,32,0.005294222384691238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,65536,0.011367110742463006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,65536,0.05166044500139025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,16384,0.007615110940403408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,16384,0.015847111741701763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,12288,0.006288888967699475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,12288,0.013796444568369122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,10240,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,10240,0.012101333174440594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,8192,0.006597333484225803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,8192,0.011027555498811932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,7168,0.006566222343179915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,7168,0.009678222239017487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,6144,0.00591555568906996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,6144,0.00963644435008367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,5120,0.006294222341643439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,4096,0.00830844458606508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,5120,0.009007111191749573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,4096,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,3584,0.006299555715587404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,3584,0.007631111476156447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,3072,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,2560,0.007074666519959767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,3072,0.006978666616810693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,2560,0.006284444282452266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,2048,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,2048,0.006616888774765863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,1536,0.004516444272465176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,1536,0.0063688887490166565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,768,0.0035137778355015647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,1024,0.003565333369705412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,768,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,1024,0.006255110932721033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,512,0.0031715554909573663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,512,0.005625777774386936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,256,0.0028808888875775864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,256,0.005884444548024072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,128,0.002532444480392668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,256,3584,0.014447111222479077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,128,0.0052675555149714155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,64,0.002556444456179937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,64,0.005555555638339784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,64,32,32,0.002471111093958219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,64,32,32,0.0052595556610160405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,16384,0.36531911955939395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,16384,0.23458756340874565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,16384,0.16963556077745226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,12288,0.2776133219401042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,12288,0.1266711155573527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,10240,0.1623928944269816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,10240,0.22745066218905977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,10240,0.10719911257425944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,65536,0.07994311385684542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,8192,0.18265333440568712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,8192,0.14521866374545628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,8192,0.08980711301167806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,7168,0.12905600335862902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,3584,0.014415999253590902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,7168,0.1615635553995768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,64,128,2560,0.01274400038851632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,7168,0.07675377527872722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,6144,0.11876800325181748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,6144,0.13738044102986655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,6144,0.06734489070044623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,5120,0.10624177588356866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,5120,0.11807644367218018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,5120,0.05725066529379951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,12288,0.18778577115800646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,16384,0.14936889542473689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,4096,0.09473688734902276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,4096,0.047022223472595215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,12288,0.10450844632254706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,3584,0.08995288610458374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,3584,0.08460177977879842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,3584,0.04183377822240194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,10240,0.08678488598929511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,3072,0.08376089069578384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,3072,0.07464800278345744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,3072,0.03704444567362467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,8192,0.0704924464225769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,2560,0.08135644594828288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,2560,0.06235111422008938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,2560,0.03286666671435038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,7168,0.06218044625388252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,2048,0.0755795571539137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,2048,0.050047109524408974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,2048,0.028365333875020344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,6144,0.05326666765742832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,1536,0.06873777839872572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,1536,0.039699554443359375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,1536,0.022973333795865376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,5120,0.04579822222391764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,1024,0.062092443307240806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,1024,0.029493331909179688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,1024,0.01834133267402649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,4096,0.0959813329908583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,768,0.030184000730514526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,768,0.022367111510700647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,768,0.01644355555375417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,4096,0.044793777995639376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,512,0.05650133556789822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,512,0.01549600064754486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,512,0.014863999353514777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,3584,0.040744000011020236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,256,0.025038222471872967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,256,0.009332444104883406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,256,0.01312355531586541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,3072,0.030790223015679255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,65536,128,0.02129688858985901
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,128,0.007662222617202335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,128,0.010416888528399997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,2560,0.026919111609458923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,64,0.006681777950790193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,64,0.01106844428512785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,65536,32,0.006907555378145642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,2048,0.022797332869635686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,65536,32,0.013454222016864352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,65536,0.34442933400472003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,1536,0.01908888916174571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,65536,0.2292382187313504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,1024,0.015480889214409722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,65536,0.18466577264997694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,16384,0.0832844442791409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,16384,0.05115911033418444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,12288,0.059912886884477407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,12288,0.06353422005971272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,12288,0.03930133250024583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,768,0.013727111121018728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,10240,0.053103999959097974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,10240,0.05388177765740288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,10240,0.03453155689769321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,65536,0.15838932991027832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,8192,0.04839644498295254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,8192,0.044758223825030856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,8192,0.029341333442264136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,16384,0.042615112331178456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,7168,0.04368622104326884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,7168,0.039044443104002215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,7168,0.0259253333012263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,12288,0.0346942212846544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,6144,0.04079733292261759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,6144,0.03356800145573086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,6144,0.023603555228975084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,10240,0.03203466534614563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,5120,0.03694222370783488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,5120,0.029436445898479883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,5120,0.020263999700546265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,8192,0.025008888708220586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,4096,0.03491288754675124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,4096,0.023919110496838886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,4096,0.017142222987280954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,7168,0.02149066660139296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,3584,0.032849778731664024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,3584,0.02141155633661482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,3584,0.015951999359660678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,16384,0.0738426645596822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,3072,0.031665778822369046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,6144,0.022996443841192458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,3072,0.01885511146651374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,3072,0.01458044515715705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,2560,0.03320088982582092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,2560,0.016145777371194627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,2560,0.013437333206335703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,512,0.012313777373896705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,2048,0.031074666314654883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,2048,0.01349866638580958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,2048,0.01236177815331353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,256,0.011095999843544431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,1536,0.028590222199757893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,1536,0.010885333021481832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,1536,0.011114666859308878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,65536,128,0.010437332921557956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,1024,0.02643288837538825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,1024,0.008642666869693333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,1024,0.008858666651778752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,5120,0.01756977703836229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,4096,0.0158906661801868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,768,0.007111111448870764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,768,0.007585778004593319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,3584,0.016151999433835346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,512,0.02200711104604933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,512,0.005311111195219888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,512,0.0069066666894488865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,3072,0.013644444445768992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,256,0.017514665921529133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,256,0.00350577798154619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,256,0.006659555352396435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,2560,0.01239733315176434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,128,0.015451555450757345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,128,0.0033119999700122406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,128,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,2048,0.011012444065676795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,64,0.0032106666929192017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,64,0.006640000061856375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,16384,32,0.0032106666929192017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,16384,32,0.0069306666652361555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,1536,0.009929777847396003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,65536,0.26136000951131183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,1024,0.007830222447713217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,65536,0.1705848905775282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,65536,0.15870132711198595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,16384,0.07487822241253324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,16384,0.044349332650502525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,768,0.007272000114123027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,12288,0.04875466558668348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,12288,0.05323555403285556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,12288,0.03469511204295688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,512,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,256,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,10240,0.04322488771544563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,10240,0.03087555699878269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,16384,128,0.006518222391605377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,8192,0.040023111634784274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,8192,0.03775022096104092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,8192,0.02609688871436649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,65536,0.141649776034885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,16384,0.03938755393028259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,7168,0.03209866748915778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,7168,0.023451555106374953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,12288,0.03238488899336921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,6144,0.033942222595214844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,6144,0.02801777919133504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,6144,0.021306667062971327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,10240,0.027452444036801655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,5120,0.03138844503296746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,5120,0.02409422232045068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,5120,0.018537777993414137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,8192,0.025331555141343012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,4096,0.029665778080622356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,16384,768,0.01920355525281694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,4096,0.02071377800570594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,4096,0.015991111596425373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,3584,0.028129776318868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,3584,0.018056000272432964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,3584,0.014765333798196582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,7168,0.020424889193640817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,3072,0.027594667341974046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,3072,0.01588355501492818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,3072,0.013759999639458127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,6144,0.018175111876593698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,2560,0.028163555595609877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,2560,0.013781333135233985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,2560,0.012829333543777466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,5120,0.016759110821617972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,2048,0.02678577767478095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,2048,0.011920000116030375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,2048,0.011414222419261932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,4096,0.014695111248228284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,3584,0.014565333724021912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,1536,0.025217778152889673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,16384,0.05890755520926582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,1536,0.010024000373151567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,1536,0.010328888893127441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,1024,0.021773333350817364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,1024,0.008378666308191087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,1024,0.007617777420414819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,3072,0.012686221963829465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,768,0.018277333842383493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,768,0.007098666495747036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,768,0.007288888924651676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,10240,0.04351822204060025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,512,0.019592000378502738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,512,0.004889777965015835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,512,0.007003555695215861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,256,0.01684888866212633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,256,0.0041848888827694785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,256,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,2560,0.011695110963450538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,128,0.015472888946533203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,128,0.0035013332962989807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,128,0.006292444550328785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,2048,0.009759999811649323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,64,0.0032026666320032547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,64,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,12288,32,0.003192000091075897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,12288,32,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,1536,0.008422222402360704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,1024,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,65536,0.21888532903459337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,65536,0.15189599990844727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,16384,0.05413066678576999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,12288,7168,0.036293334431118436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,16384,0.06203999784257677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,65536,0.15152000056372747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,16384,0.042618665430280894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,12288,0.042746666404936046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,12288,0.04445866743723551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,12288,0.033287998702791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,768,0.0069724445541699724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,10240,0.03973599937227037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,10240,0.03538399934768677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,512,0.006592888798978593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,10240,0.029373334513770208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,8192,0.03047555685043335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,8192,0.03666755557060242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,8192,0.025088888075616624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,256,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,7168,0.033238222201665245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,7168,0.0258906665775511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,7168,0.022312889496485393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,12288,128,0.006350222147173352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,6144,0.023016000787417095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,6144,0.02053600052992503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,65536,0.13392889499664307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,5120,0.029197331931855943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,5120,0.019719999697473314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,16384,0.038211554288864136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,5120,0.018236445056067575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,12288,0.029275556405385334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,4096,0.027495111028353374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,4096,0.017484444710943434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,4096,0.015451555450757345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,3584,0.02622844444380866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,10240,0.02515911062558492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,3584,0.015038222074508667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,3584,0.0144177774588267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,3072,0.02572000026702881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,8192,0.022034666604465906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,3072,0.013295110729005603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,3072,0.013413333230548434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,2560,0.027106665902667578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,7168,0.019747555255889893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,2560,0.01311555587583118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,2560,0.012368000215954252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,2048,0.025285333395004272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,2048,0.011758222348160215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,6144,0.01771377854877048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,2048,0.01107377807299296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,1536,0.023744000328911677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,1536,0.009644444617960189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,1536,0.009302222066455418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,5120,0.015824000040690105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,1024,0.020584889584117465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,4096,0.01384888920519087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,1024,0.008020444048775567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,1024,0.007611555357774098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,768,0.01791555517249637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,768,0.006257777826653586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,3584,0.014128888646761576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,768,0.006931555353932911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,512,0.018529777725537617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,512,0.0052151112920708126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,3072,0.012290666500727335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,512,0.006591111007663939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,256,0.016515556308958266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,256,0.004192000048028098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,256,0.006456888798210356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,2560,0.010560000108347999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,2048,0.008859555754396651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,128,0.015441776977645027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,128,0.003559999995761447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,128,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,1536,0.00793688909875022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,64,0.00319377767542998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,64,0.0064204442832205035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,10240,32,0.00319377767542998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,10240,32,0.006803555621041193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,65536,0.13711288240220812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,65536,0.18012355433570015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,1024,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,65536,0.10248711374070908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,768,0.006286222073766921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,16384,0.0480657782819536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,16384,0.05008711086379158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,16384,0.029988444513744775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,512,0.006572444405820634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,12288,0.04008622301949395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,12288,0.03656889001528422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,12288,0.02437422176202138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,10240,0.030692444907294378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,10240,0.0355679988861084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,256,0.006239999913507038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,10240,0.02181511123975118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,8192,0.0265964451763365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,8192,0.03360622127850851
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,8192,0.018583999739752877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,7168,0.03051466743151347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,10240,128,0.006225777582989798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,7168,0.023183999790085688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,7168,0.016912889149453905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,6144,0.02905244297451443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,6144,0.020270221763186984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,6144,0.01549600064754486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,65536,0.08502488666110569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,16384,0.029382222228580054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,5120,0.017498667041460674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,5120,0.01419911119672987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,12288,0.021713778376579285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,4096,0.025450666745503742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,10240,0.019768888751665752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,4096,0.0148026661740409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,4096,0.012171555724408893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,3584,0.02419111132621765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,8192,0.01591111057334476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,3584,0.013444444371594323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,10240,6144,0.03155466583040025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,3584,0.011360888679822287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,3072,0.011844444606039258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,3072,0.023370666636361018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,3072,0.010822222464614444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,2560,0.02571644385655721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,7168,0.014640889234013028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,2560,0.010392888552612728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,2560,0.0101724440852801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,2048,0.02392622166209751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,2048,0.009027555584907532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,2048,0.009359999663299983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,6144,0.013637333280510373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,1536,0.021871111459202234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,1536,0.007941332956155142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,1536,0.00795911086930169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,5120,0.012650666965378655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,1024,0.018571555614471436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,1024,0.005903111149867375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,1024,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,4096,0.01132266637351778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,768,0.01752000053723653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,768,0.0046986668474144405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,3584,0.010593777729405297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,768,0.006265777680608962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,512,0.01717155509524875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,3072,0.009014222357008193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,512,0.0039022221333450745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,512,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,256,0.016812443733215332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,2560,0.007962666451931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,256,0.003563555578390757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,256,0.005888888819350137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,128,0.015121777852376303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,128,0.0032177778581778207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,2048,0.007326222128338284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,128,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,64,0.0028968888024489083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,64,0.006226666685607698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,8192,32,0.0028497777465316984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,8192,32,0.006203555398517185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,1536,0.00686488880051507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,65536,0.15730399555630153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,65536,0.1313555505540636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,65536,0.09322133329179551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,16384,0.04638488756285774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,1024,0.006255110932721033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,16384,0.04673422376314799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,16384,0.028054220808876887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,768,0.005956444475385878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,12288,0.038506666819254555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,12288,0.036990222003724836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,12288,0.02272355556488037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,512,0.0058808889653947614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,256,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,10240,0.034613334470325045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,10240,0.028684443897671167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,10240,0.02013422217633989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,8192,128,0.005620444400442972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,8192,0.03211911188231574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,8192,0.024312888582547505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,8192,0.017887999614079792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,7168,0.029320889049106177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,7168,0.02151022189193302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,7168,0.01605244477589925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,65536,0.0795431137084961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,6144,0.02808088726467556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,6144,0.018908444378111098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,6144,0.014771555860837301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,16384,0.025389333566029865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,5120,0.025758221745491028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,5120,0.016407110624843173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,5120,0.013611555927329592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,4096,0.0248346659872267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,4096,0.01421333352724711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,12288,0.02269511090384589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,4096,0.011398221883508893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,3584,0.023352889551056757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,3584,0.012323555847009024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,10240,0.017493334081437852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,3584,0.011178666518794166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,3072,0.022686221533351477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,8192,0.015407111909654407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,3072,0.011358222199810876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,3072,0.010423999693658618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,2560,0.024880000286632117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,2560,0.010016000105275048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,2560,0.009697777529557547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,7168,0.01540177729394701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,2048,0.02236977716286977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,2048,0.008715555899673039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,2048,0.00831288927131229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,6144,0.012799111505349478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,1536,0.019893333315849304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,1536,0.0075795559419526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,1536,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,5120,0.012059555285506778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,1024,0.017906667457686532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,1024,0.005589333259397083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,1024,0.006285333385070165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,4096,0.010750222537252637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,768,0.01789955629242791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,768,0.0042275554604000514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,768,0.005960000058015187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,3584,0.009305777649084726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,512,0.016492444607946608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,3072,0.007997333175606197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,512,0.004015999949640698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,8192,5120,0.026712000370025635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,512,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,256,0.017179555363125272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,256,0.003499555504984326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,256,0.005535111245181825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,2560,0.007376888559924231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,7168,128,0.015271110667122735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,128,0.003228444399105178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,128,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,64,0.0028488888508743713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,64,0.0058328890138202244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,7168,32,0.0031999999450312722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,2048,0.007242666350470648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,7168,32,0.005903999838564131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,65536,0.13582932949066162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,1536,0.006270222365856171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,65536,0.12372267246246338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,16384,0.04389866524272495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,65536,0.08437422249052261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,16384,0.038972444004482694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,16384,0.02642755541536543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,1024,0.006064888917737537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,12288,0.03661600086424086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,12288,0.03161244591077169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,12288,0.021056888831986323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,768,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,10240,0.02755911151568095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,10240,0.018918222851223417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,512,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,8192,0.030301332473754883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,256,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,8192,0.022777777579095628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,8192,0.016972444123691983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,7168,128,0.005618666609128316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,7168,0.02808533443344964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,7168,0.020367999871571858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,7168,0.015220445063379077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,6144,0.02657866643534766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,6144,0.015101333459218344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,6144,0.014104889498816596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,65536,0.07352533605363634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,5120,0.02472977836926778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,5120,0.013775111072593264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,16384,0.023574221465322707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,5120,0.013026666310098437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,12288,0.018944000204404194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,4096,0.023659555448426142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,4096,0.012235555383894177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,4096,0.011053333679835001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,10240,0.018914666440751817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,3584,0.022456000248591106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,3584,0.010729778144094678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,8192,0.014807111687130399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,3584,0.01089333328935835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,3072,0.021664000219768945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,7168,0.013167999684810638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,3072,0.009635555247465769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,3072,0.010026666853162978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,2560,0.02369511127471924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,2560,0.008637333081828224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,2560,0.009302222066455418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,6144,0.012327999704413943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,2048,0.021320889393488567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,2048,0.00794400026400884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,2048,0.00794666674402025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,5120,0.011019555230935415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,4096,0.009128889275921715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,1536,0.006671111202902264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,1536,0.006649777707126405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,1024,0.01722400055991279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,1024,0.004558222161398994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,3584,0.008435555630260045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,1024,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,3072,0.007655999726719326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,768,0.004071111066473855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,768,0.006225777582989798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,2560,0.007266666326257918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,512,0.016186666157510545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,512,0.003567111161020067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,512,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,2048,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,256,0.01657777859105004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,256,0.0033235556135574975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,256,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,1536,0.00656355544924736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,128,0.01447111037042406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,128,0.003167111012670729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,128,0.005568000177542369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,1024,0.005953777581453323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,64,0.0031555555760860443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,64,0.005723555468850666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,6144,32,0.0032017777363459268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,6144,32,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,768,0.00591111100382275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,65536,0.11651644441816543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,65536,0.11949511369069417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,512,0.005620444400442972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,16384,0.04189244574970669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,65536,0.07713599999745686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,16384,0.03546222382121616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,256,0.005566222386227713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,16384,0.02458222210407257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,12288,0.03516355488035414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,12288,0.027257778578334387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,12288,0.019888000355826486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,10240,0.03154311246342129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,10240,0.023399111297395494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,10240,0.017883555756674874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,6144,128,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,65536,0.0704888900121053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,8192,0.01996799972322252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,8192,0.015607999430762397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,16384,0.02257333364751604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,7168,0.027225777506828308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,7168,0.018467555443445843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,12288,0.018345778187115986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,7168,0.014488000008794995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,6144,0.025806221697065566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,6144,0.016491555505328707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,6144,0.013530666629473368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,10240,0.016034666034910414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,5120,0.023875556058353845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,5120,0.014013333453072442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,5120,0.012447111308574677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,8192,0.014408000641398959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,4096,0.02274488906065623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,4096,0.012445333103338877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,4096,0.0107022225856781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,7168,0.012728888955381183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,3584,0.021412443783548143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,3584,0.011203555597199334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,3584,0.010202666951550378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,10240,0.032864888509114586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,6144,0.011761777930789523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,3072,0.020945777495702107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,3072,0.010211555494202508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,3072,0.009200000100665623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,2560,0.021249777740902368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,2560,0.008904889225959778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,5120,0.010217777556843227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,2560,0.008115555677149031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,2048,0.01872711049185859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,2048,0.007905777957704332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,2048,0.007130666739410824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,4096,0.008361777497662438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,3584,0.007992000215583378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,3072,0.00758755538198683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,1536,0.005919999960396025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,1536,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,2560,0.007263110743628607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,1024,0.016533333394262526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,1024,0.004243555582231945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,1024,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,2048,0.0069715554515520734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,768,0.017236444685194228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,768,0.0042204442951414315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,768,0.005922666854328579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,512,0.015503111812803479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,512,0.0035911111368073356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,512,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,1536,0.006309333360857434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,256,0.016551110479566786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,256,0.0032248888164758682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,256,0.005617777920431561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,1024,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,128,0.0151235560576121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,1536,0.018823999497625563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,128,0.0031786666562159858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,128,0.005564444594913059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,64,0.00282222218811512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,64,0.005766222046481238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,5120,32,0.003170666595300039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,5120,32,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,768,0.005879111174080107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,65536,0.09203733338250054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,512,0.0058808889653947614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,65536,0.07077333662245008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,6144,768,0.01719911065366533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,16384,0.035601778162850276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,256,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,16384,0.028189334604475234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,16384,0.02239199976126353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,12288,0.03031022349993388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,12288,0.02307022280163235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,5120,128,0.005557333429654439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,12288,0.01866755551762051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,10240,0.02774222195148468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,10240,0.020044444335831534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,10240,0.016720000240537856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,8192,0.02609155575434367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,8192,0.01715377800994449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,8192,0.01537333263291253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,65536,0.06844800048404269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,7168,0.02431022293037838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,7168,0.015617777903874716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,7168,0.013802666631009845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,16384,0.02167466613981459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,6144,0.023175110419591267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,6144,0.013779555757840475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,12288,0.017630222770902846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,6144,0.012819555898507437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,5120,0.02183644473552704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,10240,0.015483554866578845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,5120,0.01144444445768992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,5120,0.01184266640080346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,8192,0.013694222602579327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,4096,0.010300444232092964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,4096,0.010248888697889116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,7168,0.01237066669596566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,3584,0.019483556350072224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,8192,0.02904799911710951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,3584,0.009034666750166152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,3584,0.008766222331258986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,6144,0.010619555910428366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,3072,0.01685688893000285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,3072,0.00813688917292489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,3072,0.007967111137178209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,5120,0.008683555656009251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,2560,0.0196106665664249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,2560,0.007655111451943715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,2560,0.007286222444640265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,2048,0.018191110756662156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,2048,0.006280888699822956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,2048,0.006603555546866522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,4096,0.007999999655617608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,1536,0.017840888765123155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,1536,0.004880889008442561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,1536,0.006566222343179915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,3584,0.007654222349325816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,1024,0.015863999724388123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,3072,0.007291555404663086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,1024,0.004208000169859992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,1024,0.006215111249023014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,2560,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,768,0.017498667041460674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,768,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,768,0.003930666794379552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,512,0.015153777268197803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,512,0.003548444559176763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,2048,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,512,0.005738666488064661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,256,0.015224888920783997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,256,0.0032319999817344877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,1536,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,256,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,1024,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,128,0.014474666780895658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,768,0.005589333259397083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,128,0.0032053333189752367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,128,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,64,0.0028400001012616684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,64,0.005575110928879843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,512,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,4096,32,0.0031591111587153543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,4096,32,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,65536,0.07918044593599108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,65536,0.0935164425108168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,256,0.0052639999323421055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,65536,0.06809510787328084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,16384,0.03448799914783902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,16384,0.025285333395004272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,16384,0.021910222040282354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,12288,0.029859556092156306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,12288,0.02018044392267863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,4096,128,0.0052782222628593445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,12288,0.018227555685573153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,10240,0.027107555005285475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,10240,0.017674666312005784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,5120,1536,0.018230222993426852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,10240,0.016695999436908297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,65536,0.06747377581066556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,8192,0.01554844445652432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,8192,0.02548533346917894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,8192,0.01479022204875946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,7168,0.023696889479955036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,7168,0.014354666074117025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,16384,0.021280889709790547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,7168,0.013493333425786761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,6144,0.022421333524915908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,6144,0.013167111410035027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,6144,0.012689777546458773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,5120,0.020920889245139226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,12288,0.0173973325226042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,5120,0.010885333021481832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,5120,0.011384000380833944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,10240,0.015461333923869662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,4096,0.009560888840092553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,8192,0.013215111361609565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,4096,0.008975110948085785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,3584,0.01805688937505086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,3584,0.008827555510732863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,7168,0.011556444068749746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,3584,0.008566222257084316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,6144,0.010046222143703038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,3072,0.007921777665615082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,3072,0.007608888877762689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,5120,0.008715555899673039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,2560,0.01955200069480472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,2560,0.0069573331210348345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,2560,0.007131555842028723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,4096,0.007995555798212687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,2048,0.018888889087571036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,2048,0.005931555396980709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,2048,0.006759111252095964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,3584,0.007626666790909237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,1536,0.017475555340449016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,1536,0.004851555658711327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,1536,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,3072,0.007317333585686154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,1024,0.016116445263226826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,65536,0.0953671137491862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,1024,0.004184000194072723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,1024,0.005946666830115848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,768,0.01755466726091173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,2560,0.007245333658324347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,768,0.003629333443111844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,768,0.005592888842026393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,512,0.014416888356208801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,512,0.0034968890249729156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,512,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,2048,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,1536,0.006221333311663733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,256,0.0031991110493739447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,256,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,1024,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,128,0.014814222852389017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,128,0.002875555513633622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,128,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,768,0.005557333429654439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,64,0.002858666703104973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,64,0.005576000031497743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3584,32,0.002838222309947014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3584,32,0.005673777725961473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,512,0.00553955551650789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,65536,0.07047288947635226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,65536,0.09173511134253608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,65536,0.06543644269307454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,256,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,16384,0.03396177623007033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,16384,0.021581333544519212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,16384,0.021202666891945735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3584,128,0.005277333160241445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,12288,0.029150221082899306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,12288,0.018073777357737224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,12288,0.017937777770890128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,65536,0.06617066595289442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,4096,4096,0.020617778102556866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,10240,0.026425777210129633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,10240,0.015846222639083862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,10240,0.01615111033121745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,16384,0.020945777495702107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,8192,0.02496533261405097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,8192,0.013819555441538492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,8192,0.014127110441525778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,7168,0.023423110445340473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,12288,0.017139555679427255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,7168,0.012987555729018318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,7168,0.01310488912794325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,6144,0.021963554951879714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,6144,0.011888888974984487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,6144,0.012351999680201212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,10240,0.015170666906568738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,5120,0.020384889509942796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,5120,0.010695999695195092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,5120,0.010680889089902243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,8192,0.01257333325015174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,4096,0.009490666290124258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,4096,0.00831288927131229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,7168,0.010584000084135266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,3584,0.01685066686736213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,3584,0.008332444561852349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,3584,0.008276444342401292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,6144,0.009311111436949836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,3072,0.016181333197487723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,3072,0.007970666719807519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,3072,0.00758755538198683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,5120,0.00866844422287411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,2560,0.018569777409235638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,2560,0.007279111279381647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,4096,0.00797333319981893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,2560,0.007256000406212277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,2048,0.017528888252046373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,2048,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,2048,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,3584,0.007600888609886169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,1536,0.016547555724779766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,1536,0.00488800017370118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,1536,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,3072,0.007349333001507654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,1024,0.015251555376582675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,1024,0.0041964443193541635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,1024,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,2560,0.006952889263629913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,768,0.01719288859102461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,768,0.0038942222793896994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,768,0.005711110929648082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,2048,0.006576889091067844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,512,0.013993778162532382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,512,0.0035582222044467926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,512,0.005920889063013925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,1536,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,256,0.015459555718633862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,256,0.003156444471743372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,256,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,1024,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,128,0.014145778285132514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,128,0.0028604444944196274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,768,0.00555288874440723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,128,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,64,0.0024968888610601425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,64,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,3072,32,0.0028613333900769553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,3072,32,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,512,0.005467555589146084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,65536,0.05979022052552965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,256,0.005215999980767568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,65536,0.06301066610548231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,16384,0.03352266550064087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,16384,0.01962577799956004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,3072,128,0.0052800000541739995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,16384,0.02054755555258857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,12288,0.02852355440457662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,12288,0.0173680004146364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,12288,0.01722044414944119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,65536,0.06498666604359944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,10240,0.026036444637510512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,10240,0.01494488947921329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,16384,0.020621332857343886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,10240,0.015951999359660678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,4096,0.0195795562532213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,8192,0.012910222013791403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,12288,0.016736888223224215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,8192,0.014019555515713163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,3072,0.01686844395266639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,7168,0.02236355510022905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,7168,0.012079999678664737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,7168,0.012911111116409302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,6144,0.020915556285116408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,6144,0.011239111423492432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,6144,0.011694221860832639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,10240,0.014448889427714877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,8192,0.011397333608733283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,5120,0.010142222046852112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,7168,0.010036444498433007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,5120,0.010029333333174387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,6144,0.009365333451165093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,4096,0.017401778035693698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,4096,0.009004444711738164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,4096,0.008007999923494127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,5120,0.008665777742862701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,3584,0.016506666938463848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,3584,0.008263111114501953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,3584,0.007697777615653143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,4096,0.007991111112965478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3584,256,0.015222221612930298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,3072,0.007592889169851939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,3072,0.007635555333561367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,3584,0.007664889097213745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,2560,0.018223999275101554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,2560,0.006726222319735422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,2560,0.007289778027269576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,2048,0.017049777838918898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,2048,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,2048,0.006573333508438534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,3072,0.007304000357786815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,1536,0.016031111280123394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,1536,0.004572444491916233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,1536,0.006241777704821692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,2560,0.0069297779765393995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,1024,0.01480977733929952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,1024,0.003987555702527364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,1024,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,2048,0.006532444308201472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,1536,0.006243555496136348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,768,0.0038968887594011095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,768,0.00592533333433999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,1024,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,512,0.013776000175211163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,512,0.0035626664757728577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,512,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,3072,4096,0.017920888132519193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,256,0.015280889140235053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,768,0.005581333405441708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,256,0.003160888950030009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,256,0.005968889014588461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,128,0.01482488877243466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,128,0.002895999906791581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,128,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,512,0.0053031109273433685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,64,0.0026782221264309357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,64,0.005592888842026393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2560,32,0.002862222285734283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2560,32,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,65536,0.08271911409166124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,256,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,65536,0.049238221512900464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,65536,0.06025511026382446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2560,128,0.005249777601824866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,16384,0.031720889939202204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,16384,0.017614222235149808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,16384,0.019892444213231403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,65536,0.06421422296100192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,12288,0.01535999940501319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,12288,0.01696711116366916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,16384,0.020280000236299302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,10240,0.0247688889503479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,10240,0.013890667094124688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,10240,0.015551111764378019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,12288,0.01571999986966451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,8192,0.022999111149046157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,8192,0.013099555340078143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,8192,0.013110222087966071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,10240,0.012142221960756512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,7168,0.02131822208563487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,7168,0.01182755579551061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,7168,0.012069332930776807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,65536,0.09019466903474595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,6144,0.018908444378111098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,8192,0.011200000014570026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,6144,0.011050666371981302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,6144,0.010033778018421596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,5120,0.017543110582563613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,5120,0.01015822258260515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,7168,0.009695111049546136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,5120,0.00906844437122345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,4096,0.016521778371598985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,4096,0.008540444076061249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,4096,0.00831200016869439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,6144,0.009315555294354757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,3584,0.01612444387541877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,3584,0.007623111208279927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,3584,0.007644444704055786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,5120,0.008344888687133789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,3072,0.016184889607959323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,3072,0.006626666833957036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,4096,0.00797155582242542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,3072,0.007415999968846639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,2560,0.018909333480728995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,2560,0.005932444499598608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,2560,0.007312888900438945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,3584,0.00776444458299213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,8192,0.024181332853105333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,2048,0.01686044368478987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,2048,0.0052879999081293745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,2048,0.006717333363162146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,3072,0.007288888924651676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,1536,0.015824889143308003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,1536,0.00462311092350218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,5120,0.018900444110234577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,1536,0.006607999818192587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,2560,0.006958222223652734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,1024,0.014470222923490735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,1024,0.0038977778620190094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,1024,0.006243555496136348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,2048,0.006622222148709827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,768,0.00387999994887246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,768,0.005945777727497949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,1536,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,512,0.013781333135233985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,512,0.0034933334423436057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,1024,0.005938666562239329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,512,0.005710222240951326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,256,0.01514933341079288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,256,0.003168888803985384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,256,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,768,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,128,0.014059555199411182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,128,0.003160888950030009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,128,0.005534222142563925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,512,0.005274666680230035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,64,0.0028728888266616394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,64,0.005581333405441708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,2048,32,0.0028488888508743713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,256,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,2048,32,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,65536,0.03780711028310988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,65536,0.08199110958311293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,65536,0.05913955635494656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,2048,128,0.005320000151793162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,16384,0.015541333291265698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,16384,0.0192248887485928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,12288,0.02681511143843333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,12288,0.013791110780504016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,12288,0.016912889149453905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,65536,0.06329511271582709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,3072,0.015860444969601102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,10240,0.02331999937693278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,16384,0.018922666708628338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,10240,0.013570666313171387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,10240,0.01442488862408532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,8192,0.0118568887313207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,8192,0.020956445071432326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,8192,0.011685333318180509
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,7168,0.01961688862906562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,12288,0.013790222505728403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,7168,0.01125511113140318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,7168,0.010596444209416708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,10240,0.012078222301271228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,6144,0.018227555685573153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,6144,0.010256888965765635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,6144,0.00980088859796524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,8192,0.010737777584128909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,5120,0.01699733402993944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,5120,0.009820444716347588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,5120,0.009337777892748514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,4096,0.016179554992251925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,7168,0.009667555491129557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,4096,0.009018667042255402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,4096,0.00794400026400884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,3584,0.016138666205936007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,6144,0.009008000294367472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,3584,0.007558222446176741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,3584,0.007890666524569193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2560,768,0.016872000363137987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,5120,0.00831288927131229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,3072,0.006743999818960826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,3072,0.0075866662793689305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,4096,0.007923555870850882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,2560,0.017885333961910672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,2560,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,2560,0.006983111302057902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,3584,0.0076044441925154785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,2048,0.01678222252262963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,2048,0.0052933332820733385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,2048,0.006622222148709827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,3072,0.007033777733643849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,1536,0.004546666724814309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,1536,0.006278222219811545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,2560,0.007257777783605788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,1024,0.014431110686726041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,1024,0.0041999999019834726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,2048,0.006579555571079254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,1024,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,768,0.015814221567577787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,768,0.0036720000207424164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,1536,0.005969777703285217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,768,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,512,0.013101333545313941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,512,0.0035048888789282907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,512,0.0058106668293476105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,1024,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,256,0.015112888481881885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,256,0.0028400001012616684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,768,0.005382222019963794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,256,0.0058755555914507965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,512,0.005522666705979242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,128,0.014117333624098035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,128,0.0028391112056043413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,256,0.0052764444715446895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,128,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,12288,0.027449778384632532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,64,0.0026088888860411118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,64,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1536,32,0.002518222149875429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1536,32,0.00554933316177792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,65536,0.02737866673204634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,65536,0.08032355705897014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,65536,0.057244446542527944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1536,128,0.005199999858935674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,16384,0.03052088949415419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,16384,0.012886222038004132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,16384,0.019183999962276883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,12288,0.025495110286606684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,65536,0.06293600135379367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,12288,0.012067555553383298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,12288,0.015219555960761176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,10240,0.021934222843911912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,16384,0.017165333032608032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,10240,0.011502222054534487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,10240,0.012606222596433429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,12288,0.013719999955760108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,8192,0.020581333173645865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,8192,0.01001422190003925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,8192,0.010727110836240979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,10240,0.01203644441233741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,7168,0.019223110543357003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,7168,0.010346666806273991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,7168,0.009323555562231276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,6144,0.008943999807039896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,8192,0.01073155552148819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,6144,0.009879111415810054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,5120,0.01683822274208069
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,7168,0.009709333380063375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,5120,0.008656000097592672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,5120,0.009012444151772393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,4096,0.016201777590645682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,6144,0.009004444711738164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,4096,0.007963555554548899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,4096,0.00794488853878445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,3584,0.015615110596021017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,5120,0.008294222255547842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,3584,0.0068986668354935115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,3584,0.007600888609886169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,3072,0.015423110789722867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,4096,0.007921777665615082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,3072,0.006251555350091722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,3072,0.007274666594134436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,3584,0.007642666498819987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,2560,0.00591555568906996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,2560,0.0069191112286514705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,3072,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,2048,0.015126221709781222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,2048,0.0052355556852287715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,2048,0.006591111007663939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,2560,0.006912888752089606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,1536,0.01463733282354143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,1536,0.004592888885074192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,1536,0.0064026667839950984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,2048,0.0064088888466358185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,1024,0.013754666679435305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,1024,0.003879111260175705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,1024,0.005954666684071223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,1536,0.006239999913507038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,768,0.01587733295228746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,768,0.003527111063400904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,768,0.005913777897755305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,1024,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,512,0.01276622215906779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,512,0.0032088889016045462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,512,0.005882666756709416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,768,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,256,0.01510933372709486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,256,0.0031582222630580268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,256,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,512,0.0052151112920708126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,128,0.014107555150985718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,128,0.0028311111446883944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,128,0.005215999980767568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,256,0.005254222287072076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,64,0.002557333351837264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,64,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,2048,768,0.016181333197487723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,1024,32,0.002862222285734283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,1024,128,0.0052266667286554975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,1024,32,0.005585777676767773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,65536,0.08023111025492351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,65536,0.02533333334657881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,65536,0.05640000104904175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,16384,0.012233778006500669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,16384,0.01756444407833947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,65536,0.06257066461775038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,16384,0.016511110795868766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,12288,0.010341333018408881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,12288,0.014093332820468478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,16384,0.031155556440353394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,12288,0.013775111072593264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,10240,0.02198133369286855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,10240,0.010447111394670276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,10240,0.01238044434123569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,10240,0.012198222180207571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,8192,0.020260444945759244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,8192,0.00962933318482505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,8192,0.010784000158309937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,8192,0.01071200023094813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,7168,0.018929777873886954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,7168,0.008660444782839881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,7168,0.01037333326207267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,6144,0.017871111631393433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,6144,0.008329778081840938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,6144,0.009512888888518015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,7168,0.009694221946928237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,5120,0.017151110702090792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,5120,0.008293333152929941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,5120,0.008989333278603023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,6144,0.009014222357008193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,4096,0.015837333268589444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,4096,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,4096,0.008003555238246918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,5120,0.008353778057628209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,3072,0.015501333607567681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,3584,0.015826667348543804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,3584,0.0069777775141927935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,3584,0.00759733302725686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,4096,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,3072,0.015475556254386902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,3072,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,3072,0.007359999749395583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,3584,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,2560,0.005950222412745158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,2560,0.007199111084143321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,3072,0.007079111205206976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1536,1536,0.01587733295228746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,2048,0.014575999644067554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,2048,0.005241777747869492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,2048,0.006794666664467917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,2560,0.0069724445541699724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,1536,0.013785777820481194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,1536,0.004603555632962121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,1536,0.006575111299753189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,2048,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,1024,0.013078221844302284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,1024,0.0038888889054457345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,1024,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,1536,0.00627377794848548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,768,0.01516533394654592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,768,0.003543111185232798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,768,0.0059262220230367445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,512,0.0127697777416971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,1024,0.005583111196756363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,512,0.003559999995761447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,512,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,256,0.014458666245142618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,768,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,256,0.0031866667171319327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,256,0.005575110928879843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,512,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,128,0.014085332552591959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,128,0.0028897778441508612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,128,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,64,0.0028693332440323303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,256,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,64,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,768,32,0.0025066667132907445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,768,32,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,65536,0.022053332792388067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,768,128,0.0052151112920708126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,65536,0.056242664655049644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,16384,0.027474666635195415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,65536,0.06249688731299507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,16384,0.011058666639857821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,16384,0.016754666964213055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,12288,0.023676445086797077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,16384,0.016520889268981088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,12288,0.00905866672595342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,12288,0.013790222505728403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,10240,0.02130399975511763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,10240,0.009357333183288574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,12288,0.013398222625255585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,10240,0.012709333664841123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,6144,0.017855111095640395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,8192,0.00831911133395301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,8192,0.010684444672531553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,10240,0.011764444410800934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,7168,0.018537777993414137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,7168,0.007987555530336168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,8192,0.010670222342014313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,7168,0.010009778042634329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,7168,0.009698666632175446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,6144,0.008632000121805403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,6144,0.009387555221716562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,6144,0.009011555876996782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,5120,0.016171556380059984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,5120,0.00795111142926746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,5120,0.00867555538813273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,1024,2560,0.01683022247420417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,5120,0.00831200016869439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,4096,0.007576000359323289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,4096,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,4096,0.007890666524569193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,3584,0.014771555860837301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,16384,0.02901511059867011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,3584,0.007080888582600488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,12288,0.02440711193614536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,3584,0.007264888948864407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,3072,0.014759999182489185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,3072,0.006688888702127669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,3072,0.00758755538198683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,3584,0.007358222372002072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,2560,0.01408622165520986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,2560,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,2560,0.007253333098358578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,3072,0.006963555183675554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,2048,0.01349866638580958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,2048,0.00510044437315729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,2048,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,2560,0.006884444504976273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,1536,0.013400000002649097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,1536,0.0047173334492577445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,1536,0.006280000011126201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,2048,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,1024,0.012727110750145383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,1024,0.0038524443904558816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,1024,0.006109333286682765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,1536,0.006034666465388403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,768,0.015125332607163323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,768,0.003499555504984326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,768,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,1024,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,512,0.012366222010718452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,512,0.0032257777121331957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,512,0.005555555638339784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,768,0.005582222094138463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,256,0.014748444159825643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,256,0.0031617778456873367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,256,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,512,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,128,0.002862222285734283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,128,0.005222222043408289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,256,0.0052435555391841466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,64,0.0025155554629034465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,64,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,512,128,0.005192888693677055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,512,32,0.002830222249031067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,512,32,0.005211555709441503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,65536,0.01532799998919169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,65536,0.07856088876724243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,65536,0.055389334758122764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,16384,0.027112000518374976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,65536,0.058261334896087646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,16384,0.009328000247478485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,16384,0.015809777710172866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,12288,0.02306755549377865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,16384,0.016136889656384785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,12288,0.009645333720578088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,12288,0.013516444298956128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,10240,0.020595555504163105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,12288,0.013312888642152151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,10240,0.008960000342792934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,10240,0.012281777958075205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,10240,0.012051555845472546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,8192,0.00833422193924586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,8192,0.0107342220014996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,8192,0.010607110957304636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,7168,0.017877333694034152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,7168,0.007785778078767989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,7168,0.0103493332862854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,768,2560,0.015487111277050443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,6144,0.017186666528383892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,7168,0.009659555223253038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,6144,0.007294221884674496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,6144,0.00943022221326828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,6144,0.009004444711738164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,5120,0.006909333169460297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,5120,0.00906666699382994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,5120,0.008330666356616551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,4096,0.015141333142916361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,4096,0.006630222416586346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,4096,0.00798044436507755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,4096,0.00792266676823298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,3584,0.014648000399271647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,3584,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,3584,0.007917332980367873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,3584,0.007609777980380588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,3072,0.014488889111412896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,3072,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,3072,0.007598222129874759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,3072,0.007292444507280986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,2560,0.013724444641007317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,2560,0.0063288890653186375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,2560,0.007069333559936947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,2560,0.006908444480763541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,2048,0.013106666505336761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,2048,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,2048,0.0069004446268081665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,65536,0.07948444286982219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,2048,0.00627377794848548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,1536,0.012436444560686747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,1536,0.004592000196377437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,1536,0.006305777778228124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,1536,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,1024,0.011526222030321756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,1024,0.0038728887836138406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,1024,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,8192,0.019916445016860962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,1024,0.005592888842026393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,768,0.014818666709793938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,768,0.003531555748648114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,768,0.005963555640644497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,512,0.011738667057620155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,768,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,512,0.0033271111961868075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,512,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,256,0.011413333316644033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,512,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,256,0.0028595555987623003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,6144,0.01753155555990007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,256,0.005643555687533484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,128,0.01071911139620675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,128,0.0025600000388092464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,256,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,128,0.005247999810510212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,64,0.002521777732504739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,64,0.0052782222628593445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,256,32,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,256,128,0.005274666680230035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,256,32,0.00526311124364535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,65536,0.01533333294921451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,65536,0.07517066929075453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,65536,0.05370133452945285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,16384,0.02621511121590932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,16384,0.008903111020723978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,65536,0.051009777519438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,16384,0.015840889679061044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,12288,0.007975111405054728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,16384,0.016194666425387066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,12288,0.013723555538389417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,4096,0.015615110596021017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,12288,0.013402666482660504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,10240,0.02016266683737437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,10240,0.0069724445541699724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,10240,0.0123857781291008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,10240,0.011762667033407422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,8192,0.018622222873899672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,8192,0.007234666910436418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,8192,0.010904889139864178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,8192,0.01069155583779017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,7168,0.017675555414623685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,7168,0.006618666566080517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,7168,0.0103511114915212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,6144,0.016524444023768108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,7168,0.009687110781669617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,6144,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,6144,0.009453333086437648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,5120,0.00620266670982043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,6144,0.009014222357008193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,5120,0.015134221977657743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,5120,0.009321778184837764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,5120,0.00832711077398724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,4096,0.01445777714252472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,4096,0.006440889090299606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,4096,0.007992000215583378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,4096,0.00795377790927887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,3584,0.013797333670987023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,3584,0.005954666684071223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,3584,0.007607111500369177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,3584,0.007599111232492659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,3072,0.006085333310895496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,3072,0.007326222128338284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,3072,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,2560,0.012784889174832238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,2560,0.0063475556671619415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,2560,0.007288888924651676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,2048,0.013065777719020844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,2048,0.005929777605666055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,2560,0.006607111129495833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,2048,0.006687110910813014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,2048,0.006607111129495833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,1536,0.0042408886882993905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,1536,0.011751111182901593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,1536,0.0062977779242727495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,1536,0.005956444475385878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,1024,0.011388444238238864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,1024,0.00388355553150177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,1024,0.005943999936183293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,768,0.0035537779331207275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,1024,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,768,0.005903999838564131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,512,0.011022222538789114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,768,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,512,0.0031759999692440033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,512,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,512,0.005379555539952383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,256,0.0028213332924577924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,256,0.005544888890451855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,512,128,0.012431110772821637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,256,0.00526577772365676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,128,0.011005333728260465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,128,0.00282577777074443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,128,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,48,128,128,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,64,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,64,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,128,32,0.002472888885272874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,128,32,0.005623111294375525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,65536,0.0127688886390792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,65536,0.04976266622543335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,16384,0.00794666674402025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,16384,0.016189333465364244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,12288,0.007595555649863348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,8192,0.01889866590499878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,12288,0.013563555147912769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,10240,0.0069573331210348345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,10240,0.012735111018021902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,8192,0.006955555743641323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,8192,0.010407999985747868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,7168,0.006302222195598815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,7168,0.01035199976629681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,6144,0.005984000033802456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,6144,0.009662222531106737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,5120,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,5120,0.008958222137557136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,4096,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,4096,0.00794311116139094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,3584,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,3584,0.007886221839321984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,3072,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,3072,0.007287110719415877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,2560,0.006261333409282897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,2560,0.006959999601046245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,2048,0.005528888768619961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,2048,0.0069297779765393995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,1536,0.0042204442951414315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,1536,0.006213333457708359
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,1024,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,1024,0.0038382220599386427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,768,0.0032053333189752367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,768,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,512,0.0031848889258172777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,512,0.005542222410440445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,256,0.0028719999310043124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,256,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,128,0.00553955551650789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,128,0.002830222249031067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,64,0.0025173332542181015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,64,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,64,32,0.0025395556456512874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,256,5120,0.015524443652894763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,64,32,0.005568000177542369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,65536,0.047576887740029224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,65536,0.012236444486512078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,16384,0.007985778152942657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,16384,0.01625155574745602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,10240,0.006996444529957241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,12288,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,12288,0.013734222286277346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,10240,0.01238044434123569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,8192,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,8192,0.01070577816830741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,7168,0.00655733338660664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,6144,0.009306666751702627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,7168,0.010335110955768162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,6144,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,5120,0.006217777729034424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,5120,0.00867377801073922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,3584,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,4096,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,4096,0.00795111142926746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,3584,0.007629333270920648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,3072,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,3072,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,2560,0.005617777920431561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,2048,0.006586666736337874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,2560,0.0069208890199661255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,2048,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,1536,0.004316444612211651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,1024,0.006053333481152852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,1536,0.006299555715587404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,1024,0.003544888976547453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,768,0.003516444315512975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,768,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,256,0.0028364445186323593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,512,0.0031679999083280563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,512,0.005412444472312927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,128,0.005573333137565189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,64,0.00253866674999396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,256,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,128,0.0028408887899584244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,32,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,48,32,64,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,48,32,32,0.0024666666156715816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,16384,0.362643559773763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,16384,0.23324799537658691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,16384,0.16791733105977377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,12288,0.18626933627658418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,12288,0.2744764486948649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,12288,0.1256426705254449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,10240,0.16178045007917616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,10240,0.22747998767428926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,12288,0.023000000251664057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,10240,0.10645422008302476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,8192,0.14735378159417048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,3072,0.013461333182122974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,8192,0.18449066744910347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,768,0.01310488912794325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,8192,0.0908622211880154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,7168,0.16170399718814427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,7168,0.12804088327619764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,7168,0.07583822144402398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,6144,0.13720088534884983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,6144,0.11803377999199761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,48,128,256,0.010391111175219217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,6144,0.0673111081123352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,5120,0.11694310771094428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,5120,0.10583821932474773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,5120,0.05685244666205513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,4096,0.09564266602198283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,4096,0.09524089097976685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,4096,0.046671112378438316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,16384,0.1396426624721951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,12288,0.10498311122258504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,10240,0.0896222194035848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,3584,0.0837084452311198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,3584,0.04673333300484551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,3072,0.08922666973537868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,3072,0.07095022333992852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,3072,0.040800889333089195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,8192,0.0724124444855584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,7168,0.06366399923960368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,2560,0.059302224053276904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,6144,0.05588622225655449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,5120,0.05076266659630669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,2560,0.03526755505137973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,2048,0.0488684442308214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,2048,0.07653688722186618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,2048,0.03070844544304742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,1536,0.06909777720769246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,1536,0.037654221057891846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,1536,0.02419288953145345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,4096,0.03999911083115472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,1024,0.06222044759326511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,1024,0.02760444415940179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,1024,0.019129777948061626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,3584,0.09568444225523208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,768,0.030488000975714788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,768,0.020396444532606337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,768,0.01720355616675483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,2560,0.08283911148707072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,512,0.05651466714011299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,512,0.014102222190962898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,512,0.01496266656451755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,3072,0.03160088923242357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,256,0.024893333514531452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,256,0.008336000144481659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,3584,0.035071998834609985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,256,0.01293955577744378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,65536,128,0.020096888144810993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,128,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,128,0.010055111514197456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,2048,0.023334221707450017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,64,0.00591555568906996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,64,0.01146311147345437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,65536,32,0.005960888746711943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,2560,0.02752888864941067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,65536,32,0.013460444079505073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,65536,0.3602142333984375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,65536,0.22184444798363578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,1536,0.01955111159218682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,65536,0.18513422542148164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,16384,0.0739191108279758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,16384,0.08011911312739055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,16384,0.05087200138303968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,1024,0.015440889530711703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,12288,0.05975111325581869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,12288,0.061386664708455406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,12288,0.03914666506979201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,10240,0.052829331821865506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,10240,0.051389333274629384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,10240,0.03424711028734843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,65536,0.16608711083730063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,8192,0.04828088813357883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,8192,0.04416266745991177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,8192,0.02968177861637539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,16384,0.04440266556209988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,7168,0.04320711228582594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,7168,0.03767111235194736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,12288,0.037167999479505755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,7168,0.025793777571784124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,6144,0.040652443965276085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,6144,0.03293244375122918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,6144,0.022999111149046157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,10240,0.029083556599087183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,5120,0.03708088729116652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,5120,0.028113779094484117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,768,0.013738666971524557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,5120,0.020231111182106864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,512,0.012067555553383298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,4096,0.02330844435426924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,4096,0.03478044602606032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,4096,0.01702666613790724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,3584,0.037348445918824934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,3584,0.020672000116772123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,3584,0.017532444662517972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,8192,0.02568711174858941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,3072,0.035311109489864774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,3072,0.01852888862291972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,3072,0.01640533407529195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,7168,0.022101332743962605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,2560,0.03292266527811686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,2560,0.015850666496488783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,2560,0.01479822231663598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,6144,0.02067733307679494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,2048,0.03053599927160475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,2048,0.013399110900031196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,2048,0.013083555632167392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,5120,0.018558222386572096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,1536,0.028819554381900366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,1536,0.010817777779367236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,1536,0.011668444507651858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,128,0.01034311122364468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,1024,0.026527111728986103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,1024,0.008575999902354347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,1024,0.009483555952707926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,65536,256,0.010679110884666443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,768,0.020257777637905545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,768,0.0069217777086628815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,768,0.008963555925422246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,4096,0.01590577761332194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,512,0.02174577779240078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,512,0.0052044445441828836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,3584,0.014460444450378418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,512,0.007028444773621029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,3072,0.01351377781894472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,256,0.003516444315512975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,256,0.006640888750553131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,2560,0.012611555556456247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,2048,0.011167999770906238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,128,0.0032195556494924757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,128,0.006367110957702001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,1536,0.00964355551534229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,64,0.002847111059559716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,64,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,16384,32,0.003169777699642711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,16384,32,0.007227555745177799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,65536,0.16842133469051787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,1024,0.008338666624493068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,65536,0.2564293278588189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,65536,0.15834221574995253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,768,0.0069368887278768755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,16384,0.05953333112928602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,16384,0.07305066453086005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,16384,0.0442826681666904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,512,0.006631999793979857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,12288,0.05753866831461588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,12288,0.048489779233932495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,12288,0.03484800126817491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,10240,0.04320711228582594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,10240,0.04725689027044508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,10240,0.030163556337356567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,256,0.006534222099516127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,8192,0.03977955712212457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,8192,0.03814488980505202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,8192,0.02623644471168518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,16384,128,0.006539555473460092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,7168,0.035954667462242976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,7168,0.03303555647532145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,7168,0.022987556126382615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,65536,0.14704710907406277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,6144,0.03389422098795573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,6144,0.028894222444958154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,16384,0.03945066531499227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,6144,0.020941333638297185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,5120,0.03119200136926439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,12288,0.03460888730155097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,5120,0.02474311159716712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,5120,0.018562666243977018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,4096,0.02972266740269131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,4096,0.02167822255028619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,4096,0.015498666299713982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,10240,0.027035555905765955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,3584,0.03216177887386746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,3584,0.018553778529167175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,3584,0.016530666086408827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,8192,0.022960889670583937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,3072,0.030716445710923936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,3072,0.016435555285877652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,3072,0.015093333191341825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,7168,0.020568889048364427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,2560,0.028964443339241877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,2560,0.014092443717850579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,256,0.017196445001496207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,2560,0.013459555804729462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,2048,0.026919111609458923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,2048,0.011688888900809817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,2048,0.012299555871221753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,6144,0.018401778406567044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,16384,128,0.015459555718633862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,5120,0.017151110702090792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,1536,0.00962044464217292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,1536,0.01105955574247572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,4096,0.01460800071557363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,3584,0.013654222091039022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,1024,0.008246222303973304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,1024,0.00831288927131229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,3072,0.012710221939616732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,768,0.018552000323931377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,768,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,768,0.007608888877762689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,512,0.019825778073734708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,512,0.004872888740566042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,512,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,2560,0.011672889192899069
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,256,0.016862221890025668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,256,0.003834666477309333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,256,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,2048,0.010112889111042023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,128,0.01510933372709486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,128,0.00349866681628757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,128,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,1536,0.008271999657154083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,64,0.003181333343187968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,64,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,12288,32,0.0031724443866146933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,12288,32,0.0069546666410234236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,1024,0.007189333438873291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,65536,0.150273773405287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,65536,0.22896620962354872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,768,0.006759111252095964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,65536,0.1492346657647027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,16384,0.053526222705841064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,16384,0.06204444169998169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,512,0.006600889066855113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,16384,0.04246311055289375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,12288,0.04417333337995741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,12288,0.04800800151295132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,12288,0.033407111962636314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,10240,0.03905422157711453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,10240,0.04117155406210158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,256,0.006213333457708359
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,10240,0.029194666279686823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,12288,128,0.006241777704821692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,8192,0.03502310978041755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,8192,0.024901333782407973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,65536,0.1425351169374254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,7168,0.03277511066860623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,7168,0.027446221974160936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,16384,0.03846044341723124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,7168,0.022298667165968154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,6144,0.031178666485680476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,12288,0.03161511156294081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,6144,0.02367111047108968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,6144,0.02017155620786879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,5120,0.029135998752382066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,5120,0.020868443780475195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,10240,0.025739555557568867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,5120,0.018005333013004728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,4096,0.027902222341961328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,4096,0.017910222212473553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,4096,0.015416888727082146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,8192,0.021819555097156104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,3584,0.0314355558819241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,3584,0.015712888704405892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,3584,0.015164444843928019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,1536,0.025676444172859192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,3072,0.029275556405385334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,3072,0.014056889547242058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,3072,0.013850666582584381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,7168,0.019648000597953796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,2560,0.027963555521435205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,12288,1024,0.021669333179791767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,2560,0.012359999948077731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,6144,0.018216888109842937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,2560,0.013016000390052795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,2048,0.025461332665549383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,2048,0.0107022225856781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,2048,0.011391110718250275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,5120,0.01668711172209846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,1536,0.00924355536699295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,1536,0.009674666656388177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,4096,0.014294221997261047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,1024,0.020818667279349435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,1024,0.007681777907742395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,1024,0.00795999997191959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,3584,0.013154666456911298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,768,0.01824711097611321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,768,0.005964444329341252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,768,0.007256000406212277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,3072,0.012300444145997366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,512,0.01891644464598762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,512,0.0047288888858424295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,512,0.006619555668698416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,2560,0.011000889043013254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,256,0.017162667380438913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,256,0.0038311113086011675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,2048,0.008769777913888296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,256,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,1536,0.007921777665615082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,128,0.014775110615624322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,128,0.0034728890491856467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,128,0.006059555543793573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,1024,0.0069884442620807225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,64,0.003068444422549672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,64,0.006596444381607904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,10240,32,0.0031759999692440033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,10240,32,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,768,0.006353777729802662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,65536,0.13209954897562662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,65536,0.17604622575971815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,512,0.006575999988449945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,65536,0.09445510970221625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,16384,0.04759022262361315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,16384,0.04655111167165968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,16384,0.029888888200124104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,12288,0.04031999905904134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,12288,0.03649066554175483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,12288,0.02409422232045068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,8192,0.03634399837917752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,10240,0.028024888700909082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,256,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,10240,0.021634666456116572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,8192,0.0329039990901947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,8192,0.024344000551435683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,8192,0.01835555500454373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,10240,128,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,65536,0.0955431130197313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,7168,0.021015110943052504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,7168,0.016851555969980028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,16384,0.025928889711697895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,6144,0.028738667567571003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,6144,0.01850666602452596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,6144,0.015496888094478183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,12288,0.022094221578703985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,10240,0.018616000811258953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,5120,0.01590666671593984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,5120,0.013946667313575745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,8192,0.01588444411754608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,4096,0.025832888152864244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,4096,0.013798221945762634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,4096,0.012223111258612739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,7168,0.015145777000321282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,3584,0.02920177910063002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,3584,0.012076444096035428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,3584,0.012420444852775998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,6144,0.01350400017367469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,3072,0.0273991111252043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,3072,0.010999110837777456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,3072,0.011741333537631564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,5120,0.012854222622182635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,4096,0.011147555377748279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,10240,1536,0.0236835562520557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,2560,0.009684444301658207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,2560,0.010363555616802638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,2048,0.02385155525472429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,2048,0.008294222255547842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,2048,0.009682666924264696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,3584,0.010194666683673859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,1536,0.021651556094487507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,1536,0.007598222129874759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,1536,0.008289777570300633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,3072,0.008660444782839881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,1024,0.019229332605997723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,1024,0.0057955553962124725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,1024,0.0069226668112807805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,2560,0.007989333735571967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,768,0.01795288920402527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,768,0.004566222015354368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,768,0.006619555668698416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,2048,0.007191999918884701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,512,0.01718933383623759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,512,0.003859555555714501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,512,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,1536,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,256,0.016565332810084026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,256,0.003216000066863166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,256,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,1024,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,128,0.015127999915017022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,128,0.0031866667171319327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,128,0.005620444400442972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,768,0.0059368887709246735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,64,0.0028799999919202593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,64,0.006056889063782162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,8192,32,0.00282577777074443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,8192,32,0.006284444282452266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,512,0.005625777774386936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,7168,0.029954665237002905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,65536,0.1529573334587945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,65536,0.12970399856567383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,10240,0.03523644473817613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,16384,0.043848888741599194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,16384,0.045927110645506114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,65536,0.09094133641984727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,16384,0.02797422144148085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,12288,0.03520177801450094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,256,0.005555555638339784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,12288,0.022533333963818018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,8192,128,0.005551110953092575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,10240,0.03429422113630507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,10240,0.029871109459135268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,10240,0.02012800011369917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,5120,0.026714666022194758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,8192,0.03148533238304986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,8192,0.02218222286966112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,8192,0.017666666044129264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,65536,0.09350044197506374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,7168,0.028767112228605483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,16384,0.02384533319208357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,7168,0.018971555762820773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,7168,0.016064888901180692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,6144,0.02787289023399353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,12288,0.020609777834680345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,6144,0.016826666063732572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,6144,0.0147724449634552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,5120,0.025613332788149517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,5120,0.014589332871966891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,5120,0.013559111290507846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,4096,0.02421866688463423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,10240,0.01681511104106903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,4096,0.013087111214796701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,4096,0.01146311147345437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,3584,0.027888000011444092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,3584,0.011343999869293638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,3584,0.011741333537631564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,8192,0.0151555554734336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,3072,0.026668444275856018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,3072,0.00999111102686988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,3072,0.011121778024567498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,8192,2560,0.025918222135967676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,2560,0.02551377813021342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,7168,0.01479822231663598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,2560,0.008901333643330468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,2560,0.01036088913679123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,2048,0.022034666604465906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,2048,0.00795288880666097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,2048,0.008336000144481659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,6144,0.012952889005343119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,1536,0.02032977839310964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,1536,0.007280888656775157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,5120,0.012906666431162091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,1536,0.007312888900438945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,1024,0.018558222386572096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,1024,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,1024,0.0069057775868309875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,4096,0.010209778116808997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,768,0.017850667238235474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,768,0.004193777839342753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,768,0.006248000181383557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,3584,0.008950221869680617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,512,0.016914667354689706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,512,0.003556444413132138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,512,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,3072,0.008290666672918532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,256,0.01685599982738495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,256,0.0032071111102898917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,2560,0.007290666302045186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,256,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,128,0.015119110544522604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,128,0.0028293333533737394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,2048,0.0069671107663048645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,128,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,64,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,64,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,7168,32,0.0028284444577164124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,1536,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,7168,32,0.006235555642180973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,65536,0.12490933471255833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,1024,0.005955555372767978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,65536,0.12148177623748779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,768,0.005884444548024072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,16384,0.04319377740224203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,16384,0.03762222329775492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,65536,0.08563111225763957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,16384,0.026147555973794725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,12288,0.030677331818474665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,512,0.005562666803598404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,12288,0.021035555336210463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,10240,0.03239733311865065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,10240,0.026050666968027752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,10240,0.01889333294497596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,256,0.00554577757914861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,8192,0.030078222354253132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,8192,0.02199733257293701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,7168,128,0.0052666668262746595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,8192,0.01700266698996226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,7168,0.02776622275511424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,7168,0.0199297782447603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,7168,0.015261333849694995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,65536,0.07271822293599446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,6144,0.026424888107511733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,16384,0.023014222582181294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,6144,0.017872000734011333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,6144,0.01407911048995124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,12288,0.018795554836591084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,5120,0.01579466627703773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,5120,0.013088888592190213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,10240,0.016564443707466125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,8192,0.014494222071435718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,7168,12288,0.038033776813083224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,4096,0.023354666100607976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,4096,0.011904888682895236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,4096,0.011055111057228513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,3584,0.0273671117093828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,7168,0.013071999781661563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,3584,0.010406222608354356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,3584,0.0110871113008923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,3072,0.02600266701645321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,6144,0.012076444096035428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,3072,0.009278222090668148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,3072,0.010430222584141625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,2560,0.02335022224320306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,2560,0.00831288927131229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,2560,0.009844444692134857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,5120,0.011530666715568967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,2048,0.020631111330456205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,2048,0.007485333416197035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,2048,0.008005333443482717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,1536,0.018924444913864136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,1536,0.0069013333155049225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,1536,0.007165333463086023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,4096,0.009304000271691216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,1024,0.017496888836224873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,1024,0.004815111143721474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,1024,0.006583111153708563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,3584,0.007996444072988298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,768,0.01771288944615258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,768,0.004218666503826777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,768,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,3072,0.007610666255156199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,512,0.01588888963063558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,512,0.0036159998012913596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,512,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,2560,0.007277333074145847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,256,0.01686755485004849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,256,0.0031928889867332247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,256,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,2048,0.0069360000391801195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,128,0.014121777481502958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,128,0.003169777699642711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,128,0.005439110928111606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,1536,0.006255110932721033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,64,0.0028524444335036804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,64,0.005888888819350137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,6144,32,0.0028853333658642242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,6144,32,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,1024,0.005923555543025334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,65536,0.1112097766664293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,768,0.005924444645643234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,65536,0.11757422818077935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,16384,0.041329777903027005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,512,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,65536,0.07670399877760145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,16384,0.032920890384250216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,16384,0.024345777100986902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,12288,0.026477333572175767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,256,0.005571555760171678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,12288,0.034973333279291786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,12288,0.01980888843536377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,10240,0.02285066743691762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,10240,0.03142311175664266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,10240,0.01808444493346744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,8192,0.028772443532943726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,8192,0.0203128887547387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,6144,128,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,8192,0.0158942225906584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,7168,0.026808889375792608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,7168,0.017837334010336135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,7168,0.014497778481907315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,65536,0.0702071123652988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,6144,0.025193777349260118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,6144,0.016442666451136272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,6144,0.01315111087428199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,16384,0.022265776991844177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,5120,0.02397511071628994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,5120,0.014344889256689282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,12288,0.018290667070282828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,5120,0.01238933371173011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,10240,0.01591199967596266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,4096,0.010997333460383944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,4096,0.010687111152542962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,8192,0.014127110441525778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,3584,0.026387555731667414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,3584,0.009763555394278632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,3584,0.010592000352011787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,7168,0.012713777522246042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,3072,0.023910222782029047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,3072,0.008785777621799046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,3072,0.009679999616410997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,6144,0.011722666521867117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,12288,0.03617422117127313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,2560,0.021969777014520433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,2560,0.008066666622956594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,2560,0.008332444561852349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,5120,0.010032888915803697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,2048,0.01963911122745938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,2048,0.0069431112044387394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,2048,0.007928888830873702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,4096,0.008299555215570662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,1536,0.017887999614079792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,1536,0.006299555715587404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,1536,0.006931555353932911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,3584,0.008017777568764156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,3072,0.007592889169851939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,1024,0.00463733325401942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,1024,0.006616888774765863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,2560,0.007261333366235097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,768,0.017160000072585214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,768,0.004294222427739037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,2048,0.006739555547634761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,768,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,512,0.003547555456558863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,512,0.015331554743978711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,512,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,6144,5120,0.024469332562552557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,256,0.01648533344268799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,256,0.0032053333189752367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,256,0.005793777604897817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,128,0.014767110347747803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,128,0.003198222153716617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,1536,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,128,0.005537777725193236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,64,0.002842666581273079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,64,0.005757333503829108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,5120,32,0.002846222163902389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,5120,32,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,65536,0.09425510962804158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,1024,0.005919111271699269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,65536,0.08919644355773926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,768,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,65536,0.0706640018357171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,512,0.005615111026499007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,16384,0.03494488861825731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,16384,0.026919111609458923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,16384,0.021987555755509272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,12288,0.021228444245126512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,256,0.005378666851255629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,12288,0.018571555614471436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,10240,0.027315555347336665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,5120,128,0.005349333501524395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,10240,0.018783999813927543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,10240,0.016934222645229764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,8192,0.02546577817863888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,8192,0.01664622293578254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,8192,0.01514488955338796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,65536,0.06816533539030287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,7168,0.02397777802414364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,16384,0.02161955502298143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,7168,0.014711111783981323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,7168,0.013796444568369122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,6144,0.02296977738539378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,12288,0.01758666667673323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,6144,0.013643555343151093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,6144,0.012732444538010491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,5120,0.021425777011447485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,10240,0.015494222442309061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,5120,0.013109332985348172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,5120,0.011726222104496427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,4096,0.020344000723626878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,4096,0.011404444773991903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,8192,0.01368977791733212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,4096,0.010055111514197456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,3584,0.02366755571630266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,3584,0.01034400032626258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,3584,0.009323555562231276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,7168,0.012040889097584618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,6144,0.010832000109884473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,3072,0.009340444372759925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,3072,0.008303110798199972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,5120,0.009650666680600908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,4096,0.007985778152942657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,2560,0.008397333323955536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,2560,0.007703999678293864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,3584,0.007663999994595845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,2048,0.018911111685964797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,2048,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,2048,0.007249777515729268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,3072,0.007320000065697565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,1536,0.018253333038753934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,1536,0.004906666775544484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,1536,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,2560,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,1024,0.016813332835833233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,1024,0.003910222401221593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,1024,0.006234666539563074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,2048,0.006619555668698416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,768,0.017508443858888414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,768,0.003843555433882607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,768,0.006263999889294307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,1536,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,512,0.015116444892353482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,512,0.003571555432346132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,4096,0.02237422267595927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,512,0.005912888795137405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,256,0.015509333875444202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,256,0.00319377767542998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,256,0.005577777822812398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,1024,0.005842666659090254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,768,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,128,0.014496889379289416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,128,0.0028719999310043124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,128,0.005281777845488654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,512,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,64,0.0025315555847353404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,64,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,4096,32,0.0028693332440323303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,4096,32,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,256,0.005229333208666907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,65536,0.07619200150171916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,65536,0.09178400039672852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,16384,0.03394133183691237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,65536,0.06718577941258748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,16384,0.024726221958796184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,16384,0.021697777840826247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,4096,128,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,12288,0.0196079992585712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,12288,0.029595557186338637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,12288,0.01811200049188402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,10240,0.02665866745842828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,10240,0.017236444685194228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,10240,0.016514667206340365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,65536,0.06722044282489352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,8192,0.025042666329277888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,16384,0.02128444446457757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,5120,1024,0.016530666086408827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,8192,0.01513155632548862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,8192,0.014711999230914645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,12288,0.017218665944205392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,7168,0.02353511088424259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,7168,0.013775111072593264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,7168,0.013372444444232516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,6144,0.022208889325459797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,6144,0.013096888860066732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,6144,0.012460444536474017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,5120,0.020675554871559143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,10240,0.015305777390797934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,5120,0.012182222472296821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,5120,0.011401777466138205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,8192,0.013249778085284762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,4096,0.010709332923094431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,7168,0.011699555648697747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,4096,0.00905688852071762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,6144,0.009703111317422655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,3584,0.022667555345429316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,3584,0.010067555639478896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,3584,0.008343111309740279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,3072,0.021701332595613267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,5120,0.008967111508051554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,12288,0.03016800019476149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,3072,0.009030222064918941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,3072,0.007625777688291337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,2560,0.019938665959570143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,2560,0.008117333054542542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,2560,0.007299555672539606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,2048,0.01867377758026123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,2048,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,2048,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,4096,0.007954667011896769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,1536,0.017847999930381775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,1536,0.004871110949251386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,1536,0.0069288888739215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,3584,0.0076248885856734375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,1024,0.016520889268981088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,1024,0.0042071110672420925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,3072,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,1024,0.006267555471923616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,2560,0.006981333096822103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,768,0.0038248888320393036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,768,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,2048,0.006599111275540457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,512,0.014494222071435718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,512,0.0035457776652442086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,512,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,1536,0.006246222390068903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,1024,0.0058782220714622075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,256,0.003132444495956103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,256,0.005555555638339784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,768,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,3072,0.02258844508065118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,128,0.0028311111446883944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,128,0.005542222410440445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,512,0.005436444448100195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,64,0.002552888873550627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,64,0.005664000080691443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,4096,2560,0.020624000165197585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3584,32,0.0028435554769304064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3584,32,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,65536,0.06720177994834052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,65536,0.09028355280558269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,256,0.005249777601824866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,16384,0.03384800089730157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,65536,0.06492800182766385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,16384,0.021134222547213238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3584,128,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,16384,0.020978666014141504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,12288,0.017831110292010836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,12288,0.029088000456492107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,12288,0.01754399968518151
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,65536,0.06574666500091553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,10240,0.026107556290096704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,10240,0.015680889288584392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,16384,0.020917332834667627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,10240,0.015989333391189575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,8192,0.024322667055659827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,12288,0.01697422232892778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,8192,0.013752000199423896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,8192,0.014087110757827759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,7168,0.02289688918325636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,7168,0.0127697777416971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,7168,0.013038222160604266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,10240,0.014615999327765571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,6144,0.022092445029152766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,6144,0.011745777395036487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,8192,0.01184977756606208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,6144,0.012551999754375882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,5120,0.020440888073709276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,5120,0.011016888750924004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,7168,0.010647999743620554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,5120,0.011067555182509951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,4096,0.017889777819315594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,4096,0.010048888623714447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,4096,0.00867555538813273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,3584,0.022625777456495497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,3584,0.008906666603353288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,3584,0.00795022232664956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,6144,0.009346666435400644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,3072,0.021752888957659405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,3072,0.008271999657154083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,5120,0.008635555704434713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,3072,0.007649777664078607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,2560,0.01961866683430142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,4096,0.007978666987684038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,2560,0.007124444676770105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,2560,0.007303111255168915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,2048,0.017872000734011333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,2048,0.005543111099137201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,3584,0.00795555528667238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,2048,0.006912888752089606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,3072,0.007329777710967594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,1536,0.004594666676388847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,1536,0.006583111153708563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,2560,0.006959999601046245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,2048,0.006648889018429651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,1024,0.0038399998512532976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,1024,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,1536,0.00611111107799742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,768,0.01716888944307963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,768,0.0035342222286595237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,768,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,1024,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,512,0.014159111513031853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,512,0.0033893332713180115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,512,0.005623111294375525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,4096,0.019010666343900893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,256,0.015160888433456421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,768,0.00553688903649648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,256,0.002899555489420891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,256,0.005622222191757626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,128,0.01477599971824222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,128,0.0028204443968004654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,128,0.005551110953092575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,512,0.005456888841258154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,64,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,64,0.005655111124118169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,3072,32,0.002540444541308615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,3072,32,0.005557333429654439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,256,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,65536,0.08846311436759101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,65536,0.059438221984439425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,3072,128,0.00527555536892679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,65536,0.06259377797444661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,16384,0.03299288948376974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,16384,0.01976355579164293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,16384,0.020573332905769348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,65536,0.0647875534163581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,12288,0.015840889679061044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,768,0.017490666773584154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,12288,0.01736000014675988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,16384,0.020468443632125854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,10240,0.02544622288809882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,10240,0.014018666413095264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,10240,0.015835555063353646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,8192,0.02353422178162469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,8192,0.01257955531279246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,256,0.015243555108706156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,8192,0.013948443863126965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3584,128,0.01422933406300015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,7168,0.02219555609756046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,7168,0.011776888536082374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,7168,0.01274400038851632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,12288,0.016571554872724745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,6144,0.020263110597928364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,6144,0.010963555839326648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,6144,0.011383111278216044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,10240,0.014325333966149224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,5120,0.01015822258260515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,5120,0.009826666778988307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,8192,0.011030221978823343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,7168,0.010029333333174387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,4096,0.01683644453684489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,6144,0.009194666312800514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,4096,0.008938666847017076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,4096,0.008018666671382057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,3584,0.020625778370433383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,3584,0.008217777642938826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,3584,0.008005333443482717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,5120,0.008643555144468943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,3072,0.01957066688272688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,3072,0.007155555817815993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,3072,0.0075031113293435835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,4096,0.007976888782448238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,2560,0.01925155520439148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,2560,0.006238222122192383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,2560,0.007252444823582967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,3584,0.007638221813572778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,3072,0.007266666326257918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,2048,0.0052044445441828836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,2048,0.0069164443347189165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,1536,0.016215110818545025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,1536,0.004567999806669024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,2560,0.006912888752089606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,1536,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,1024,0.015160888433456421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,1024,0.0038657776183552216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,1024,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,2048,0.006573333508438534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,768,0.01717866626050737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,768,0.003506666670242945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,768,0.005945777727497949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,1536,0.006224888894293044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,512,0.013439999686347114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,512,0.0033582221302721235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,1024,0.005769777629110549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,512,0.005879111174080107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,256,0.014479110638300577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,256,0.0031902222997612427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,256,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,768,0.005622222191757626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,128,0.014767110347747803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,128,0.0028657778683635923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,128,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,512,0.005283555636803309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,64,0.002833777831660377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,64,0.0052693333062860705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2560,32,0.002850666642189026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,256,0.005209777918126848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,1536,0.016878222425778706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2560,32,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,65536,0.04798488815625509
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,65536,0.08126133018069796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,65536,0.059827552901373975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,16384,0.031035555733574763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,16384,0.016033778587977093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,16384,0.019831111033757527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2560,128,0.0052266667286554975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,12288,0.026904889278941687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,12288,0.013391999734772576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,12288,0.01681155628628201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,65536,0.06418133444256253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,10240,0.012031111452314587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,16384,0.01998488936159346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,10240,0.015508444772826301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,8192,0.02273422148492601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,12288,0.015800888339678448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,8192,0.013458666702111563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,8192,0.01277599980433782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,7168,0.02054755555258857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,7168,0.012471111284361945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,3072,1024,0.015589333242840238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,7168,0.011979555918110741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,6144,0.019104000594880845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,10240,0.013395555317401886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,6144,0.011729777687125735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,6144,0.011048888994587792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,5120,0.017234666479958426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,8192,0.01071200023094813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,5120,0.011009777585665384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,5120,0.00928177767329746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,4096,0.016809778081046212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,7168,0.010040889183680216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,4096,0.00959911114639706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,4096,0.008351999852392409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,6144,0.009145778086450364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,3584,0.008632000121805403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,3584,0.007655111451943715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,5120,0.008632889224423302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,3072,0.020572443803151447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,3072,0.006629333313968446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,4096,0.007977777885066139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,3072,0.007273777491516537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,2560,0.0192577772670322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,2560,0.005920889063013925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,2560,0.007286222444640265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,3584,0.007641777396202087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,2048,0.018185777796639335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,2048,0.005224888937340842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,2048,0.006581333362393909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,3072,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,2560,0.006989333364698622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,1536,0.016528000434239704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,1536,0.004573333180612988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,1536,0.006248888870080312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,2048,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,1024,0.015503111812803479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,1024,0.0038248888320393036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,12288,0.0285173323419359
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,1024,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,768,0.016875555117925007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,768,0.003818666769398583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,768,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,1536,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,512,0.013527111046844058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,512,0.0031804444475306403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,512,0.00554577757914861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,1024,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,256,0.015458666616015963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,256,0.0028799999919202593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,256,0.00554577757914861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,768,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,512,0.005335111171007156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,128,0.0028826666788922418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,128,0.005583999885453119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,64,0.002492444382773505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,64,0.005744888964626525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,256,0.005249777601824866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,2048,32,0.0028239999794297745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,2048,32,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,65536,0.03721066647105747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,65536,0.08034221993552314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,65536,0.05860533316930135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,2048,128,0.005215999980767568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,16384,0.030563556485705908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,16384,0.013419555293189155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,16384,0.01944533321592543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,12288,0.02676533328162299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,5120,0.018246221873495314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,12288,0.011993777420785693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,65536,0.0634568863444858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,12288,0.016511999898486666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,10240,0.023045332895384893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,10240,0.017273777061038546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,10240,0.01444355481200748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,16384,0.0191600008143319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,8192,0.020854221449957952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,8192,0.014783110883500842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,8192,0.011698666546079846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,7168,0.019443555010689628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,12288,0.013796444568369122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,7168,0.014105778601434497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,7168,0.010676444404655032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,10240,0.012103999654452005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,6144,0.018241778016090393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,6144,0.012494222157531314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,6144,0.009930666950013902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,8192,0.010689777632554373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,5120,0.01717422240310245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,5120,0.011352889239788055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,5120,0.008974221845467886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,4096,0.01644888851377699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,7168,0.009852444132169088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,4096,0.009689778089523315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,4096,0.007958222594526079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,3584,0.020584000481499564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,6144,0.009073778159088558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,3584,0.008979555633332994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,3584,0.00795555528667238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,5120,0.008631111019187504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,3072,0.019553777244355943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,3072,0.006515555497672822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,3072,0.007636444436179266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,4096,0.008036444584528605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,2560,0.018532445033391316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,2560,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,2560,0.00703111125363244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2560,2048,0.01721244388156467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,3584,0.007599111232492659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,2048,0.017091555727852713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,2048,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,2048,0.006713777780532837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,1536,0.015838222371207345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,1536,0.004543999830881755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,3072,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,1536,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,1024,0.015113777584499784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,1024,0.0038853333228164246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,2560,0.007038222418891058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,1024,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,768,0.016514667206340365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,768,0.0036115555299652945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,2048,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,768,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,512,0.013346666263209449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,512,0.0033288889875014624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,512,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,1536,0.006034666465388403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,256,0.014820444915029736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,256,0.0031928889867332247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,256,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,1024,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1536,128,0.01482488877243466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,768,0.005420444326268301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,128,0.0028648889727062653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,128,0.005247999810510212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,512,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,64,0.00282222218811512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,64,0.005559999909665849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1536,32,0.002521777732504739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1536,32,0.005582222094138463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,256,0.0052800000541739995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,65536,0.026836444934209187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,65536,0.07920266522301568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,65536,0.056941330432891846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,16384,0.029822223716311987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1536,128,0.005274666680230035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,16384,0.013565333353148567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,16384,0.018207111292415194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,12288,0.024404444628291663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,12288,0.011732444167137146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,65536,0.06320977873272367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,12288,0.014465777410401238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,16384,0.016537777251667447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,10240,0.010784000158309937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,10240,0.012703999876976013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,8192,0.019984000258975558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,12288,0.013415999710559845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,8192,0.01035911093155543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,8192,0.011518222590287527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,7168,0.01901244454913669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,7168,0.009199110998047723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,7168,0.010379555324713389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,10240,0.012087999946541257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,6144,0.018228444788191054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,6144,0.00798044436507755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,6144,0.009711999860074785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,8192,0.01069777790043089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,10240,0.024647111694018047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,5120,0.0074933336840735535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,5120,0.01719999975628323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,5120,0.00887022250228458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,7168,0.009733333355850643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,4096,0.016195555528004963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,4096,0.007638221813572778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,4096,0.00795733349190818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,6144,0.009005332986513773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,3584,0.017841777867741056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,3584,0.007272889216740926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,5120,0.00832266691658232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,3584,0.00794666674402025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,4096,0.007976888782448238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,3072,0.01757600075668759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,3072,0.006288888967699475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,3072,0.0074417781498697065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,2560,0.016843555702103507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,2560,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,3584,0.007430222299363878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,2560,0.006966222491529252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,2048,0.016137777103318107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,2048,0.005728000154097875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,2048,0.006751110984219446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,3072,0.007287110719415877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,1536,0.0052062223354975385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,1536,0.01445777714252472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,1536,0.006295111030340195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,1024,0.013783111340469785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,1024,0.004230222354332606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,1024,0.0059164443777667145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,2560,0.006629333313968446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,2048,0.0064035554726918536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,768,0.015113777584499784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,768,0.003846222327815162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,1536,0.0058319999112023254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,768,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,512,0.012732444538010491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,512,0.0035502223504914176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,1024,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,512,0.005615111026499007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,256,0.01479022204875946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,256,0.0032177778581778207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,768,0.0052764444715446895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,256,0.005569777968857024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,128,0.01479288935661316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,512,0.005261333452330695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,128,0.0028577778074476454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,128,0.005626666463083691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,64,0.0028284444577164124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,64,0.00552977787123786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,256,0.005078222188684675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,1024,32,0.0028400001012616684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,1024,32,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,1024,128,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,65536,0.022447999980714586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,65536,0.05641422006818983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,16384,0.027855111493004694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,3584,0.021307556165589228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,16384,0.014719110396173267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,16384,0.017549332645204332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,65536,0.062374220954047314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,12288,0.024111110303137038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,12288,0.010010666317409938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,16384,0.016528889536857605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,12288,0.014096000128322177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,10240,0.009665778113736046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,10240,0.012716444830099741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,12288,0.01347733371787601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,8192,0.020258666740523446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,8192,0.009032888544930352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,8192,0.010672888822025724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,10240,0.012228444218635559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,7168,0.01891999940077464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,7168,0.008098666866620382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,7168,0.010324444207880232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,6144,0.018187556001875136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,8192,0.010762666662534079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,6144,0.007594666547245449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,6144,0.009718221922715506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,7168,0.010018666585286459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,5120,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,5120,0.00909511082702213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,6144,0.009022222624884712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,4096,0.016152888536453247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,4096,0.0075093333919843035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,4096,0.008164444731341468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,5120,0.008653333617581261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,3584,0.017465778523021273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,3584,0.00702133360836241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,3584,0.00794311116139094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,4096,0.00795377790927887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,3072,0.016190222567982145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,3072,0.006884444504976273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,2048,128,0.01479111115137736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,3072,0.007608888877762689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,3584,0.007648888561460707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,2560,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,2560,0.015223999818166098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,2560,0.007019555403126611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,3072,0.007288888924651676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,2048,0.005224888937340842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,2048,0.015517334143320719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,2048,0.0068897778789202375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,1536,0.014441778262456259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,2560,0.0069306666652361555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,1536,0.004520888957712385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,1536,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,1024,0.003908444609906939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,1024,0.013772444592581855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,2048,0.0062977779242727495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,1024,0.005984000033802456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,768,0.003567999849716822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,768,0.01568622224860721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,768,0.005910222315125995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,1536,0.006057777752478917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,512,0.013065777719020844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,512,0.0031973332580592898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,512,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,1024,0.0057760001056724125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,768,0.005567111074924469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,256,0.014880888991885714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,256,0.0028968888024489083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,256,0.005202666752868229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,128,0.014111111561457316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,128,0.0028479999552170434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,512,0.0052897776994440295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,128,0.0053582220441765254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,64,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,64,0.005636444522274866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,768,32,0.002569777684079276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,256,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,768,32,0.0055440002017551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,65536,0.07756444480684069
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,768,128,0.00519822206762102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,65536,0.029314666986465454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,65536,0.05591377947065565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,16384,0.01037599974208408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,16384,0.027785778045654297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,16384,0.016178665889634024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,65536,0.06249599986606174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,12288,0.023678221636348303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,12288,0.008953777452309927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,12288,0.013855111267831592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,16384,0.016536888149049547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,10240,0.02125777800877889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,10240,0.00794222205877304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,10240,0.012609778179062737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,12288,0.013302221894264221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,8192,0.019293333093325298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,8192,0.008336889247099558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,8192,0.01068800025516086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,10240,0.012033777932325998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,7168,0.008284444610277811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,7168,0.018350222044520907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,7168,0.0103502223889033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,6144,0.007304888632562425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,8192,0.010665777656767102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,6144,0.017532444662517972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,6144,0.009720000127951304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,7168,0.009480000370078618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,5120,0.016547555724779766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,5120,0.0069679998689227635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,5120,0.008736888567606607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,6144,0.009036444127559662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,4096,0.01516355574131012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,4096,0.007611555357774098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,4096,0.007714666426181793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,5120,0.008435555630260045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,3584,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,3584,0.015962666935390897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,3584,0.007920000288221572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,4096,0.00795022232664956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,3072,0.014771555860837301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,3072,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,3072,0.007608888877762689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,3584,0.007322666545708974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,2560,0.014114666316244336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,2560,0.006575999988449945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,2560,0.007251555720965068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,3072,0.007263110743628607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,2048,0.013753777576817406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,2048,0.006248000181383557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,2048,0.006625777731339137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,2560,0.006636444479227066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,1536,0.013431111143694984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,1536,0.004249777644872665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,1536,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,2048,0.006512889017661412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,1024,0.01274400038851632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,1024,0.0038542221817705366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,1024,0.006127999888526068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,1536,0.0058782220714622075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,768,0.01482488877243466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,768,0.003519999898142285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,768,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,1024,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,512,0.0031795555518733132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,512,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,768,0.005437333136796951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,256,0.0032311110860771606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,256,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,512,0.005405333307054307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,128,0.01276177747382058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,128,0.002879111096262932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,256,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,128,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,64,0.002604444407754474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,64,0.005557333429654439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,512,32,0.0025475554996066624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,512,128,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,512,32,0.005631111148330901
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,65536,0.015139554937680563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,65536,0.054823110500971474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,65536,0.05976533227496677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,1024,10240,0.022184888521830242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,16384,0.026774222652117412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,16384,0.009029332962301042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,16384,0.016139555308553908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,12288,0.02336799932850732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,16384,0.016524444023768108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,12288,0.008289777570300633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,12288,0.013560888667901358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,10240,0.020941333638297185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,12288,0.013526221944226159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,10240,0.007710222568776872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,10240,0.012366222010718452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,10240,0.012051555845472546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,8192,0.007252444823582967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,8192,0.010946667028797997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,8192,0.010690666735172272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,7168,0.017847999930381775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,7168,0.006656888872385025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,7168,0.01054044481780794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,7168,0.009758222434255812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,65536,0.0785084433025784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,6144,0.006555555595291986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,6144,0.009290667043791877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,6144,0.009015999734401703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,5120,0.015449777245521545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,5120,0.006212444355090459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,5120,0.008961777720186446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,5120,0.008339555727110969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,4096,0.014502222339312235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,4096,0.005903999838564131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,4096,0.007594666547245449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,4096,0.007596444752481248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,3584,0.014123555686738757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,3584,0.006521777974234686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,3584,0.00758133331934611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,10240,0.02165599995189243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,3072,0.013793778088357715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,3584,0.007270221908887227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,3072,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,3072,0.007287110719415877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,3072,0.0069644442862934535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,2560,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,2560,0.006834666762087081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,2560,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,2048,0.01275111072593265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,2048,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,768,5120,0.01663466625743442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,2048,0.006629333313968446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,2048,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,1536,0.00453955555955569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,1536,0.006227555374304454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,1536,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,1024,0.011749332977665795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,1024,0.0038968887594011095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,1024,0.005954666684071223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,1024,0.005671999934646819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,768,0.01311466677321328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,768,0.003572444534964032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,768,0.005961777849329843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,768,0.005280888742870755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,512,0.011065777805116443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,512,0.0032275555034478507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,512,0.0052586665583981415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,512,0.0052800000541739995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,256,0.011742221812407175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,256,0.0032071111102898917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,256,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,256,0.005247999810510212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,512,0.012079999678664737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,128,0.002798222212327851
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,128,0.005769777629110549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,256,128,0.0052533331844541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,64,0.0026462222966882917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,64,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,256,32,0.0025066667132907445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,512,256,0.013756444056828817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,256,32,0.00553955551650789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,65536,0.014481777946154276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,65536,0.07389422257741292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,65536,0.051199999120500356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,16384,0.02677244444688161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,65536,0.05306577682495117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,16384,0.00795199970404307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,16384,0.016122667325867545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,16384,0.01648355523745219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,12288,0.022615111536449854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,12288,0.007643555601437886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,12288,0.013419555293189155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,10240,0.019888889458444383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,12288,0.013758222262064615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,10240,0.006980444822046492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,10240,0.01240355521440506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,8192,0.01854933301607768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,10240,0.011697777443461947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,65536,0.07747466696633233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,8192,0.006966222491529252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,8192,0.01071911139620675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,7168,0.017522666189405654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,7168,0.006568888823191325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,8192,0.010716444088353051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,7168,0.010313777459992303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,6144,0.006274666637182236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,7168,0.009364444348547194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,6144,0.016892444756295946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,6144,0.009681777821646797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,6144,0.008963555925422246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,5120,0.015498666299713982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,5120,0.006443555570311016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,5120,0.00867377801073922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,5120,0.008399999803966945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,4096,0.01481599940194024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,8192,0.01886755559179518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,4096,0.00565155554148886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,4096,0.008031999899281396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,4096,0.00793866647614373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,3584,0.014208000567224292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,3584,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,3584,0.007647111184067196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,3584,0.0074622225430276655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,3072,0.007341333561473423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,6144,0.01716444392999013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,3072,0.013815110756291283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,3072,0.005616000129116907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,3072,0.007304888632562425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,2560,0.013770666387346057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,2560,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,2560,0.0069751110341813825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,2560,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,2048,0.01310755560795466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,2048,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,2048,0.006647111227114995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,1536,0.006291555447710886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,2048,0.006585777633719974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,1536,0.012067555553383298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,1536,0.004515555583768421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,1536,0.006189333481921091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,1024,0.012448888685968188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,1024,0.0036159998012913596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,1024,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,1024,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,768,0.013403555585278405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,768,0.003543111185232798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,512,0.0031884445084465873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,768,0.005906666732496685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,768,0.005423999908897612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,512,0.011104889214038849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,256,0.002879111096262932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,512,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,512,0.00554666668176651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,256,0.011866666376590729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,256,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,256,0.005254222287072076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,128,128,0.011755555868148804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,128,0.0028151110228565005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,32,128,128,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,128,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,64,0.002479111154874166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,64,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,128,32,0.0028355556229750314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,128,32,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,65536,0.04889600144492256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,65536,0.011345778074529437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,16384,0.007320889168315464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,16384,0.01613333324591319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,12288,0.006912888752089606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,12288,0.013738666971524557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,10240,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,10240,0.012362666428089142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,8192,0.006649777707126405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,8192,0.011024889018800525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,7168,0.006294222341643439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,7168,0.01035377797153261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,6144,0.006256000035338932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,6144,0.00941955546538035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,5120,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,5120,0.009304000271691216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,4096,0.0058853332367208265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,4096,0.007998222278224098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,3584,0.006280000011126201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,3584,0.007658667034573025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,3072,0.005959110955397288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,3072,0.007344000041484833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,2560,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,2560,0.013063111239009433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,1536,0.005912000106440649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,2560,0.007017778025733099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,2048,0.005938666562239329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,768,0.003576888806290097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,2048,0.006574222197135289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,1536,0.00453955555955569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,1024,0.0038453332251972626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,1024,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,768,0.00589155571328269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,512,0.003212444484233856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,512,0.0059164443777667145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,256,0.002846222163902389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,256,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,128,0.0025519999778932994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,128,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,64,0.002510222295920054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,64,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,64,32,0.0024746666765875286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,64,32,0.005199111170238919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,65536,0.008975110948085785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,65536,0.047576887740029224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,16384,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,16384,0.016137777103318107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,12288,0.006250666661394968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,12288,0.013059555656380124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,10240,0.006245333287451003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,10240,0.012097777591811286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,8192,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,8192,0.010690666735172272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,7168,0.00629066675901413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,7168,0.00998133338159985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,6144,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,4096,0.007602666815121968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,1536,0.012094222009181976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,6144,0.009672000176376766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,5120,0.006090666684839461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,5120,0.009213333328564962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,4096,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,3584,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,3584,0.007306666837798224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,3072,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,3072,0.007290666302045186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,2560,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,2560,0.0070017774899800616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,2048,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,768,0.0035084444615576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,2048,0.006562666760550604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,1536,0.0042444442709287005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,512,0.0052800000541739995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,1536,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,1024,0.0038364442686239877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,1024,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,768,0.005982222242487802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,512,0.003152888889114062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,256,0.00281866660548581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,256,0.00527555536892679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,32,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,128,0.002536000063021978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,128,0.005569777968857024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,64,0.0026453334010309647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,32,32,64,0.00555288874440723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,32,32,32,0.002522666628162066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,16384,0.32392801178826225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,12288,0.2269199954138862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,16384,0.17320533593495688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,12288,0.1861413319905599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,16384,0.23226843939887154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,12288,0.12816088729434544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,10240,0.18902577294243705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,10240,0.10911822319030762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,8192,0.15520889229244655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,8192,0.14555466175079346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,8192,0.09302044577068752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,32,256,128,0.011354666617181567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,7168,0.13387821780310735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,7168,0.08097777764002483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,6144,0.12813066111670599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,6144,0.11554133892059326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,6144,0.073161780834198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,10240,0.16039021809895834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,5120,0.11441422833336724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,5120,0.09744355413648818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,5120,0.060912887255350746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,7168,0.14550222290886775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,4096,0.11016533109876846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,4096,0.08132177591323853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,4096,0.048278222481409706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,16384,0.13862044281429714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,12288,0.10488355822033352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,3584,0.0694426695505778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,10240,0.09582133425606622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,3584,0.09683555364608765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,3584,0.04368977745374044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,3072,0.06021333403057522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,3072,0.08939022488064235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,3072,0.038145777251985334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,2560,0.08301155434714423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,2560,0.050981332858403526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,8192,0.07235733005735609
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,2560,0.03276355399025811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,2048,0.07637244462966919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,2048,0.04194044404559665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,2048,0.028380443652470905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,7168,0.06406578090455797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,6144,0.05537155601713392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,1536,0.032702220810784235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,1536,0.022666666242811415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,5120,0.04710666669739617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,1536,0.06931200292375353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,1024,0.02355466617478265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,1024,0.018567111757066514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,4096,0.03923022084765964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,3584,0.03497155507405599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,768,0.018573333819707234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,768,0.01676177812947167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,3072,0.030623998906877305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,512,0.05627466572655571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,512,0.013451555536852943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,512,0.01498133275243971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,2560,0.0274008893304401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,256,0.024508444799317256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,256,0.008360889222886827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,256,0.01350400017367469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,2048,0.027100443840026855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,128,0.01960266629854838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,128,0.006578666468461354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,128,0.01034400032626258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,1536,0.019197333190176222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,64,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,64,0.01038933296998342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,65536,32,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,65536,32,0.01038577738735411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,65536,0.22556622823079428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,65536,0.32160888777838814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,65536,0.17858400609758165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,1024,0.06266399886873034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,65536,768,0.031022220849990845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,16384,0.07238488727145724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,16384,0.09222666422526042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,16384,0.04956355690956116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,12288,0.05894311269124349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,12288,0.0714977780977885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,12288,0.03869511021508111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,65536,0.16178754965464273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,10240,0.05224177903599209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,10240,0.05435466766357422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,10240,0.03408799899948968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,16384,0.04383822282155355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,8192,0.0475466681851281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,8192,0.044490665197372437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,8192,0.028656889994939167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,12288,0.03555555476082696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,7168,0.052433778842290245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,7168,0.03914666506979201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,7168,0.02697244452105628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,1024,0.015810666812790763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,6144,0.04822044571240743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,6144,0.03409511182043288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,6144,0.023966223001480103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,10240,0.03192889028125339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,5120,0.04420622189839681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,5120,0.029158221350775823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,5120,0.022093334131770666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,8192,0.024884444144037034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,4096,0.04257511099179586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,4096,0.02459466622935401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,4096,0.01793599956565433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,7168,0.02219999995496538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,768,0.013454222016864352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,3584,0.03746133380466037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,3584,0.021016889148288306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,3584,0.016173332929611206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,3072,0.03523644473817613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,3072,0.018975110517607797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,3072,0.014896000425020853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,6144,0.020042666130595736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,2560,0.03278933299912347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,2560,0.01608533329433865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,2560,0.013764444324705334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,256,0.01038577738735411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,2048,0.03121244576242235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,512,0.012703110774358114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,2048,0.013184888495339287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,2048,0.012300444145997366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,1536,0.028985778490702312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,1536,0.010648888846238455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,1536,0.01072533345884747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,65536,128,0.010255110760529837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,1024,0.026366222235891554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,1024,0.008277333445019191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,1024,0.008914666871229807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,5120,0.017650667164060805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,768,0.019164444671736825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,768,0.007284444239404466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,768,0.0074942219588491656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,4096,0.015449777245521545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,512,0.021996445126003687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,512,0.004880889008442561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,512,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,3584,0.014642667439248828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,256,0.01717066599263085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,256,0.00351555562681622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,256,0.006267555471923616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,3072,0.013247999880048962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,16384,128,0.015159999330838522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,128,0.003173333282272021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,128,0.006266666783226862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,2560,0.01237155579858356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,64,0.002867555452717675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,64,0.005919999960396025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,16384,32,0.0028764444092909494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,16384,32,0.006550222221348021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,2048,0.012764444781674279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,1536,0.009006222089131674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,65536,0.2552231152852376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,1024,0.0075911109646161394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,16384,0.05827199750476413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,16384,0.06713066498438518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,65536,0.15391822655995688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,16384,0.043508443567487926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,12288,0.04776444368892246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,12288,0.052941332260767617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,12288,0.034098668230904475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,768,0.0069617778062820435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,10240,0.04269511169857449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,10240,0.045103109545177884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,10240,0.029829333225886028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,512,0.0069466663731469055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,8192,0.03939111034075419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,8192,0.03795999950832791
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,256,0.006575999988449945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,8192,0.025757332642873127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,7168,0.04584088921546936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,7168,0.0329075555006663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,16384,128,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,7168,0.02443555494149526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,65536,0.14735021856096056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,6144,0.04289244280921089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,6144,0.029401779174804688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,6144,0.02198933396074507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,5120,0.03908088803291321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,16384,0.03945066531499227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,5120,0.025901334153281316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,5120,0.019223110543357003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,4096,0.034930666287740074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,4096,0.022255111071798537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,4096,0.01683644453684489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,12288,0.03532355692651536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,3584,0.032906668053732976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,3584,0.019816888703240287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,3584,0.015252444479200574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,10240,0.027613333529896204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,3072,0.030858665704727173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,3072,0.01774133410718706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,3072,0.013842666314707862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,8192,0.022865777214368183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,2560,0.029684444268544514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,2560,0.012308444413873883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,2560,0.01274311128589842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,7168,0.020262221495310467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,2048,0.027425777581002977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,2048,0.010647111468844943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,2048,0.011386666860845355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,6144,0.018222222725550335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,5120,0.01647733317481147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,4096,0.014516444669829475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,1536,0.009452444811662039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,1536,0.010129777921570672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,3584,0.013448889056841532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,1024,0.02295822236273024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,1024,0.00794400026400884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,1024,0.007656888829337225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,3072,0.012595555848545499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,768,0.006634666687912411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,768,0.006747555401590135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,2560,0.011581333147154914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,65536,0.16683644718594023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,512,0.004596444467703502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,512,0.006262222097979651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,2048,0.011562666959232755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,256,0.01752444439464145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,256,0.003532444437344869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,256,0.006299555715587404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,1536,0.008287111090289222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,1024,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,128,0.0031679999083280563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,128,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,768,0.006641777853171031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,64,0.0031555555760860443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,64,0.006257777826653586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,12288,32,0.003206222214632564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,12288,32,0.006083555519580841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,512,0.006559111177921295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,65536,0.1471573379304674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,256,0.006304889089531369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,65536,0.21972533067067465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,16384,0.05253244439760844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,65536,0.14096889230940077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,16384,0.06048088603549533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,16384,0.041301333241992526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,12288,128,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,12288,0.04702755477693346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,12288,0.03255022234386868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,65536,0.14083555009629992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,10240,0.039093332158194646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,10240,0.03897066579924689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,16384,0.03801244497299194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,10240,0.028463111983405218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,8192,0.03639288743336996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,1536,0.025864000121752422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,8192,0.03264177838961283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,8192,0.024337778488794964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,7168,0.04474044508404202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,7168,0.029314666986465454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,7168,0.021908443835046556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,12288,0.0331911113527086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,6144,0.040656887822681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,6144,0.025983111725913152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,6144,0.019866666860050626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,10240,0.02674666709370083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,768,0.018571555614471436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,5120,0.036764442920684814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,512,0.02023555503951179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,5120,0.022706665926509436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,5120,0.017287110288937885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,4096,0.019704888264338177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,4096,0.03301599952909682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,8192,0.022207111120224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,4096,0.014459555347760519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,3584,0.017662222186724346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,3584,0.013797333670987023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,7168,0.019552888141738046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,3072,0.030016889174779255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,3072,0.01590222285853492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,3072,0.012730666332774691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,12288,128,0.015487111277050443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,6144,0.017518222332000732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,2560,0.028472890456517536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,2560,0.01239377756913503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,2560,0.012423999607563019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,2048,0.02665244374010298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,2048,0.010991999672518836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,2048,0.01090844472249349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,5120,0.015944000747468736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,1536,0.024000888069470722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,1536,0.009200000100665623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,4096,0.014099554883109199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,1536,0.00852177788813909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,1024,0.021280889709790547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,1024,0.0075333333677715724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,3584,0.013058666553762225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,1024,0.007269333634111617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,768,0.01868888901339637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,768,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,768,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,3072,0.012048000262843238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,512,0.018550222118695576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,512,0.004874666531880696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,512,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,2560,0.010712888505723743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,256,0.01750666730933719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,256,0.0035128887328836652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,256,0.006221333311663733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,2048,0.008936888641781276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,12288,0.04386933313475715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,128,0.0148053334818946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,128,0.0033377777371141645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,128,0.005932444499598608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,1536,0.007713778151406183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,64,0.0032017777363459268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,64,0.005950222412745158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,10240,32,0.0031617778456873367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,1024,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,10240,32,0.006292444550328785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,768,0.006604444649484422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,65536,0.17486133840348986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,65536,0.13270310560862222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,16384,0.047296888298458524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,16384,0.04627644353442722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,512,0.0060159998635451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,65536,0.09415022532145183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,16384,0.030346665117475722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,12288,0.03972533345222473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,12288,0.03298577666282654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,12288,0.02389066749148899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,10240,0.027549333042568628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,10240,0.03542933199140761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,256,0.006105777704053455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,10240,0.021520889467663232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,8192,0.024018666810459558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,8192,0.032784889141718544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,8192,0.01923466722170512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,7168,0.040214220682779946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,7168,0.020617778102556866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,7168,0.017525333497259352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,10240,128,0.00564444437623024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,6144,0.0365973346763187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,6144,0.018271999226676095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,6144,0.016167999969588388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,65536,0.09193511141671075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,5120,0.03455911080042521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,16384,0.02549955579969618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,5120,0.01579466627703773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,5120,0.014460444450378418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,12288,0.020703110429975722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,4096,0.031098667118284438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,4096,0.01351111133893331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,4096,0.012739555703269111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,10240,0.019550222489568923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,3584,0.011986666255527072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,3584,0.011703111231327057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,8192,0.015805333852767944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,3072,0.028397334946526423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,3072,0.010999110837777456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,3072,0.011050666371981302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,7168,0.01514488955338796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,2560,0.02608977754910787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,2560,0.009654222263230218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,2560,0.010244444840484196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,6144,0.013794666363133324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,2048,0.02380888826317257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,2048,0.008458666503429413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,2048,0.008996444443861643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,10240,3584,0.031896889209747314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,5120,0.012161778079138862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,1536,0.007246221933099959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,1536,0.007272889216740926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,4096,0.010909332997269101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,1024,0.0192266669538286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,1024,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,1024,0.00629066675901413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,3584,0.00982488857375251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,768,0.018426666657129925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,768,0.004449777719047335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,768,0.005946666830115848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,3072,0.008783111141787635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,512,0.017839999662505258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,512,0.0035048888789282907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,512,0.0058808889653947614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,2560,0.0076657773719893555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,256,0.016252444850073922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,256,0.0032551110618644287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,256,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,2048,0.007112888826264276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,128,0.014765333798196582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,128,0.002867555452717675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,128,0.005584888988071018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,1536,0.0069022224181228215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,64,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,64,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,8192,32,0.002830222249031067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,1024,0.005961777849329843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,8192,32,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,65536,0.1506835619608561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,768,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,512,0.005943999936183293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,65536,0.0911911129951477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,16384,0.04219111137919956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,16384,0.045534223318099976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,16384,0.02765422231621212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,256,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,12288,0.03774311145146688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,12288,0.03338399860594008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,12288,0.02247911029391819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,8192,128,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,10240,0.03363022208213806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,10240,0.028565333949195013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,10240,0.020046222541067336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,65536,0.07492978043026395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,8192,0.03105688757366604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,8192,0.021969777014520433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,8192,0.01776266594727834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,16384,0.023546665906906128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,7168,0.04072177741262648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,7168,0.018956444329685636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,7168,0.01646222174167633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,12288,0.01926933394538032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,6144,0.03750933210055033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,6144,0.01667555504375034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,6144,0.015247111519177755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,5120,0.03459111187193129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,5120,0.014455111490355598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,5120,0.013800889253616333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,10240,0.016896888613700867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,4096,0.029976887835396662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,4096,0.012895111408498553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,4096,0.011714666253990598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,8192,0.015024888846609326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,3584,0.02918755677011278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,3584,0.011023999916182624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,3584,0.011353777514563667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,7168,0.01313066648112403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,3072,0.02716444432735443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,3584,0.029414223300086126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,3072,0.010077333284748925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,6144,0.01241422196229299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,3072,0.01035288886891471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,2560,0.025762667258580525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,2560,0.008966222405433655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,2560,0.009668444593747457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,5120,0.011223110887739392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,2048,0.00796088907453749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,2048,0.00792266676823298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,4096,0.009962666365835402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,1536,0.019935111204783123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,1536,0.007278222176763747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,1536,0.006610666712125142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,3584,0.00903200027015474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,3072,0.008059555457697975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,1024,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,1024,0.006253333141406377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,2560,0.007237333390447829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,768,0.017872888180944655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,768,0.0041786668201287585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,768,0.0059368887709246735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,2048,0.0069431112044387394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,8192,1536,0.02162488963868883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,512,0.016857778032620747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,512,0.003546666767862108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,512,0.005884444548024072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,256,0.016819554898473952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,256,0.0032391111469931076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,256,0.005555555638339784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,1536,0.006292444550328785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,128,0.014446222119861178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,128,0.0031742221779293488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,128,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,1024,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,64,0.0028568889117903183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,64,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,7168,32,0.0029031110720502008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,7168,32,0.005277333160241445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,768,0.005719999886221356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,512,0.005619555711746216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,65536,0.12314400408003066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,65536,0.11951377656724717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,256,0.00554577757914861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,65536,0.08521511157353719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,16384,0.03607733382119073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,16384,0.04268355502022637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,16384,0.026293334033754136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,12288,0.02885599931081136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,12288,0.035637332333458796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,12288,0.020994666549894545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,65536,0.1264515585369534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,10240,0.03233333428700765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,7168,128,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,10240,0.024805333879258897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,10240,0.01924444403913286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,8192,0.02983199887805515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,8192,0.021839999490314062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,8192,0.01667288939158122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,7168,0.039918220705456205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,7168,0.019225777851210702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,7168,0.015169777803950839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,65536,0.07188710901472303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,6144,0.035916444328096174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,16384,0.022987556126382615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,6144,0.017469333277808297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,6144,0.014290667242474027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,12288,0.018567111757066514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,5120,0.03272799981964959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,5120,0.015467555986510383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,5120,0.012832000023788877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,10240,0.016181333197487723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,4096,0.028969777954949275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,4096,0.012064889073371887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,8192,0.014471999473041959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,4096,0.011223999990357293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,3584,0.027760000692473516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,3584,0.010346666806273991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,3584,0.010696888797812991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,7168,0.01278133359220293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,3072,0.026479111777411565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,3072,0.009371555513805812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,3072,0.00996977753109402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,2560,0.02403555479314592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,6144,0.0118568887313207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,2560,0.00867555538813273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,2560,0.00848355558183458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,2048,0.02163377735349867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,2048,0.007285333342022366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,5120,0.010673777924643623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,2048,0.007037333316273159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,1536,0.019719110594855416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,1536,0.006562666760550604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,1536,0.006599111275540457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,4096,0.009353777600659264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,1024,0.017901332841979135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,1024,0.004588444613748127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,1024,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,3584,0.008007110820876228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,768,0.017877333694034152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,768,0.004187555362780889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,768,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,3072,0.007299555672539606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,512,0.01629955569903056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,512,0.0035297779573334586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,512,0.005776888794369168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,2048,0.0233297778500451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,2560,0.007110222346252865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,256,0.01681866745154063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,256,0.003170666595300039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,256,0.005619555711746216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,6144,128,0.014820444915029736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,2048,0.006673777682913675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,128,0.003192000091075897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,128,0.0053582220441765254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,64,0.0028320000403457214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,64,0.0052639999323421055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,1536,0.006257777826653586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,6144,32,0.002854222224818336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,6144,32,0.0052693333062860705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,65536,0.10889155334896511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,1024,0.005955555372767978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,65536,0.07748622364468045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,768,0.00574311117331187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,16384,0.03250755535231696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,16384,0.040799998574786715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,16384,0.02402577797571818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,512,0.005571555760171678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,12288,0.026384888423813715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,12288,0.03432800041304694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,12288,0.019659555620617337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,10240,0.03087377879354689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,7168,1024,0.01921866668595208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,256,0.0052800000541739995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,10240,0.022351110974947613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,10240,0.01796444422668881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,8192,0.019832889238993328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,8192,0.0285537772708469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,6144,128,0.0055440002017551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,8192,0.015856888559129503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,7168,0.017550221747822232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,7168,0.01443822185198466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,65536,0.07004444466696845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,6144,0.03211022085613675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,6144,0.015574221809705099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,6144,0.01369955556260215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,16384,0.02196888956758711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,5120,0.03212533394495646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,5120,0.013767110804716745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,12288,0.01796711153454251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,5120,0.012164444559150271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,4096,0.029158221350775823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,4096,0.011040888726711273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,10240,0.01571911076704661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,4096,0.010568000376224518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,8192,0.013754666679435305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,3584,0.00963644435008367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,3584,0.00977244476477305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,7168,0.012560000022252401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,3072,0.02334755493534936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,3072,0.008622222476535374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,3072,0.008348444269763099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,6144,0.01128088848458396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,2560,0.021583999196688335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,2560,0.007992888490358988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,2560,0.007327111230956183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,5120,0.009692444569534725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,2048,0.020284444093704224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,2048,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,2048,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,4096,0.008328888979223039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,3584,0.007643555601437886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,1536,0.018605333235528734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,1536,0.005966222120655908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,1536,0.0062871111763848206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,3072,0.0076142218377855085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,1024,0.01683644453684489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,1024,0.004534222185611725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,1024,0.0059262220230367445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,2560,0.007298666569921706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,768,0.017539555827776592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,768,0.003907555507289039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,768,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,2048,0.006653333289755716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,512,0.01549600064754486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,512,0.003640000190999773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,512,0.005575110928879843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,1536,0.006248000181383557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,256,0.014819555812411837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,256,0.003208000005947219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,256,0.0053084443012873335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,128,0.014462222655614218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,128,0.0028560000161329904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,128,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,1024,0.005947555518812603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,64,0.0028035555862718155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,64,0.0052737775776121355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,5120,32,0.0028106667515304354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,5120,32,0.0052684446175893145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,768,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,65536,0.09127822187211777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,65536,0.08641333050198025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,512,0.005571555760171678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,65536,0.0702622201707628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,16384,0.03449777762095133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,16384,0.02608000073168013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,16384,0.022985777921146814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,256,0.005211555709441503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,12288,0.02951999836497837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,12288,0.020958221620983545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,12288,0.018567999203999836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,10240,0.026995556222067937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,5120,128,0.005234666582610872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,10240,0.018213333355055917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,10240,0.017123555143674214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,8192,0.025232889586024817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,8192,0.015846222639083862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,8192,0.015119110544522604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,65536,0.06764266888300578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,7168,0.036107556687461004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,7168,0.01443911095460256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,16384,0.021425777011447485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,7168,0.013799111048380533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,6144,0.033026665449142456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,6144,0.013391111459996967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,12288,0.017197334104114108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,6144,0.012774222426944308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,5120,0.03029511041111416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,5120,0.012050666742854647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,10240,0.015228443675571017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,5120,0.011721777419249216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,4096,0.02849510974354214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,4096,0.012617777619096967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,4096,0.010029333333174387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,8192,0.013435555828942193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,3584,0.02436622149414486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,3584,0.011899555722872416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,3584,0.008339555727110969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,7168,0.011721777419249216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,3072,0.023379556006855432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,3072,0.010688888529936472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,6144,0.010001777774757808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,3072,0.007656888829337225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,2560,0.021750221649805706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,65536,0.11529333061642116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,2560,0.007251555720965068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,2560,0.007289778027269576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,5120,0.00869688888390859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,2048,0.019889778561062284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,2048,0.005991999887757831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,2048,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,1536,0.018636445204416912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,4096,0.008044444852405125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,1536,0.004690666579537922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,1536,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,1024,0.017145777742067974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,1024,0.0038862224254343244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,3584,0.007660444411966536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,1024,0.005921777751710679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,768,0.017881777551439073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,768,0.0038542221817705366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,3072,0.007256888680987888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,768,0.0059573331640826324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,512,0.01516355574131012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,512,0.003552000141806073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,512,0.005246222019195557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,2560,0.0069351109365622205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,7168,0.035411556561787925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,256,0.015440889530711703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,256,0.003152888889114062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,256,0.005573333137565189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,2048,0.006628444625271692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,4096,128,0.014149333039919535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,128,0.0028808888875775864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,128,0.0053031109273433685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,1536,0.006228444476922353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,64,0.0025048889219760895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,64,0.005380444642570283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,1024,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,4096,32,0.002789333255754577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,4096,32,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,768,0.005551110953092575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,65536,0.07993421951929729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,65536,0.08985777695973714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,65536,0.0684551133049859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,512,0.005567111074924469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,5120,3584,0.027099554737408955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,16384,0.023698666029506262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,16384,0.021758221917682227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,256,0.0052933332820733385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,12288,0.02882844540807936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,12288,0.019335110982259113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,12288,0.018202667435010273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,10240,0.01695999999841054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,10240,0.026592888765864905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,4096,128,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,10240,0.016593777471118502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,8192,0.024658666716681585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,8192,0.014717333846622042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,8192,0.014680888917711047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,7168,0.035487112071779035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,7168,0.01352355546421475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,7168,0.013398222625255585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,65536,0.06654044654634264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,6144,0.03339111142688327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,6144,0.012723555167516073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,6144,0.012721777790122561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,5120,0.030159112479951646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,16384,0.020963556236690942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,5120,0.011719110939237805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,5120,0.011439111497667102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,12288,0.01700177788734436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,4096,0.011529777612951068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,4096,0.0262435558769438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,4096,0.008738666772842407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,10240,0.015059555570284525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,3584,0.011049778097205691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,3584,0.0239991115199195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,3584,0.00833155545923445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,8192,0.01274311128589842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,3072,0.022642667094866436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,3072,0.00942311104800966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,7168,0.011023999916182624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,3072,0.007634667058785756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,2560,0.009026666482289633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,6144,0.009400000174840292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,2560,0.0069848886794514125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,2048,0.019948444432682462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,2048,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,5120,0.008678221868144141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,2048,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,1536,0.018599111172888014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,1536,0.004563555535342959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,4096,0.007947555846638149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,1536,0.006251555350091722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,3584,0.007611555357774098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,1024,0.016912889149453905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,1024,0.00388711111413108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,1024,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,3072,0.007378666765160031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,768,0.01790844400723775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,768,0.0038497779104444715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,768,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,512,0.01500266624821557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,2560,0.0069671107663048645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,512,0.0034435554924938413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,512,0.005585777676767773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,256,0.014817777607176038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,2048,0.006592000110281839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,256,0.0031475555151700974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,256,0.00554666668176651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,128,0.014944000376595391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,128,0.0028773333049482773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,1536,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,128,0.005561777700980504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,64,0.0028444443725877335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,64,0.005218666460778978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,1024,0.005616000129116907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3584,32,0.0028133332315418455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3584,32,0.005571555760171678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,768,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,65536,0.06348088714811537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,65536,0.08811733457777236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,512,0.005600000007285013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,65536,0.06607377529144287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,16384,0.03310399916436937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,16384,0.02083911167250739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,16384,0.02106133268939124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,256,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,12288,0.028706666496064927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,12288,0.017630222770902846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,12288,0.017873777283562552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3584,128,0.005195555587609609
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,10240,0.015312888556056552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,10240,0.016169778174824186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,65536,0.06548888815773858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,8192,0.024002666274706524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,16384,0.020797333783573575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,8192,0.013454222016864352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,8192,0.014357333381970724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,7168,0.032292445500691734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,12288,0.016883555385801528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,7168,0.012687111066447364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,7168,0.01293244461218516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,6144,0.03130133284462823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,6144,0.011746666497654386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,10240,0.014535110857751636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,6144,0.012096000214417776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,5120,0.028189334604475234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,5120,0.010737777584128909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,5120,0.010441777606805166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,8192,0.012048888537618848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,4096,0.025080889463424683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,4096,0.009866666462686326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,7168,0.010016889207892949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,4096,0.008731555607583787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,3584,0.02330222229162852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,3584,0.01000088867213991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,3584,0.007657777931955125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,6144,0.009288888838556077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,3072,0.021780444516075984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,3072,0.008974221845467886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,3072,0.007355555891990662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,5120,0.008658666577604082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,2560,0.020447111792034574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,2560,0.00629688882165485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,2560,0.006981333096822103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,2048,0.019270223047998216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,4096,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,2048,0.005311111195219888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,2048,0.006568000134494569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,1536,0.01720177796151903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,3584,0.007596444752481248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,1536,0.004551110996140374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,1536,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,1024,0.016145777371194627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,1024,0.003916444463862313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,3072,0.007260444263617198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,1024,0.005929777605666055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,2560,0.007032000356250339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,768,0.0036720000207424164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,768,0.005720888988839255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,2048,0.006597333484225803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,512,0.014132445057233175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,512,0.003253333270549774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,1536,0.006248888870080312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,512,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,1024,0.005671999934646819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,256,0.00315022220214208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,256,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,768,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,128,0.014771555860837301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,128,0.0028773333049482773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,128,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,512,0.005559999909665849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,64,0.00253866674999396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,64,0.00554933316177792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,3072,32,0.002524444419476721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,3072,32,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,256,0.005203555441564984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,65536,0.08616177903281318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,65536,0.05855821900897556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,65536,0.06372088856167264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,16384,0.03353244397375319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,16384,0.03228888909022013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,16384,0.01919644408755832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,3072,128,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,16384,0.020560888780487906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,12288,0.01613066593805949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,12288,0.02815199891726176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,12288,0.017493334081437852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,10240,0.02530755599339803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,10240,0.01385155568520228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,10240,0.015740444262822468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,65536,0.06451377603742811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,8192,0.02342044479317135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,8192,0.01236977759334776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,16384,0.02055111030737559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,8192,0.013777777552604675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,7168,0.03357955482270982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,12288,0.016535111599498324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,7168,0.011686222420798408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,7168,0.012837332983811697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,6144,0.02783999840418498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,6144,0.010716444088353051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,10240,0.013751111096805997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,6144,0.011328000161382886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,5120,0.024031110935741003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,5120,0.010000000397364298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,8192,0.011008888483047485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,5120,0.009703111317422655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,7168,0.010024000373151567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3584,2560,0.021386666430367365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,4096,0.023904000719388325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,4096,0.009293333523803288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,4096,0.007992888490358988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,3584,0.022287110487620037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,6144,0.009311999711725447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,3584,0.008994667066468133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,3584,0.007874666816658443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,3072,0.019919999771647982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,3072,0.007982222570313348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,5120,0.00869066682126787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,3072,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,2560,0.0188746667570538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,2560,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,2560,0.007263999846246507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,4096,0.00796088907453749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,2048,0.01757600075668759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,2048,0.005249777601824866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,2048,0.006583111153708563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,3584,0.007485333416197035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,1536,0.016807110773192514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,3072,0.0075786668393347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,1536,0.004552888787455029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,1536,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,1024,0.015284443895022074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,1024,0.003900444342030419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,2560,0.006983111302057902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,1024,0.00599911105301645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,768,0.016922665966881644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,768,0.0036755556033717263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,768,0.00556355549229516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,2048,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,512,0.014107555150985718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,512,0.0035146665241983202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,512,0.005561777700980504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,1536,0.005998222364319696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,256,0.014506666196717156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,256,0.00315022220214208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,1024,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,256,0.005229333208666907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2560,128,0.014077333940400017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,128,0.0028133332315418455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,128,0.005568888866239124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,768,0.005640888793600931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,64,0.002524444419476721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,64,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2560,32,0.0027875554644399216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2560,32,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,65536,0.050808889998330005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,512,0.005433777968088786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,65536,0.0795253316561381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,16384,0.030841777722040813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,65536,0.062041779359181724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,256,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,16384,0.01588800052801768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,16384,0.020281778441535104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,12288,0.013196444345845116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,12288,0.026916444301605225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2560,128,0.005196444276306364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,12288,0.016858667135238647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,10240,0.012156444291273752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,10240,0.02421333392461141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,10240,0.015480889214409722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,10240,0.02568888829814063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,65536,0.06396622127956815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,8192,0.010459555519951714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,8192,0.013056000073750814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,16384,0.02016799979739719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,7168,0.033839109871122576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,7168,0.010680889089902243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,7168,0.011687999798191918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,12288,0.015284443895022074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,6144,0.029784000582165186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,6144,0.012708444562223224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,6144,0.010207110808955299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,10240,0.01237155579858356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,5120,0.011633777783976661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,8192,0.011071999867757162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,5120,0.009000889129108852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,4096,0.023711111810472276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,7168,0.009884444375832876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,4096,0.010599999792046018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,4096,0.008236444658703273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,3584,0.023355555203225877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,6144,0.009348444640636444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,3584,0.009659555223253038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,3584,0.007697777615653143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,3072,0.021624000536070928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,3072,0.006616888774765863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,5120,0.008563555777072906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,3072,0.007267555428875818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,2560,0.020265777905782063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,2560,0.005615111026499007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,4096,0.007639110916190677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,2560,0.007297777467303806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,2048,0.01852888862291972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,3584,0.007685333490371704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,2048,0.005222222043408289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,2048,0.006640000061856375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,3072,0.007330666813585494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,1536,0.01648622254530589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,1536,0.004399111287461387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,2560,0.006741333338949416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,1536,0.006615110983451207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,1024,0.01514755520555708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,1024,0.0038515557017591265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,2048,0.006593777901596493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,1024,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,768,0.01723288827472263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,768,0.003528888854715559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,768,0.0059075554211934405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,768,0.017641777793566387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,1536,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,512,0.003148444410827425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,512,0.013582222163677216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,512,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,1024,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,256,0.0028213332924577924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,256,0.0053226666318045715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,128,0.014847111370828418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,768,0.005583999885453119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,128,0.0026559999419583213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,128,0.005261333452330695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,64,0.0025155554629034465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,512,0.005222222043408289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,64,0.005542222410440445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,2048,32,0.0025537777692079544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,2048,32,0.005306666509972678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,65536,0.03672444489267137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,65536,0.0783511135313246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,256,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,65536,0.05994666947258843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,16384,0.030406223403082952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,3072,256,0.0144533332851198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,16384,0.013447110851605734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,2048,128,0.005296000176005893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,16384,0.019640889432695176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,12288,0.026398221651713055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,12288,0.011615999870830111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,12288,0.01679111189312405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,10240,0.02297244469324748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,65536,0.06307022439108954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,10240,0.010372444159454769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,10240,0.01444355481200748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,16384,0.018588445252842374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,8192,0.020689777202076383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,8192,0.009657777845859528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,8192,0.011076444553004371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,7168,0.02836622132195367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,7168,0.01019200020366245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,12288,0.013721778160995908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,7168,0.010371555884679159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,6144,0.02717511190308465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,6144,0.008970666262838576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,10240,0.01180355581972334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,6144,0.009674666656388177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,5120,0.007981333467695449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,5120,0.026657778355810378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,8192,0.011003555523024665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,5120,0.008831111093362173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,4096,0.022594667143291895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,7168,0.009674666656388177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,4096,0.0069671107663048645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,4096,0.007954667011896769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,6144,0.009318222602208456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,3584,0.02162844439347585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,3584,0.00740444411834081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,3584,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,5120,0.008281778130266402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,3072,0.019892444213231403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,3072,0.006278222219811545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,3072,0.007300444775157505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,4096,0.007995555798212687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,2560,0.01889155473974016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,2560,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,2560,0.0069884442620807225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,2048,0.006262222097979651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,3584,0.007641777396202087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,2048,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,1536,0.016527111331621807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,3072,0.007306666837798224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,1536,0.004594666676388847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,1536,0.006248888870080312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,2560,0.006990222467316522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,1024,0.014767999450365702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,1024,0.0038337777886125776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,1024,0.006160000132189856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,2048,0.006600889066855113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,768,0.017105778058369953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,768,0.003536000019974179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,768,0.005761777775155173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,512,0.013439111411571503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,1536,0.006265777680608962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,512,0.0032239999208185407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,512,0.005630222045713001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,1024,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,256,0.014449778530332776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,256,0.0031786666562159858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,256,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,768,0.005616888817813661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,128,0.014287999934620328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,128,0.0028728888266616394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,512,0.005537777725193236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,128,0.005576888720194499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,64,0.002601777720782492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,64,0.0052435555391841466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1536,32,0.002524444419476721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,256,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1536,32,0.005587555468082428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,65536,0.026320889592170715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,65536,0.07695200045903523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,65536,0.058099554644690625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1536,128,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,16384,0.029875556627909344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,16384,0.01514311134815216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,16384,0.018774222996499803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,12288,0.02404266595840454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,65536,0.06277422110239665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,12288,0.012981333666377597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,12288,0.014117333624098035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,10240,0.021910222040282354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,16384,0.016530666086408827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,10240,0.011724444727102915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,10240,0.012708444562223224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,8192,0.019938665959570143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,12288,0.013449778159459433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,8192,0.010999999940395355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,8192,0.0107004443804423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,7168,0.0236826671494378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,10240,0.011776888536082374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,7168,0.00980088859796524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,7168,0.010266666611035665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,6144,0.02237155536810557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,8192,0.011072888970375061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,6144,0.008975110948085785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,6144,0.00963644435008367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,5120,0.022316444251272414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,7168,0.009705777797434065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,5120,0.008012444608741337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,5120,0.008990222381220924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,4096,0.007644444704055786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,4096,0.020597333709398907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,4096,0.00796977761718962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,6144,0.009123555488056606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,3584,0.018233777748213876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,3584,0.007487110793590546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,5120,0.008634666601816813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,3584,0.007638221813572778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,3072,0.0185208883550432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,4096,0.00792266676823298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,3072,0.006272000157170826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,3072,0.007317333585686154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,2560,0.016528889536857605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,3584,0.007321777443091075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,2560,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,2560,0.0070017774899800616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,2048,0.006963555183675554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,3072,0.007275555696752336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,2048,0.006616000086069107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,1536,0.014464888307783337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,2560,0.006631111105283101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,1536,0.005879999862776862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,1536,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,1024,0.013825777504179211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,2048,0.006587555425034628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,1024,0.004569777597983678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,1024,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,768,0.016135111451148987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,1536,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,5120,0.026369776990678575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,768,0.003915555361244413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,768,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,512,0.013071999781661563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,512,0.0038515557017591265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,1024,0.005629333357016246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,8192,0.022656889425383672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,512,0.005581333405441708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,256,0.0032017777363459268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,256,0.014149333039919535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,768,0.005568888866239124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,256,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,128,0.013748444616794586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,128,0.0028479999552170434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,512,0.0052168890833854675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,128,0.0052435555391841466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,64,0.0028080000645584534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,64,0.005404444618357553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,1024,32,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,256,0.005241777747869492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,1024,32,0.002834666727317704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,65536,0.021818665994538203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,1024,128,0.00508355556262864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,65536,0.05804355276955498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,16384,0.028145776854621038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,16384,0.010437332921557956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,16384,0.017133333616786532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,65536,0.06270222531424628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,12288,0.023698666029506262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,12288,0.011390222443474663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,12288,0.013410666750537025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,16384,0.01647555496957567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,10240,0.021595555875036452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,10240,0.009995555712117089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,10240,0.012048000262843238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,12288,0.013752000199423896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,8192,0.019904888338512845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,8192,0.009352000223265754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,8192,0.010713777608341642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,10240,0.012048000262843238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,7168,0.02125599980354309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,7168,0.008988444175985124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,7168,0.010062221851613786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,8192,0.010727110836240979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,6144,0.01990133358372582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,6144,0.008021333151393467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,6144,0.009487999810112847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,7168,0.009691555466916826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,5120,0.007610666255156199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,5120,0.008970666262838576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,2048,256,0.014462222655614218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,6144,0.009297777381208207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,4096,0.007616000043021307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,4096,0.007974222302436829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,5120,0.008357333640257517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,3584,0.016999999682108562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,3584,0.007298666569921706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,3584,0.007608888877762689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,4096,0.008005333443482717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,3072,0.016886222693655226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,3072,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,3072,0.007649777664078607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,3584,0.007463999920421177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,2560,0.015450666348139444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,2560,0.006581333362393909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,2560,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,3072,0.007316444483068254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,2048,0.0052800000541739995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,2048,0.015171556009186639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,2048,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,1536,0.014858666393491956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,2560,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,1536,0.0052586665583981415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,1536,0.006303999986913469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,2048,0.0063075555695427795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,1024,0.013755555782053204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,1024,0.0042462220622433555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,1024,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,1536,0.006262222097979651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,768,0.016144000821643405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,768,0.003916444463862313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,768,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,1024,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,512,0.0035502223504914176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,512,0.005421333428886201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,768,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,256,0.015472888946533203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,256,0.0031679999083280563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,256,0.0052595556610160405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,512,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,256,0.0052586665583981415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,128,0.0028640000770489373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,128,0.00526311124364535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,768,128,0.0052328887912962176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,64,0.00286666676402092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,64,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,768,32,0.002519999941190084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,768,32,0.005446222093370225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1536,2048,0.018238221605618794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,65536,0.017256889078352187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,65536,0.0759155551592509
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,65536,0.05712622404098511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,16384,0.026891556051042344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,16384,0.011119999819331698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,65536,0.06219377782609728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,16384,0.016063110695944894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,12288,0.023024888502226934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,12288,0.008753777378135258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,16384,0.016527111331621807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,12288,0.01313244468635983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,10240,0.02091466718249851
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,10240,0.007742221984598372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,12288,0.013528000149461957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,10240,0.012392889294359418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,8192,0.018935999936527677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,8192,0.008307555483447181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,10240,0.012094222009181976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,8192,0.01038933296998342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,7168,0.022338666849666174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,7168,0.007663999994595845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,8192,0.01071911139620675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,7168,0.010282666318946415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,6144,0.01993955506218804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,6144,0.007615110940403408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,7168,0.009741333623727163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,6144,0.009303111169073317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,5120,0.01885955532391866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,5120,0.006911110960774952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,6144,0.009008000294367472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,5120,0.008980444735950893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,5120,0.008280889027648503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,4096,0.017514665921529133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,4096,0.007100444700982835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,4096,0.007658667034573025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,4096,0.0076142218377855085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,3584,0.01646488904953003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,3584,0.0069226668112807805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,3584,0.007635555333561367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,3584,0.007257777783605788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,3072,0.015976889265908133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,3072,0.006450666735569636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,3072,0.007264888948864407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,3072,0.0069351109365622205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,2560,0.015447111593352424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,2560,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,2560,0.0069448889957533936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,1024,2048,0.01646222174167633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,2048,0.014125333891974555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,2048,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,2560,0.006642666541867786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,2048,0.006858666737874349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,1536,0.013786666923099093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,1536,0.004556444370084339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,1536,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,2048,0.006574222197135289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,1024,0.003830222205983268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,1024,0.013048888908492194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,1024,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,1536,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,768,0.015429332852363586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,768,0.0035377778112888336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,768,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,1024,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,512,0.0123831108212471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,512,0.0031759999692440033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,512,0.005554666535721884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,768,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,256,0.014769777655601501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,256,0.0032222221295038858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,256,0.005532444351249271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,512,0.005224888937340842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,512,128,0.013105777402718862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,128,0.0028408887899584244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,128,0.005223999834722943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,256,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,64,0.0026311110705137253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,64,0.0055440002017551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,512,32,0.0028239999794297745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,512,32,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,512,128,0.004937777916590373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,65536,0.016169778174824186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,65536,0.05394755469428169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,16384,0.026100445124838088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,65536,0.05198755529191759
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,16384,0.008950221869680617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,16384,0.015815110670195687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,12288,0.023005333211686876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,16384,0.01607466737429301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,12288,0.00833511104186376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,12288,0.013217777841620974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,10240,0.020072888996866014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,12288,0.013575110998418597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,10240,0.0075937774446275495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,10240,0.012411555482281579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,8192,0.018906666172875296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,65536,0.07658577627605863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,8192,0.007315555380450354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,10240,0.011690666278203329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,8192,0.010686222049925061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,7168,0.006612444503439798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,8192,0.01072177787621816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,7168,0.01754666699303521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,7168,0.010055111514197456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,6144,0.01682222220632765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,6144,0.006251555350091722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,7168,0.009699555734793345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,6144,0.009316444396972656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,5120,0.016230222251680162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,6144,0.009013333254390294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,5120,0.006219555520349079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,5120,0.009019555317031013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,5120,0.008288000192907121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,4096,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,4096,0.014754666222466363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,4096,0.018217777212460835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,4096,0.007631999750932057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,3584,0.013863110707865821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,3584,0.006261333409282897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,4096,0.00794933322403166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,5120,0.018191999859280057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,3584,0.007425777614116669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,3072,0.013570666313171387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,3584,0.007249777515729268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,3072,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,3072,0.007438222567240397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,2560,0.013736888766288757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,3072,0.0069617778062820435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,2560,0.006304889089531369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,2560,0.007314666277832455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,2048,0.012477333347002665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,2560,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,2048,0.005764444255166584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,2048,0.0063582224150498705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,2048,0.0063031112982167145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,1536,0.004585777719815572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,1536,0.012723555167516073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,1536,0.005960000058015187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,1024,0.011751111182901593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,1024,0.003843555433882607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,1536,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,1024,0.0058853332367208265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,768,0.014436443646748861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,1024,0.005623999983072281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,768,0.0034977777136696708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,768,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,512,0.011386666860845355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,768,0.005554666535721884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,512,0.0031679999083280563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,512,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,512,0.013070222404268054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,256,0.013068444199032254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,512,0.005247999810510212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,256,0.002921777880854077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,256,0.005230222311284807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,128,0.012093333734406365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,256,0.0052666668262746595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,64,0.002557333351837264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,128,0.002528888897763358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,128,0.005197333378924264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,256,128,0.005048000150256687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,64,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,256,32,0.002543111021320025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,256,32,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,65536,0.0144177774588267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,65536,0.07230844762590197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,65536,0.049313777022891574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,65536,0.05026577909787496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,16384,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,16384,0.015488889482286243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,16384,0.016544888416926067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,12288,0.022006221943431433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,12288,0.007609777980380588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,12288,0.01311822235584259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,12288,0.013162666724787818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,10240,0.006979555719428592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,10240,0.012271111210187277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,768,128,0.013136000268989138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,10240,0.011739555332395764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,8192,0.017831999394628737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,8192,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,8192,0.010724444356229572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,8192,0.010399999717871347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,7168,0.016876444220542908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,7168,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,7168,0.009736888938479954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,7168,0.009642666412724389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,6144,0.016107555892732408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,6144,0.006303999986913469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,6144,0.009296889106432596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,6144,0.008991999758614434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,5120,0.015451555450757345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,5120,0.005970666805903117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,5120,0.00868000007337994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,4096,0.007686222592989604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,4096,0.014154665999942355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,5120,0.008271999657154083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,4096,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,3584,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,4096,0.007630222373538547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,3584,0.007368889119890001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,3584,0.007334222396214803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,3072,0.013417777915795645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,3072,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,3072,0.007288888924651676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,3072,0.007267555428875818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,2560,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,2560,0.006602666858169768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,2560,0.006918222126033571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,2048,0.012402666939629449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,2048,0.005577777822812398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,2048,0.006642666541867786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,1536,0.004545777622196409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,1536,0.011747555600272285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,2048,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,1536,0.006226666685607698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,1536,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,1024,0.011697777443461947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,1024,0.0038399998512532976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,1024,0.005581333405441708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,1024,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,768,0.013731555806265937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,768,0.003301333429084884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,768,0.005475555443101459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,768,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,512,0.010375111467308467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,512,0.0031999999450312722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,512,0.005591111050711737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,256,0.011383111278216044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,256,0.0028026666906144884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,512,0.005203555441564984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,256,0.00526577772365676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,128,0.010100444157918295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,256,0.00518844442235099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,128,0.0028497777465316984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,128,0.005196444276306364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,64,0.002640888922744327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,16,128,128,0.004977777600288391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,64,0.004920888692140579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,128,32,0.002507555608948072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,128,32,0.005208888815508948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,65536,0.010743111371994019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,65536,0.04754044281111824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,16384,0.0069884442620807225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,16384,0.015665777855449252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,12288,0.006191111273235745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,12288,0.013077333569526672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,10240,0.005948444621430502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,10240,0.011734222372372946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,8192,0.006298666612969504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,8192,0.010678222609890832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,7168,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,7168,0.009755555954244401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,6144,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,6144,0.009297777381208207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,5120,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,5120,0.008622222476535374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,4096,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,4096,0.00794933322403166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,3584,0.005983110931184556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,3584,0.007302222152551015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,3072,0.005618666609128316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,3072,0.007239999870459239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,2560,0.005970666805903117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,2560,0.006948444578382704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,2048,0.005583999885453119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,2048,0.006343110981914733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,1536,0.0042453333735466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,1536,0.005920889063013925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,1024,0.003654222107595868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,1024,0.005877333382765452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,768,0.0031946665710873077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,768,0.005382222019963794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,512,0.0031840000301599503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,512,0.00526577772365676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,256,0.0028648889727062653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,256,0.005195555587609609
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,128,0.00286311118139161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,128,0.0052168890833854675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,64,0.002516444358560774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,64,0.005183111048407025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,64,32,0.0024826667375034755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,64,32,0.004913777940803104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,65536,0.009192888935407003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,65536,0.04662222332424588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,16384,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,16384,0.015792000624868605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,12288,0.006211555666393704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,12288,0.013181333740552267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,10240,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,10240,0.01204622205760744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,8192,0.011053333679835001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,8192,0.005573333137565189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,7168,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,7168,0.00998755544424057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,6144,0.006298666612969504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,6144,0.009481777747472128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,5120,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,5120,0.00868266655339135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,4096,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,4096,0.00793422261873881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,256,65536,0.074654221534729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,3584,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,3072,0.007287110719415877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,3584,0.007489778101444244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,3072,0.005906666732496685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,2560,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,2048,0.00647199981742435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,2560,0.007003555695215861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,2048,0.005285333428117964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,1536,0.0042408886882993905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,1536,0.006211555666393704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,1024,0.003546666767862108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,1024,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,768,0.0034337778472238113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,768,0.00573333352804184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,512,0.0032115555885765287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,512,0.005018666800525454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,256,0.0028479999552170434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,256,0.005551110953092575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,128,0.0028684443483750024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,128,0.0052533331844541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,64,0.002504000026318762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,64,0.005254222287072076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,16,32,32,0.0025191110455327565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,16,32,32,0.0052177777720822235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,16384,0.3429315619998508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,16384,0.23130310906304252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,16384,0.1733297771877713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,12288,0.20583111710018584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,12288,0.24447200033399794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,12288,0.13382755385504827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,10240,0.17776710457272002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,10240,0.205878218015035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,16384,0.026386666629049513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,10240,0.11311466164059109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,8192,0.15416977140638563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,8192,0.15567555692460802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,8192,0.09701422188017104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,10240,0.019919999771647982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,7168,0.13302666611141628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,7168,0.14505244625939265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,7168,0.08048088683022393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,2560,0.013073777986897362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,6144,0.12816266218821207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,16,128,3584,0.013463999662134381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,6144,0.11497422059377034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,6144,0.07018399900860257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,5120,0.09658933348125881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,5120,0.12140533659193252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,5120,0.059580445289611816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,4096,0.10337511036131118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,4096,0.08044799831178454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,4096,0.04833333359824287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,16384,0.13702489270104304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,3584,0.09640177753236558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,3584,0.0691244469748603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,3584,0.043515556388431124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,12288,0.10473599698808457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,3072,0.08953866693708633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,3072,0.05998488929536608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,3072,0.03822044531504313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,10240,0.08704177538553874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,2560,0.08255910873413086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,8192,0.07129866547054715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,2560,0.05070666803254021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,2560,0.03251466817326016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,2048,0.07635111278957792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,2048,0.041466666592492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,2048,0.02830222249031067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,6144,0.05494844582345751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,7168,0.06260622209972806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,1536,0.06913066572613187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,1536,0.03205688794453939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,1536,0.022151110900772944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,1024,0.06248888704511854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,1024,0.023333332604832117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,1024,0.01822933389080895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,5120,0.046557333734300405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,4096,0.04736977815628052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,768,0.031239999665154353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,768,0.01871200071440803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,768,0.016137777103318107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,512,0.05647377835379707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,512,0.013397333522637686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,512,0.014868444866604276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,3584,0.03489511211713155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,256,0.024154666397306655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,256,0.00831733312871721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,256,0.012913777596420713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,3072,0.03085244364208645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,65536,128,0.01924533314175076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,128,0.006485333459244833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,128,0.010047111246320937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,2560,0.02721244427892897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,64,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,64,0.01033155537313885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,2048,0.023238221804300945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,65536,32,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,65536,32,0.010336000058386061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,1024,0.017196445001496207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,65536,0.321486234664917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,1536,0.019370666808552213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,65536,0.22319467862447104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,768,0.015027556154463025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,16384,0.08732977840635513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,16384,0.0730453332265218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,65536,0.18280800183614096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,16384,0.05003911256790161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,12288,0.07629155450397067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,12288,0.06700177987416585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,12288,0.04313688808017307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,10240,0.06549155712127686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,10240,0.054655998945236206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,10240,0.03760800096723769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,8192,0.058187557591332324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,512,0.012930666406949362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,8192,0.043615112702051796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,8192,0.030799110730489094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,7168,0.05713333023918999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,7168,0.037632889217800565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,7168,0.026877333720525105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,65536,0.16858488983578154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,6144,0.04871199859513176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,16384,0.05269955595334371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,6144,0.03244533472590976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,6144,0.0244835548930698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,12288,0.034641779131359525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,5120,0.027791112661361694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,10240,0.02892000145382351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,5120,0.02163377735349867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,8192,0.02474311159716712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,7168,0.021589333812395733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,4096,0.02370133333735996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,4096,0.017433777451515198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,3584,0.03765511181619432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,3584,0.020589333441522386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,3584,0.01685688893000285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,6144,0.019714666737450492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,3072,0.03552800085809495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,3072,0.018400000201331247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,3072,0.015463110473420886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,5120,0.017501332693629794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,2560,0.03353510962592231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,2560,0.015488000379668342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,2560,0.013760888742076026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,4096,0.015295111470752291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,2048,0.03215288784768846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,2048,0.013207111093733044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,2048,0.012404444317022959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,3584,0.014475555883513557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,1536,0.029172443681293066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,1536,0.010703110860453712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,1536,0.010775110787815519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,256,0.009986666341622671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,1024,0.02585244509908888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,1024,0.008450667063395182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,1024,0.008655110994974772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,3072,0.01352355546421475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,768,0.01959733333852556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,768,0.006918222126033571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,768,0.007632888853549957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,2560,0.01459111107720269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,65536,128,0.010140444669458602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,512,0.0225591113169988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,512,0.005026666654480828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,512,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,256,0.01754044493039449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,256,0.0031964443624019623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,5120,0.044673777288860746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,256,0.006284444282452266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,128,0.015468445089128284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,128,0.0031662223239739737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,128,0.006238222122192383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,2048,0.010875555376211802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,64,0.00281866660548581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,64,0.006298666612969504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,16384,32,0.0028355556229750314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,1536,0.009351111120647853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,16384,32,0.006232000059551663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,65536,0.24524267514546713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,16384,4096,0.039735111925337054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,65536,0.16428800423940024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,16384,0.058911111619737416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,16384,0.06661333640416463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,65536,0.1547280020183987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,16384,0.043318221966425575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,12288,0.06557244724697538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,12288,0.052060445149739586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,1024,0.00832444429397583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,12288,0.036659555302725896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,10240,0.05724089013205635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,10240,0.04446311129464043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,10240,0.032045334577560425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,768,0.007895111209816402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,8192,0.050104889604780406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,8192,0.03703733285268148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,512,0.006958222223652734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,8192,0.02536444365978241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,7168,0.046448889705869884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,256,0.006344888773229387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,7168,0.03261244297027588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,7168,0.025064888927671645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,6144,0.042380445533328585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,16384,128,0.006361777583758037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,6144,0.02882933285501268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,6144,0.020635555187861126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,5120,0.03966755668322245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,5120,0.025381333298153345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,5120,0.019428445233239066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,65536,0.14293955432044134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,4096,0.0347306662135654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,4096,0.022629333866967097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,16384,0.038839999172422625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,4096,0.015722667177518208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,3584,0.032949331733915545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,3584,0.01960444450378418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,3584,0.014696000350846184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,12288,0.030790223015679255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,3072,0.031261334816614784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,3072,0.01756177842617035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,3072,0.013404444687896304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,10240,0.026351110802756414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,2560,0.02977777851952447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,2560,0.015235554840829639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,2560,0.013035555680592855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,8192,0.022665777140193518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,2048,0.027426666683620874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,2048,0.013432000246312885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,2048,0.01127466642194324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,7168,0.01994755533006456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,6144,0.01813155578242408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,1536,0.026250667042202417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,1536,0.008598222500748104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,1536,0.009925333162148794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,5120,0.0162808895111084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,1024,0.022031999296612207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,1024,0.0069448889957533936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,1024,0.007295110987292395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,4096,0.014303111367755465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,768,0.01886399918132358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,768,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,768,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,3584,0.01350044459104538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,512,0.019889778561062284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,512,0.0038808890514903595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,512,0.0063235556913746735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,3072,0.014058666096793281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,256,0.01685688893000285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,256,0.0032035555276605818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,256,0.006241777704821692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,12288,128,0.01477777792347802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,128,0.003124444435040156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,2560,0.011571555501884885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,128,0.005932444499598608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,64,0.0026213334252436957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,64,0.006218666831652324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,12288,32,0.0028035555862718155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,12288,32,0.006226666685607698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,65536,0.1480382283528646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,2048,0.009322666459613377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,65536,0.20713510778215197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,1536,0.008273777862389883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,65536,0.1415004465315077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,16384,0.059362669785817467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,16384,0.05264800124698215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,16384,0.041025777657826744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,1024,0.007658667034573025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,12288,0.06368622514936659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,12288,0.043881777260038585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,768,0.007259555160999298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,12288,0.03269244564904107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,10240,0.038006222910351224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,10240,0.06039110819498698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,10240,0.02962755494647556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,8192,0.04871111114819845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,512,0.006637333167923822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,8192,0.03188088867399428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,8192,0.02437066700723436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,7168,0.043882668018341064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,7168,0.02901244494650099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,256,0.006256888724035687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,7168,0.021290666527218286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,6144,0.040615999036365084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,12288,128,0.006288888967699475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,6144,0.025752888785468206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,6144,0.020277332928445604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,5120,0.03730400072203742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,5120,0.02241777711444431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,5120,0.017027555240525138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,65536,0.13840977350870767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,4096,0.03346133232116699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,16384,0.04233244392606947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,4096,0.019506666395399306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,4096,0.014769777655601501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,3584,0.03222489025857713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,3584,0.017502221796247695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,3584,0.014064000712500678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,12288,0.030040889978408813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,3072,0.030749334229363337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,10240,0.02569155560599433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,3072,0.015788444214397006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,3072,0.01254488858911726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,2560,0.028431110911899145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,2560,0.014203555054134794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,8192,0.021930666433440313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,2560,0.011389333340856763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,2048,0.026707554856936138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,2048,0.010909332997269101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,2048,0.01072088877360026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,7168,0.019464888506465487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,1536,0.02367111047108968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,1536,0.009148444566461775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,1536,0.00797155582242542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,6144,0.017439999514155917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,1024,0.020630222227838304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,1024,0.006577777779764599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,1024,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,5120,0.015849777393870883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,768,0.018925334016482036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,768,0.004858666823969947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,768,0.006611555400821898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,4096,0.013763555222087435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,512,0.018535999788178336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,512,0.0034737777378824023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,512,0.005938666562239329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,3584,0.014461333552996317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,256,0.01720977822939555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,256,0.003167111012670729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,256,0.006250666661394968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,3072,0.01275644451379776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,2560,0.010904889139864178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,10240,128,0.015468445089128284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,128,0.0028906667398081887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,128,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,2048,0.008653333617581261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,64,0.002850666642189026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,64,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,10240,32,0.0028284444577164124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,10240,32,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,1536,0.00775911079512702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,65536,0.1730266677008735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,1024,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,65536,0.13163555992974177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,16384,0.046895109944873385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,16384,0.044976890087127686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,65536,0.09860889116923015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,768,0.006974221931563483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,16384,0.029747555653254192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,12288,0.03280444608794318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,12288,0.0551164448261261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,12288,0.02489600082238515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,10240,0.027520888381534155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,10240,0.04828000068664551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,10240,0.023041778140597876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,8192,0.04583466715282864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,8192,0.023992000354660883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,8192,0.018760888112915885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,512,0.006572444405820634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,7168,0.04054222173160977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,7168,0.02056533263789283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,256,0.006173333360089197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,7168,0.017775111728244357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,6144,0.03797511259714762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,6144,0.018160000443458557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,6144,0.015595555305480957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,10240,128,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,65536,0.07872266901863946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,5120,0.015799999237060547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,5120,0.014718222949239941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,16384,0.02603911028967963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,4096,0.031721777386135526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,12288,0.020471110939979553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,4096,0.01331999980741077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,4096,0.012404444317022959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,10240,0.016968000266287062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,3584,0.011913778053389655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,3584,0.011400889191362592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,8192,0.015103111664454142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,3072,0.02847466535038418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,3072,0.010692444112565784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,3072,0.010934222075674268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,7168,0.014175999495718213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,2560,0.026709333062171936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,2560,0.009666666388511658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,2560,0.01033155537313885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,6144,0.012484444512261285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,2048,0.02475999957985348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,2048,0.008285333712895712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,2048,0.00887822194231881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,5120,0.011696000066068439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,1536,0.020970667401949566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,1536,0.007259555160999298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,1536,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,4096,0.011375111010339526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,1024,0.01955200069480472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,1024,0.005214222189452913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,1024,0.006238222122192383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,3584,0.009657777845859528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,768,0.018220444520314533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,768,0.003549333247873518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,768,0.005940444353553984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,3072,0.008403555386596257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,512,0.01754666699303521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,512,0.0034968890249729156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,512,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,2560,0.008658666577604082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,256,0.016840888394249808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,256,0.0029760001020299066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,256,0.00553688903649648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,2048,0.007273777491516537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,128,0.014589332871966891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,128,0.0028080000645584534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,128,0.00554044461912579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,1536,0.006616888774765863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,64,0.002520888836847411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,64,0.005559999909665849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,8192,32,0.0028906667398081887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,1024,0.0063982220987478895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,8192,32,0.0052426668504873914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,768,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,65536,0.14964355362786189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,65536,0.1269271108839247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,16384,0.04497866829236349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,65536,0.09266666571299235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,16384,0.04145510991414388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,512,0.005658666706747479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,16384,0.02743822170628442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,12288,0.033415112230512835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,12288,0.05807377894719442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,12288,0.023203555080625746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,256,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,10240,0.05128355489836799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,10240,0.028559999333487615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,8192,128,0.0052595556610160405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,10240,0.021006221572558086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,65536,0.07431555456585355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,8192,0.024188444018363953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,8192,0.01831644442346361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,16384,0.023807111713621352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,7168,0.04110222061475118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,7168,0.01868711080816057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,12288,0.01904177831278907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,7168,0.01661422186427646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,6144,0.03751111030578613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,6144,0.016526222229003906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,6144,0.015464888678656684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,5120,0.0337777766916487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,5120,0.014402666025691561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,10240,0.01646399994691213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,5120,0.01368622233470281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,4096,0.03149244520399306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,4096,0.012484444512261285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,4096,0.011686222420798408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,8192,0.014838222000333997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,3584,0.028791109720865887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,3584,0.011036444041464063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,5120,0.034910221894582115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,3584,0.011051555474599203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,7168,0.013095999757448832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,3072,0.02751733362674713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,3072,0.010049777726332346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,3072,0.01036711119943195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,2560,0.025742222865422566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,2560,0.008929777476522658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,6144,0.012033777932325998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,2560,0.009696000152164036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,2048,0.023007111416922674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,2048,0.007937777373525832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,2048,0.007650666766696506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,5120,0.011168888873524137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,1536,0.02025422288311852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,1536,0.006842666616042455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,1536,0.00656444455186526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,4096,0.010094222095277574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,1024,0.018915555543369718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,8192,3584,0.030163556337356567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,1024,0.004588444613748127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,1024,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,768,0.018557333283954196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,768,0.003855111284388436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,768,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,3584,0.008978666530715095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,512,0.01721244388156467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,3072,0.007623111208279927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,512,0.0035031110876136353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,512,0.005583999885453119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,256,0.016117332710160148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,256,0.003192000091075897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,256,0.005434666656785541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,2560,0.0069395556218094295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,128,0.014727110664049784
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,2048,0.006907555378145642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,128,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,128,0.0052871112194326185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,64,0.0028773333049482773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,64,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,1536,0.006256888724035687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,7168,32,0.002524444419476721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,7168,32,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,1024,0.005943999936183293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,65536,0.12279821766747369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,65536,0.11845689349704319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,768,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,65536,0.08365422487258911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,16384,0.04238577683766683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,16384,0.03548889027701484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,16384,0.025824889540672302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,12288,0.028453333510292902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,12288,0.05224622289339701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,12288,0.021557332740889654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,512,0.00556355549229516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,10240,0.04611466659439934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,10240,0.024481778343518574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,10240,0.019262222780121695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,256,0.0052533331844541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,8192,0.041490667396121554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,8192,0.021201777789327834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,8192,0.016573333077960543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,7168,0.03817066550254822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,7168,0.018967999352349173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,7168,128,0.005225777626037598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,7168,0.015544000599119397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,6144,0.03507288959291246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,6144,0.017154667112562392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,6144,0.013960000541475085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,65536,0.07174222336875068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,5120,0.03325244453218248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,5120,0.01513066722287072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,16384,0.022822222775883142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,5120,0.01293511109219657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,12288,0.018561777141359117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,4096,0.029662221670150757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,4096,0.011725333001878528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,4096,0.011385777758227454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,10240,0.016136889656384785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,8192,0.014130666851997375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,3584,0.01034311122364468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,3584,0.010370666782061258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,7168,0.012714666624863943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,3072,0.027097778187857732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,3072,0.009186666872766282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,3072,0.00978577799267239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,6144,0.01202488856183158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,7168,8192,0.046479112572140165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,2560,0.02402399977048238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,2560,0.008463111188676622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,2560,0.008212444682916006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,2048,0.021499555971887376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,2048,0.007359111474619971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,5120,0.01070755554570092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,2048,0.00720711135201984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,1536,0.019554666346973844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,1536,0.006221333311663733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,1536,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,4096,0.009359111388524374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,1024,0.004224888980388641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,1024,0.01791111131509145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,1024,0.005947555518812603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,768,0.01791911158296797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,768,0.0035706667436493766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,3584,0.007883555359310573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,768,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,512,0.016138666205936007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,512,0.0032026666320032547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,512,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,3072,0.007264888948864407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,256,0.015127110812399121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,256,0.0031600000543726813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,256,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,2560,0.007308444215191736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,128,0.014783110883500842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,128,0.0028159999185138275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,128,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,2048,0.006609777609507243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,64,0.002528000002106031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,64,0.0052871112194326185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,6144,32,0.002519999941190084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,1536,0.0064106666379504735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,6144,32,0.00526311124364535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,1024,0.005929777605666055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,65536,0.10772177908155654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,65536,0.11440355247921413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,768,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,16384,0.04084088736110263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,65536,0.07776533232794867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,16384,0.03205777870284186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,16384,0.023748444186316595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,12288,0.025910221868091162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,12288,0.05070133341683281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,512,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,12288,0.019869332512219746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,10240,0.022075555390781824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,10240,0.01795733306143019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,256,0.005230222311284807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,8192,0.04116444455252753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,8192,0.01947377787695991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,6144,128,0.005382222019963794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,8192,0.015455999308162265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,7168,0.03809600075085958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,7168,0.017312000195185345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,65536,0.06939822435379028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,7168,0.014343111051453484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,16384,0.02199555602338579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,6144,0.015560888581805758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,6144,0.013345777988433838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,12288,0.017907554904619854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,10240,0.015683554940753512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,5120,0.013737777868906656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,5120,0.01202933324707879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,8192,0.014016888207859464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,4096,0.027514666318893433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,4096,0.010775110787815519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,4096,0.01036800030204985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,7168,0.01239911135700014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,3584,0.027847111225128174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,3584,0.009645333720578088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,3584,0.009751111268997192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,6144,0.01146488885084788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,3072,0.02366133365366194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,3072,0.008719111482302347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,3072,0.008660444782839881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,5120,0.009637333452701569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,2560,0.020959999826219346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,2560,0.007927999728255803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,2560,0.00793866647614373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,4096,0.008016000191370646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,2048,0.01958044370015462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,2048,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,2048,0.006592000110281839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,3584,0.007654222349325816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,3072,0.007646222081449296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,1536,0.018538667096032035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,1536,0.005210666606823604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,1536,0.00665511108107037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,2560,0.007264888948864407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,1024,0.016982222596804302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,1024,0.003858666867017746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,1024,0.00591822216908137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,2048,0.006561777657932705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,768,0.017846221725145977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,768,0.0035537779331207275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,768,0.005576000031497743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,1536,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,512,0.015816888875431485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,6144,3584,0.0280675556924608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,512,0.0032213332338465583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,512,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,1024,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,256,0.0028648889727062653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,256,0.0055440002017551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,768,0.005566222386227713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,128,0.002781333401799202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,128,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,512,0.005621333503060871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,64,0.0031715554909573663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,64,0.005561777700980504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,5120,32,0.002578666640652551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,5120,32,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,256,0.005230222311284807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,65536,0.09198577536476983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,65536,0.09075022406048244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,65536,0.0713955561319987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,16384,0.03406577640109592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,16384,0.026358221968015034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,5120,128,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,16384,0.022022222479184467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,12288,0.05190755592452156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,12288,0.020558221472634208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,12288,0.018530666828155518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,65536,0.06754222181108263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,16384,0.02125866711139679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,10240,0.018093332648277283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,10240,0.01679111189312405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,12288,0.017500445246696472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,8192,0.04009866714477539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,8192,0.015803555647532146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,8192,0.015040889382362366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,10240,0.04692177640067207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,7168,0.03545511100027297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,7168,0.014142221874660917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,7168,0.013510222236315409
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,6144,0.035593777894973755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,10240,0.014808000789748298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,6144,0.0332640012105306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,6144,0.013052444491121503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,6144,0.012727999852763282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,5120,0.030233777231640283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,5120,0.011911999848153857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,5120,0.011702222128709158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,8192,0.013438222308953604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,4096,0.028804444604449805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,4096,0.010661332971519895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,5120,0.03152799937460158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,4096,0.009640889035330879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,3584,0.026047110557556152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,7168,0.011720000041855706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,3584,0.012038222617573209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,3584,0.008399999803966945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,3072,0.023352889551056757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,3072,0.011081777513027191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,3072,0.007655111451943715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,6144,0.01036800030204985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,2560,0.02126311096880171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,2560,0.009676444033781687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,2560,0.007241778075695038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,5120,0.009072889056470659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,2048,0.01994400057527754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,2048,0.007997333175606197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,2048,0.006603555546866522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,4096,0.007990222010347579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,1536,0.017881777551439073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,3584,0.007890666524569193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,1536,0.006607999818192587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,1536,0.006545777950021956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,1024,0.01716800034046173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,1024,0.004930666751331753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,1024,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,3072,0.007287110719415877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,768,0.017904000149832833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,768,0.004577777865860197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,768,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,2560,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,512,0.014800000521871777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,512,0.0038337777886125776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,512,0.00554577757914861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,2048,0.0064186664919058485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,256,0.014817777607176038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,256,0.003162666741344664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,256,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,1536,0.006168889088763132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,1024,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,128,0.014831110835075378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,128,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,128,0.005280888742870755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,768,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,64,0.0028248888750871024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,64,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,4096,32,0.002583111118939188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,4096,32,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,512,0.0052675555149714155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,65536,0.08027644289864434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,65536,0.08875822358661228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,256,0.0052328887912962176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,65536,0.06799733638763428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,16384,0.024148444334665935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,16384,0.021716444028748408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,4096,128,0.005280888742870755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,12288,0.051628443929884166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,12288,0.01904800037542979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,12288,0.01830222209294637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,128,0.014800000521871777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,5120,256,0.016569778323173523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,10240,0.016561778055297006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,10240,0.045520000987582736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,10240,0.01627466744846768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,65536,0.06645244359970093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,8192,0.040287110540601946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,8192,0.01478488908873664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,8192,0.014644443988800049
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,7168,0.03592533204290602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,7168,0.013806222213639153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,16384,0.020900444851981267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,7168,0.013401777380042605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,6144,0.033067554235458374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,6144,0.012429333395428128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,6144,0.01236177815331353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,5120,0.03055466545952691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,5120,0.011342222491900126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,5120,0.011167999770906238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,12288,0.01719555589887831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,10240,0.014771555860837301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,4096,0.027820444769329492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,4096,0.010048000348938836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,4096,0.009015999734401703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,8192,0.012661333713266583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,3584,0.011687111523416309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,3584,0.007932444413503012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,7168,0.01073333372672399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,3072,0.022671111755900916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,6144,0.009662222531106737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,3072,0.009641778137948778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,3072,0.007354666789372762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,2560,0.021280889709790547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,2560,0.008625778059164682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,2560,0.006978666616810693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,5120,0.008567999634477828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,2048,0.019596444235907663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,2048,0.006949333681000604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,4096,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,2048,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,1536,0.017839999662505258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,1536,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,3584,0.007596444752481248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,1536,0.006269333263238271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,1024,0.016336000627941556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,1024,0.004857777721352047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,1024,0.005592888842026393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,3072,0.007319110963079665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,768,0.017839999662505258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,768,0.0042275554604000514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,768,0.005573333137565189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,2560,0.006873777757088344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,512,0.01479022204875946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,512,0.0038702223036024305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,512,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,2048,0.006567111031876669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,256,0.015488889482286243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,256,0.002855111120475663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,1536,0.006214222146405115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,256,0.005487999982304043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,128,0.01477688882086012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,128,0.0027991111079851785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,1024,0.005616000129116907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,4096,10240,0.04591644472546048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,128,0.00553955551650789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,64,0.0028168888141711554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,64,0.005277333160241445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3584,32,0.0028764444092909494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3584,32,0.005557333429654439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,768,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,65536,0.06290666924582587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,65536,0.08670577738020156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,65536,0.06569866339365642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,512,0.005274666680230035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,16384,0.02052533295419481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,16384,0.03299466768900553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,16384,0.020627554919984605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,12288,0.017276444368892245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,12288,0.04658311274316576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,256,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,12288,0.017616000440385606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,10240,0.042640888028674655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,10240,0.015156444576051501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3584,128,0.005231110999981563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,10240,0.01612977849112617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,8192,0.03888088795873854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,8192,0.013307555682129331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,65536,0.06572800212436251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,8192,0.014071111877759298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,16384,0.020794666475719877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,7168,0.035507556464936994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,7168,0.012450666891203986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,7168,0.013053333593739403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,12288,0.016876444220542908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,6144,0.03264266583654616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,6144,0.011401777466138205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,6144,0.012081777883900536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,5120,0.028368890285491943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,10240,0.01440622243616316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,5120,0.01070666644308302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,5120,0.010519111322032081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,4096,0.02566844390498267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,4096,0.009648000200589498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,4096,0.008416888614495596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,8192,0.011707555916574268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,3584,0.022847111026446026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,3584,0.009299555586444007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,7168,0.010450666977299584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,3584,0.007600000335110559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,3072,0.021869333253966436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,6144,0.009343110852771336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,3072,0.009026666482289633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,3072,0.007430222299363878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,2560,0.020240889655219186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,2560,0.00798044436507755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,5120,0.008395555946562026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,2560,0.007247111035717859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,2048,0.018571555614471436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,2048,0.00656088896923595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,2048,0.00657066661450598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,4096,0.007992000215583378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,1536,0.017829333742459614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,1536,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,1536,0.006288888967699475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,3584,0.007670222057236566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,1024,0.0052684446175893145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,1024,0.005932444499598608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,3072,0.007272889216740926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,768,0.01773244473669264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,768,0.00435644429590967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,768,0.005888888819350137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,2560,0.0069146665434042616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,512,0.01442488862408532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,512,0.003991999973853429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,2048,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,512,0.005564444594913059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,256,0.014879999889267815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,256,0.003539555602603488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,1536,0.006016888966162999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,256,0.005224888937340842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,128,0.014478221535682678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,128,0.0031786666562159858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,1024,0.005762666463851929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,128,0.00526577772365676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,64,0.0028124445428450904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,64,0.005224888937340842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,3072,32,0.0024968888610601425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,768,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,3072,32,0.0052693333062860705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,65536,0.057652446958753795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,65536,0.08559199836519028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,512,0.00553955551650789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,65536,0.06304088566038343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,16384,0.03228088882234361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,16384,0.01891377733813392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,256,0.005239999956554837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,16384,0.020541333489947848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,12288,0.047127111090554126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,12288,0.015446222490734525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,3072,128,0.005245333330498801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,12288,0.017186666528383892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,10240,0.04066666629579332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,10240,0.013951111170980664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,10240,0.015680889288584392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,65536,0.0648568868637085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,8192,0.03680977887577481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,8192,0.012127999630239276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,8192,0.013549333645237817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,16384,0.020464888877338834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,7168,0.03324000040690104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,7168,0.011412444214026133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,7168,0.012434666355450949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,6144,0.030249777767393324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,12288,0.01640266676743825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,6144,0.010613333019945355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,6144,0.011672889192899069
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,16384,0.033246222469541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,5120,0.024002666274706524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,5120,0.009679999616410997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,5120,0.009351111120647853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,10240,0.013297778036859302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,4096,0.024106666445732117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,8192,0.011030221978823343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,4096,0.008951110972298516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,4096,0.008265777594513362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,7168,0.009999111294746399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,3584,0.008663111262851292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,3584,0.007652444144090016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,6144,0.009350222018029954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,3072,0.02127911150455475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,3072,0.00832799987660514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,5120,0.008624888956546783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,3072,0.007247111035717859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,4096,0.007814221911960179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,2560,0.0068400001360310455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,2560,0.007248000138335758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3584,3584,0.02568000058333079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,3584,0.007365333537260692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,2048,0.005904888941182031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,2048,0.01789688898457421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,2048,0.006577777779764599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,1536,0.01662400033738878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,3072,0.007234666910436418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,1536,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,1536,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,2560,0.006720888945791457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,1024,0.004568888909286923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,1024,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,2048,0.006575111299753189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,768,0.017295110556814406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,768,0.0042035554846127825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,768,0.005621333503060871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,1536,0.006250666661394968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,512,0.013780444032616086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,512,0.003847111016511917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,512,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,1024,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,256,0.003543111185232798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,256,0.005218666460778978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,768,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,128,0.014755555325084262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,128,0.0028728888266616394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,512,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,128,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,64,0.002536000063021978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,64,0.005535999933878581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2560,32,0.0028159999185138275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2560,32,0.005240889059172736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,256,0.005201777650250329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,65536,0.07852622535493639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,65536,0.05156088868776957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,65536,0.060625778304206006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2560,128,0.005229333208666907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,16384,0.030800888935724895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,16384,0.015662221444977652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,16384,0.019883554842736986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,12288,0.04923289020856222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,12288,0.013056000073750814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,12288,0.01702222228050232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,65536,0.06367378102408515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,10240,0.04160888989766439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,16384,0.019875556230545044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,10240,0.011730666789743634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,10240,0.015442666080262927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,8192,0.036601778533723615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,8192,0.011060444845093621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,12288,0.01460977726512485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,8192,0.012639111114872826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,7168,0.033195555210113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,7168,0.010738666686746808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,7168,0.011365333365069495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,10240,0.012416889270146688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,6144,0.029129776689741347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,6144,0.009786666267448002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,6144,0.009667555491129557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,8192,0.011023110813564725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,5120,0.027129777603679236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,5120,0.008938666847017076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,7168,0.009674666656388177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,5120,0.008965333302815756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,4096,0.025639111797014873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,4096,0.010718222293588849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,4096,0.00794400026400884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,6144,0.008952000074916417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,3584,0.022666666242811415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,3584,0.009684444301658207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,3584,0.007640889121426477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,5120,0.00833333366447025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,3072,0.021831999222437542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,3072,0.006572444405820634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,3072,0.007622222105662028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,4096,0.008177777959240807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,3584,0.007479111353556315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,2560,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,2560,0.0069857777820693115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,3072,0.007024889190991719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,3072,1024,0.016166221764352586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,2048,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,2560,0.00674933319290479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,2048,0.006607111129495833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,2048,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,1536,0.017557332913080852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,1536,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,1536,0.006280000011126201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,1536,0.006304889089531369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,1024,0.005008888741334279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,1024,0.005921777751710679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,3584,0.02127644419670105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,768,0.017840888765123155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,768,0.0041697778635554844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,768,0.005919999960396025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,1024,0.0058666666348775225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,512,0.01350044459104538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,512,0.003524444583389494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,512,0.005296000176005893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,768,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,1024,0.01549600064754486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,256,0.015470221638679504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,2560,0.020113777783181932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,256,0.0032097777972618737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,256,0.0052533331844541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,512,0.0052391112678580815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,128,0.01475822263293796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,128,0.002871111035346985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,128,0.005250666704442766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,64,0.0028826666788922418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,64,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,2048,32,0.0028560000161329904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,2048,32,0.0052471111218134565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,65536,0.03654933306905959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,65536,0.07745599746704102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,256,0.0052586665583981415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,65536,0.059113780657450356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,16384,0.013306666579511432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,16384,0.03029688861634996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,16384,0.01943288909064399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,2048,128,0.00526311124364535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,12288,0.011404444773991903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,12288,0.046443555090162486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,12288,0.016736888223224215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,10240,0.03846577803293864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,65536,0.06362044148974948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,10240,0.010478222535716163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,10240,0.013767110804716745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,16384,0.018276444739765592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,8192,0.031946665710873075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,8192,0.010213333699438307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,8192,0.012042666474978128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,12288,0.013524444566832649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2560,256,0.014813333749771118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,7168,0.02978844443957011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,7168,0.009885333478450775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,7168,0.01037599974208408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,6144,0.028858668274349634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,6144,0.008993777963850234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,10240,0.01201333353916804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,6144,0.009682666924264696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,5120,0.024488889508777197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,5120,0.0076044441925154785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,8192,0.01071999967098236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,5120,0.008993777963850234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,7168,0.00998755544424057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,4096,0.0074551113777690465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,4096,0.007719111111429002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,3584,0.02194044490655263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,6144,0.008989333278603023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,3584,0.0069751110341813825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,3584,0.007659555309348636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,5120,0.008462222086058723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,3072,0.02092355489730835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,3072,0.006256888724035687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,3072,0.007306666837798224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,4096,0.007923555870850882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,2560,0.01923022170861562
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,2560,0.005592888842026393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,2560,0.007056000332037608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,3584,0.007627555893527136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,2048,0.017901332841979135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,2048,0.006076444354322221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,2048,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,1536,0.016511110795868766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,3072,0.007252444823582967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,1536,0.0052693333062860705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,1536,0.006217777729034424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,1024,0.0151573336786694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,2560,0.006892444358931647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,1024,0.004234666625658671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,1024,0.0058853332367208265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,2048,0.006608888920810487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,768,0.017549332645204332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,768,0.004007110993067424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,768,0.00555288874440723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,1536,0.005923555543025334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,512,0.013502221968438892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,512,0.0032257777121331957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,512,0.005541333307822545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,1024,0.005649777750174205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,256,0.01515111161602868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,256,0.0031555555760860443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,256,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,768,0.005570666657553778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,128,0.014501333236694336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,128,0.002887111157178879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,512,0.0052800000541739995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,128,0.005241777747869492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,64,0.0028204443968004654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,64,0.005568000177542369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,256,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1536,32,0.002838222309947014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1536,32,0.005404444618357553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,65536,0.025635555386543274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1536,128,0.0052391112678580815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,65536,0.058182219664255776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,65536,0.06263999806510077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,16384,0.015056000815497505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,16384,0.018565333551830716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,2560,0.020917332834667627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,12288,0.03567466802067227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,12288,0.013040000365840064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,12288,0.01404088901148902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,16384,0.0173662222094006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,10240,0.03080444534619649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,10240,0.011418666276666852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,10240,0.012412444584899478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,8192,0.028026666906144884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,12288,0.013447999954223633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,8192,0.011047999891969891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,8192,0.010668444136778513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,10240,0.011746666497654386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,7168,0.024716445141368445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,7168,0.009672889278994666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,7168,0.009909333454238044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,2048,0.01885600056913164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,6144,0.008287111090289222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,6144,0.023227555884255305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,6144,0.009359999663299983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,8192,0.01069866700304879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,5120,0.023173333870040044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,5120,0.008016889293988546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,5120,0.008622222476535374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,7168,0.009402666654851701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,4096,0.007244444555706448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,6144,0.008976889153321585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,4096,0.007978666987684038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,3584,0.018611555298169453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,5120,0.008616000413894653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,3584,0.0069955554273393415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,3584,0.00794666674402025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,3072,0.018177777528762817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,4096,0.007647111184067196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,3072,0.00628977765639623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,3072,0.007248000138335758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,2560,0.017154667112562392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,3584,0.007297777467303806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,2560,0.0058808889653947614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,2560,0.006970666348934174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,3072,0.007311999797821045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,2048,0.0063866666621632045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,2048,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,2560,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,1536,0.015099555253982544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,1536,0.005591111050711737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,1536,0.006228444476922353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,2048,0.006567111031876669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,1024,0.014410666293568082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,1024,0.004511111312442356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,1024,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,1536,0.005897777775923411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,768,0.015792888071801927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,768,0.004249777644872665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,768,0.005521777603361342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,1024,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,512,0.013396444420019785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,512,0.0035048888789282907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,2048,1024,0.015955555770132277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,512,0.005520888914664586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,768,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,256,0.014896000425020853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,256,0.003145777723855443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,256,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,128,0.014420444766680399
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,128,0.0028817777832349143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,512,0.005244444641802046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,128,0.005554666535721884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,64,0.002522666628162066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,256,0.004932444542646408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,64,0.0052933332820733385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,1024,32,0.002834666727317704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,1024,32,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,65536,0.021571555071406897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,1024,128,0.0052782222628593445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,65536,0.07561600208282471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,65536,0.057406220171186656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,16384,0.010664000279373594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,16384,0.016520000166363187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,65536,0.06253244479497273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,12288,0.029487109846538965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,12288,0.01130488928821352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,16384,0.01648533344268799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,12288,0.013720889058378009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,12288,0.013805333111021252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,10240,0.009667555491129557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,10240,0.01240888900227017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,10240,0.012079999678664737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,8192,0.023027555810080633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,8192,0.008989333278603023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,8192,0.01068800025516086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1536,4096,0.024383111132515803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,7168,0.022006221943431433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,7168,0.008656000097592672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,8192,0.01071911139620675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,7168,0.010051555931568146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,6144,0.020978666014141504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,6144,0.007679111427730984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,6144,0.009367999931176504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,7168,0.009694221946928237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,5120,0.01863555610179901
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,5120,0.006748444504208035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,5120,0.008976889153321585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,6144,0.008993777963850234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,4096,0.018593778212865193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,4096,0.007289778027269576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,4096,0.00794577764140235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,5120,0.008612444831265343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,3584,0.017165333032608032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,3584,0.007252444823582967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,4096,0.007630222373538547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,3584,0.007599111232492659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,3072,0.01682400041156345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,3072,0.006625777731339137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,3072,0.007363555332024892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,3584,0.007306666837798224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,2560,0.016137777103318107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,2560,0.006617777877383762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,16384,0.029478222131729126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,65536,0.07666933536529541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,2560,0.00674044465025266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,3072,0.007248889240953658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,2048,0.005214222189452913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,2048,0.015119999647140503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,2048,0.006636444479227066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,1536,0.00489066665371259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,1536,0.006269333263238271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,2560,0.006955555743641323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,1024,0.013793778088357715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,2048,0.00629066675901413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,1024,0.0041884444653987885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,1024,0.0058222222659322954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,1536,0.006236444330877728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,768,0.015328889091809591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,768,0.0036880001425743103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,768,0.005578666511509154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,1024,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,512,0.012424888710180918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,512,0.003489777859714296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,512,0.005423111220200856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,768,0.00554933316177792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,256,0.014159111513031853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,256,0.0028719999310043124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,256,0.005295111073387994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,128,0.013091555900043912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,512,0.005355555564165115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,128,0.002871111035346985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,128,0.005223111146026188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,256,0.005225777626037598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,64,0.002509333400262727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,64,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,768,32,0.0028435554769304064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,768,32,0.0052675555149714155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,65536,0.017077333397335477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,768,128,0.004936000125275718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,65536,0.07518666982650757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,65536,0.0565813316239251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,16384,0.02678311036692725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,16384,0.011115555961926779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,16384,0.015989333391189575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,65536,0.05692355500327217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,12288,0.023640000157886084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,12288,0.00868711123863856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,12288,0.013398222625255585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,16384,0.01646488904953003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,10240,0.007987555530336168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,10240,0.020915556285116408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,10240,0.012079999678664737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,4096,0.020922667450375024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,12288,0.013386666774749756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,8192,0.01955111159218682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,8192,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,8192,0.010661332971519895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,10240,0.012040889097584618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,7168,0.022416889667510986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,7168,0.007661333514584436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,7168,0.01034311122364468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,8192,0.01071911139620675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,6144,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,6144,0.02054666644997067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,6144,0.009645333720578088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,7168,0.009663110805882348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,5120,0.006899555524190267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,5120,0.00867555538813273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,6144,0.009085333181752099
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,4096,0.015798222687509324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,4096,0.007123555574152205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,5120,0.008274666965007782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,4096,0.00795199970404307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,3584,0.015257777439223396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,3584,0.006958222223652734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,3584,0.007629333270920648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,4096,0.007607999775144789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,3072,0.01612622208065457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,3072,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,1024,2048,0.015834665960735746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,3584,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,3072,0.007274666594134436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,2560,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,2560,0.014434667097197639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,2560,0.006638222270541721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,3072,0.006953777538405524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,2048,0.014431110686726041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,2048,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,2048,0.006245333287451003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,2560,0.006903111106819577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,1536,0.0141004439857271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,1536,0.004550222307443619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,1536,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,2048,0.006292444550328785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,1024,0.013064888616402945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,1024,0.0038853333228164246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,1024,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,1536,0.00591111100382275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,768,0.0036764442920684814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,768,0.015116444892353482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,768,0.005567111074924469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,512,0.00318755561278926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,512,0.012092444631788464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,1024,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,512,0.005577777822812398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,768,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,256,0.01443199978934394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,256,0.005231110999981563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,256,0.003141333245568805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,512,0.005375999957323074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,128,0.012053333222866058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,128,0.0052151112920708126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,256,0.004976888911591636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,128,0.0028071111689011254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,64,0.0026551110463009942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,64,0.00489688871635331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,512,32,0.0025395556456512874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,16384,0.027394667267799377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,512,128,0.004933333231343163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,512,32,0.005247999810510212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,65536,0.01629155543115404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,65536,0.07281688849131267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,65536,0.04954399996333652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,16384,0.026766222384240892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,16384,0.008974221845467886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,65536,0.05204533206091987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,12288,0.008295111358165741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,12288,0.022655111220147874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,16384,0.015481778317027621
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,16384,0.016131555040677387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,12288,0.013447110851605734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,10240,0.02091288897726271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,10240,0.007639110916190677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,12288,0.013619555367363824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,10240,0.012030222349696688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,8192,0.01886044442653656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,10240,0.011725333001878528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,8192,0.0069360000391801195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,8192,0.01071200023094813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,8192,0.010438222024175854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,7168,0.017843556072976854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,7168,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,7168,0.010076444182131026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,6144,0.01682844426896837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,6144,0.00629688882165485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,7168,0.00963466697269016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,6144,0.009323555562231276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,5120,0.015577778220176697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,5120,0.006255110932721033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,6144,0.008983111215962304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,5120,0.008958222137557136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,4096,0.015442666080262927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,5120,0.008303110798199972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,4096,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,10240,0.02607911162906223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,4096,0.007648000286685095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,3584,0.014138667119873894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,3584,0.00591555568906996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,4096,0.007879999776681265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,3584,0.00795022232664956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,3072,0.013801777528391944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,3072,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,3584,0.007644444704055786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,3072,0.007260444263617198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,2560,0.013728000223636627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,3072,0.007296889192528195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,2560,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,2560,0.006970666348934174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,2048,0.01274311128589842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,2048,0.005948444621430502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,2560,0.0069066666894488865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,2048,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,2048,0.006631111105283101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,1536,0.01239111108912362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,1536,0.004511111312442356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,1536,0.0062871111763848206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,1024,0.012055111428101858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,1536,0.005894222193294101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,1024,0.003666666646798452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,1024,0.005888888819350137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,768,0.014128888646761576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,1024,0.005893333504597346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,768,0.003506666670242945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,768,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,768,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,512,0.0032195556494924757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,512,0.005446222093370225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,256,0.012774222426944308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,512,0.005197333378924264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,256,0.0028657778683635923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,256,0.005205333232879639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,128,0.01072177787621816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,128,0.0028533333291610083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,256,0.005208000126812193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,128,0.00518222235971027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,64,0.002834666727317704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,256,128,0.005053333524200651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,64,0.005212444398138259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,256,32,0.0025333333760499954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,256,32,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,65536,0.014118222726715935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,65536,0.07185688945982191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,65536,0.04683822062280443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,16384,0.007947555846638149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,16384,0.025399110383457605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,16384,0.015607110328144498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,65536,0.05032533407211304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,16384,0.016170667277442086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,12288,0.022621333599090576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,12288,0.007274666594134436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,12288,0.013111111190583972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,12288,0.013448889056841532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,10240,0.019208888212839764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,10240,0.0070977773931291364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,10240,0.012058667010731168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,10240,0.011699555648697747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,8192,0.018547554810841877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,8192,0.006508444332414203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,8192,0.010674667027261523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,8192,0.010399999717871347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,7168,0.016873778568373788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,7168,0.006640000061856375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,6144,0.006254222244024277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,7168,0.009959999885823991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,7168,0.009827555881606208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,6144,0.016170667277442086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,5120,0.005917333480384614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,6144,0.009347555538018545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,6144,0.008946666287051307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,5120,0.014816888504558139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,5120,0.008664888640244802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,5120,0.008278222547637092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,4096,0.014840000205569796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,4096,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,4096,0.007710222568776872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,4096,0.007647111184067196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,3584,0.014446222119861178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,3584,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,3584,0.007575111256705389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,3584,0.007276444799370236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,3072,0.013169777890046438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,3072,0.005551110953092575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,3072,0.007644444704055786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,768,1536,0.014505777094099255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,3072,0.0069324444565508105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,2560,0.01276266657643848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,2560,0.0059279998143513995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,2560,0.007188444336255391
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,2560,0.00686488880051507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,2048,0.01275022245115704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,1536,0.004230222354332606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,2048,0.005573333137565189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,2048,0.006272000157170826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,1536,0.006272000157170826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,2048,0.006305777778228124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,1024,0.005948444621430502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,1536,0.01239377756913503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,1536,0.006252444452709622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,1024,0.011717333561844297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,1024,0.0038488888078265717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,768,0.013807111316257052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,1024,0.005582222094138463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,768,0.003516444315512975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,768,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,768,0.005421333428886201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,512,0.010703999963071613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,512,0.0032151111712058387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,512,0.0052382221652401825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,512,0.005397333453098933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,256,0.012065778175989786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,256,0.0028577778074476454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,128,0.002789333255754577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,256,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,256,0.004914666629499859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,128,128,0.01071200023094813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,128,0.005057777795526716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,64,0.0026577777332729767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,8,128,128,0.005218666460778978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,64,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,128,32,0.002814222127199173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,128,32,0.005208000126812193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,65536,0.010100444157918295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,65536,0.04685422115855747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,16384,0.006968888971540663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,16384,0.015782222151756287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,12288,0.005910222315125995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,8192,0.006578666468461354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,12288,0.013392888837390475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,10240,0.006230222268237009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,10240,0.011787555283970304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,8192,0.010676444404655032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,7168,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,7168,0.00998488896422916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,6144,0.00592533333433999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,6144,0.009342222577995723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,5120,0.005609777652555042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,5120,0.008960888617568547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,4096,0.005246222019195557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,3072,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,4096,0.008001777860853408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,3584,0.005879111174080107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,3584,0.007340444458855524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,512,5120,0.01963644391960568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,3072,0.0069671107663048645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,2560,0.006311999840868845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,2560,0.006879111131032308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,2048,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,2048,0.006243555496136348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,768,0.0033333332588275275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,1536,0.004520888957712385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,1536,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,1024,0.0038408889538711975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,256,0.0028151110228565005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,1024,0.005895111295912001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,768,0.005543111099137201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,512,0.0032275555034478507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,64,0.002480888946188821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,512,0.0052586665583981415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,32,0.0025519999778932994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,256,0.005557333429654439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,64,128,0.0028239999794297745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,128,0.0052791109515561005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,64,0.004917333523432414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,64,32,0.004939555707905027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,65536,0.008981333010726506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,65536,0.046867556042141385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,16384,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,16384,0.015491555134455362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,12288,0.005949333310127258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,12288,0.013337777720557319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,7168,0.010002666877375709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,10240,0.005633777628342311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,10240,0.012062222593360476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,8192,0.005664888769388199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,5120,0.00868888861603207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,8192,0.01073688848151101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,7168,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,6144,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,6144,0.009635555247465769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,5120,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,4096,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,4096,0.007613333563009898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,2560,0.006959111326270633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,3584,0.005912000106440649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,3584,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,3072,0.0053022222386466126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,3072,0.007274666594134436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,1024,0.003526222374704149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,2560,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,2048,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,768,0.005575110928879843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,2048,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,512,0.0031795555518733132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,1536,0.004228444563017951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,1536,0.005953777581453323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,1024,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,768,0.0033093332830402586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,512,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,256,0.0028728888266616394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,256,0.005222222043408289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,128,0.002817777709828483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,128,0.004869333157936732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,64,0.0024968888610601425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,64,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,8,32,32,0.002500444443689452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,8,32,32,0.005249777601824866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,16384,0.33951110310024685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,12288,0.25744088490804035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,12288,0.21334755420684814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,16384,0.301750209596422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,16384,0.18353777461581758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,12288,0.13258488972981772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,10240,0.21720622645484078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,10240,0.1830151081085205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,10240,0.11322577794392903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,8192,0.16858932707044813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,8,256,512,0.011688888900809817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,8192,0.09409511089324951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,7168,0.14676177501678467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,7168,0.1545599963929918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,7168,0.08063999811808269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,6144,0.128166225221422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,6144,0.0699804425239563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,8192,0.1750497817993164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,6144,0.13059377670288086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,5120,0.1028862198193868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,5120,0.12240533034006755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,5120,0.05880977710088094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,4096,0.11031466722488403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,4096,0.08540088600582546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,4096,0.04727377825313144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,16384,0.13663466771443686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,3584,0.09578133291668361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,3584,0.07327999671300252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,12288,0.10265866915384929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,10240,0.08778844277064006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,3584,0.0427653325928582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,3072,0.06389600038528442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,3072,0.09161866373485988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,3072,0.03730222251680162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,8192,0.0710737771458096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,2560,0.08421422375573052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,2560,0.053732444842656456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,2560,0.0322364436255561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,2048,0.07767999834484525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,7168,0.07292888561884563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,2048,0.04388977752791511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,2048,0.028410666518741187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,6144,0.0555573304494222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,1536,0.07100977500279744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,1536,0.0340666671593984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,1536,0.022258667482270136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,1024,0.06293155749638875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,1024,0.0236799998415841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,1024,0.01788444485929277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,5120,0.0497920016447703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,768,0.031160000297758315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,768,0.018696000178654987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,768,0.01623644431432088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,4096,0.03837155633502536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,512,0.05624444617165459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,512,0.013880000346236758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,512,0.014930667148696052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,3584,0.04563911093605889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,256,0.023452444208992854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,3072,0.030919111437267725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,256,0.008753777378135258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,256,0.012999110751681857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,65536,128,0.01960000064637926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,128,0.006280000011126201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,128,0.010428444378905825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,2560,0.028413332170910303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,64,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,64,0.010303999814722273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,65536,32,0.005583999885453119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,65536,32,0.010406222608354356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,1536,0.023335110810067918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,2048,0.02750933335887061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,65536,0.3205386797587077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,65536,0.22537867228190103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,65536,0.18090489175584581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,1024,0.018069333500332303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,16384,0.0920942227045695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,16384,0.08602755599551731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,16384,0.05313511027230156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,12288,0.0766106645266215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,12288,0.06599555412928264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,12288,0.04245955414242215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,65536,0.17160444789462617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,10240,0.056573331356048584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,16384,0.04373777906099955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,10240,0.03688177797529433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,8192,0.05908089213901096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,8192,0.043042666382259794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,12288,0.03424088822470771
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,8192,0.03142755561404758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,7168,0.05384355452325609
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,7168,0.03715555535422431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,7168,0.02699111070897844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,10240,0.02930844492382473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,8192,0.024540444215138752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,6144,0.032607999112870954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,6144,0.024699555502997503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,7168,0.02643199927277035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,5120,0.044975999328825206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,5120,0.028019554085201685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,5120,0.0214675549003813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,768,0.015440000428093804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,10240,0.06731555859247844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,4096,0.040693332751592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,4096,0.023779556155204773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,4096,0.01816177699300978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,3584,0.03864533371395535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,3584,0.02051555613676707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,3584,0.016862221890025668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,512,0.01297688898113039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,6144,0.025285333395004272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,3072,0.018458666072951425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,3072,0.014910222755538093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,256,0.010048888623714447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,5120,0.018329777651362948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,2560,0.03392533461252848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,2560,0.015695111619101632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,65536,128,0.01034044474363327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,2560,0.013815110756291283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,2048,0.013223999904261695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,2048,0.03137333194414774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,2048,0.012090666426552666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,4096,0.01551555593808492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,1536,0.029512001408471003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,1536,0.010664000279373594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,1536,0.01072266697883606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,1024,0.026447110705905493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,1024,0.008359111017651027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,1024,0.00870311094654931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,3584,0.014436443646748861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,768,0.02023555503951179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,768,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,768,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,3072,0.013271110753218332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,512,0.02198133369286855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,512,0.004883555488453971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,512,0.006719111154476802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,6144,0.049913777245415576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,256,0.01719555589887831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,2560,0.014782221780882942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,256,0.003539555602603488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,256,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,128,0.014861333701345654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,2048,0.01277599980433782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,128,0.003148444410827425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,128,0.006280888699822956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,64,0.0028577778074476454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,1536,0.010384000009960597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,64,0.006284444282452266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,16384,32,0.0028355556229750314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,16384,32,0.006278222219811545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,1024,0.008672888908121321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,65536,0.24390755759345162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,768,0.007675555845101674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,16384,0.08173511425654094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,65536,0.15060355928209093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,16384,0.06673777765697904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,16384,0.046247111426459424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,12288,0.06917688581678602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,12288,0.051869332790374756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,12288,0.03587022092607286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,16384,3072,0.03642311029964023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,10240,0.05838933255937365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,10240,0.04389422138532003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,10240,0.03254844413863288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,512,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,8192,0.05071022113164266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,8192,0.038329776790406965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,8192,0.026467555099063452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,256,0.006286222073766921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,65536,0.14470488495296902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,7168,0.03509155578083462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,7168,0.045477333996031016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,16384,128,0.006588444527652528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,7168,0.02332088847955068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,16384,0.03950488898489211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,6144,0.0306693348619673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,6144,0.04294311006863912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,6144,0.020569778150982328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,5120,0.026689777771631878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,5120,0.03923911187383864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,5120,0.018975110517607797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,4096,0.03587288988961114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,4096,0.02236622240808275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,4096,0.015443555182880826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,12288,0.030435555511050757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,3584,0.033188445700539485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,3584,0.020607110526826646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,3584,0.014695111248228284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,10240,0.026425777210129633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,8192,0.022487110561794702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,3072,0.018202667435010273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,3072,0.013610666824711693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,7168,0.022072000635994807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,2560,0.029645333687464397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,6144,0.018086221483018663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,2560,0.01611288885275523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,2560,0.012479110724396177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,5120,0.016711999972661335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,2048,0.028311110205120508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,2048,0.013750221994188098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,2048,0.011065777805116443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,4096,0.01442488862408532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,1536,0.027170666389995154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,1536,0.008936000367005667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,1536,0.009713778065310584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,3584,0.016020443704393175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,1024,0.022653333014912073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,1024,0.007333333293596904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,1024,0.007321777443091075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,768,0.0188746667570538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,3072,0.01239111108912362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,768,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,768,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,512,0.019588443968031142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,512,0.0038515557017591265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,512,0.006329777754015393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,65536,0.1641591125064426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,2560,0.011671110987663269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,256,0.0032177778581778207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,256,0.01739288866519928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,256,0.006295111030340195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,2048,0.01055555542310079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,128,0.014826666977670459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,128,0.0028453332682450614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,128,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,64,0.002560888934466574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,64,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,12288,32,0.0028515555378463534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,12288,32,0.006239999913507038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,1536,0.009124444590674506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,65536,0.2055955595440335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,65536,0.14555733733707005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,1024,0.008192000289758047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,65536,0.14346933364868164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,16384,0.07811644342210558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,16384,0.058155556519826256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,16384,0.040594667196273804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,12288,0.0676177806324429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,12288,0.04494755466779073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,12288,0.031778666708204485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,768,0.007287110719415877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,10240,0.06085599793328179
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,512,0.006565333240562015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,10240,0.03875999980502658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,10240,0.02867377797762553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,256,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,8192,0.03279555506176419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,12288,128,0.0063582224150498705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,8192,0.023922666907310486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,7168,0.0465555555290646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,7168,0.029498666524887085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,7168,0.022620444496472675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,65536,0.13382221592797175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,6144,0.04105510976579454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,6144,0.026175111532211304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,16384,0.038331554995642766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,6144,0.01923111081123352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,5120,0.03640266590648227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,5120,0.022967999180157978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,5120,0.01719555589887831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,12288,0.029511110650168523
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,4096,0.03387555480003357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,4096,0.019513777560657926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,12288,3072,0.03244533472590976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,4096,0.014386667145623101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,3584,0.03222044308980306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,3584,0.017616000440385606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,3584,0.013432888521088494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,10240,0.02562933332390255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,3072,0.030785779158274334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,3072,0.016294222739007738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,3072,0.012085333466529846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,8192,0.02169688873820835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,2560,0.029492444462246362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,2560,0.014386667145623101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,2560,0.012097777591811286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,7168,0.019336000084877014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,6144,0.01772177716096242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,2048,0.010997333460383944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,2048,0.011031111081441244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,5120,0.017888888716697693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,1536,0.024381332927280005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,1536,0.009300444689061906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,1536,0.008551999926567078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,4096,0.013832888669437833
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,1024,0.021253334151373968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,1024,0.006559999866618051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,1024,0.007292444507280986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,3584,0.015048000547620984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,768,0.018922666708628338
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,768,0.004585777719815572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,768,0.006600889066855113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,3072,0.013568888935777875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,512,0.018590221802393597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,512,0.003546666767862108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,512,0.006216889040337668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,2560,0.011045333411958484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,256,0.017214222086800467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,256,0.00318755561278926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,256,0.005966222120655908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,2048,0.009700444837411245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,1536,0.00868800034125646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,128,0.014870221416155497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,128,0.002862222285734283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,128,0.0059573331640826324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,64,0.0028782222006056043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,1024,0.0074728892909155945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,64,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,10240,32,0.0028159999185138275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,10240,32,0.006235555642180973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,768,0.007170666423108842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,65536,0.17093155119154188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,65536,0.1303911076651679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,512,0.006269333263238271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,65536,0.09861510992050171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,16384,0.07413955529530843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,16384,0.047127111090554126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,16384,0.032582223415374756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,12288,0.03427733315361871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,12288,0.06272177563773261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,12288,0.02446222139729394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,10240,0.054471999406814575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,10240,0.028452446063359577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,256,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,10240,0.022665777140193518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,8192,0.04925955666436089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,8192,0.023618666662110224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,8192,0.018976888722843595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,10240,128,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,7168,0.040770666466818914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,7168,0.021203555994563635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,7168,0.016855110724767048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,65536,0.07809244261847602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,6144,0.03895466526349386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,16384,0.024359999431504145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,6144,0.018935999936527677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,6144,0.016135111451148987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,12288,0.019726221760114033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,5120,0.03559822175237868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,5120,0.016114667057991028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,5120,0.014699555105633207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,4096,0.03278133273124695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,4096,0.014135110709402295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,10240,0.0178195552693473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,4096,0.012139555480745105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,3584,0.03080000148879157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,3584,0.012290666500727335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,3584,0.011608888705571493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,8192,0.015107555521859063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,3072,0.02846755584081014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,3072,0.011178666518794166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,3072,0.011057777537239922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,7168,0.01384622189733717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,6144,0.01295199990272522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,2560,0.009755555954244401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,2560,0.01036177741156684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,5120,0.011712888876597086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,2048,0.023521777656343248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,2048,0.008316444853941599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,4096,0.010776888993051318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,2048,0.008275555239783393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,10240,2048,0.02703999976317088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,1536,0.022574222750133936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,1536,0.007309333317809635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,1536,0.007263999846246507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,1024,0.0196079992585712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,1024,0.005223999834722943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,1024,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,3584,0.009323555562231276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,768,0.018558222386572096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,768,0.003949333396222857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,768,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,3072,0.008657777474986183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,512,0.01721600029203627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,512,0.0035439998739295532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,512,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,2560,0.008231110870838165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,256,0.016869333055284288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,256,0.0032115555885765287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,2048,0.007648000286685095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,256,0.005584888988071018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,128,0.014942222171359591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,128,0.0028657778683635923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,1536,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,128,0.005575110928879843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,64,0.0028577778074476454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,64,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,8192,32,0.002528000002106031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,1024,0.006258666515350342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,8192,32,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,768,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,65536,0.1480684412850274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,65536,0.1249128845002916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,65536,0.09116444322797988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,16384,0.07646844122144911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,16384,0.042691555288102895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,16384,0.028267555766635474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,12288,0.06176888942718506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,512,0.00565155554148886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,12288,0.034105777740478516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,12288,0.023243554764323767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,256,0.005551110953092575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,10240,0.05156533254517449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,10240,0.029206222958034937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,8192,128,0.005548444473081165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,10240,0.02055288851261139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,8192,0.0455662210782369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,8192,0.025324443976084392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,8192,0.017440888616773818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,65536,0.07456711265775892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,7168,0.019322666856977675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,16384,0.023518222901556227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,7168,0.015784889459609985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,6144,0.03900622328122457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,6144,0.01683466633160909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,6144,0.015095111396577625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,12288,0.018907555275493197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,5120,0.03387466735310025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,5120,0.014862222803963555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,5120,0.013437333206335703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,8192,0.045369777414533824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,10240,0.016495111915800307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,4096,0.013401777380042605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,4096,0.01181777815024058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,8192,0.015032889114485847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,3584,0.029822223716311987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,3584,0.01144177797767851
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,3584,0.011355555719799466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,7168,0.013152889079517789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,3072,0.028375112348132666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,3072,0.010327999790509542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,3072,0.010310221877362993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,6144,0.012044444680213928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,2560,0.026327111654811438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,2560,0.009184888667530483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,2560,0.009671111073758867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,5120,0.011046222514576383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,2048,0.023769777682092454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,2048,0.008240888516108194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,2048,0.0075839997993575195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,4096,0.009762666291660732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,1536,0.021896888812383015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,1536,0.006960888703664144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,1536,0.0069368887278768755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,3584,0.008966222405433655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,1024,0.019289778338538278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,1024,0.004524444540341695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,1024,0.0063031112982167145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,3072,0.007687110867765214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,768,0.018265777164035372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,768,0.0038933331767717996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,2560,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,8192,2560,0.02642222245534261
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,768,0.0059075554211934405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,512,0.016736888223224215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,512,0.003378666730390655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,512,0.005597333527273602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,256,0.015965332587560017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,2048,0.0069617778062820435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,256,0.0031884445084465873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,256,0.005254222287072076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,128,0.014148443937301636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,128,0.0028320000403457214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,128,0.005201777650250329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,1536,0.006576889091067844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,64,0.0028284444577164124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,64,0.005583999885453119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,7168,32,0.002542222125662698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,1024,0.006048000107208888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,7168,32,0.005559999909665849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,65536,0.12464533911810981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,65536,0.11755822764502631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,65536,0.08197866545783149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,768,0.00573511090543535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,16384,0.03781422310405307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,16384,0.0632844434844123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,512,0.005597333527273602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,16384,0.025806221697065566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,12288,0.03012266755104065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,12288,0.0530008872350057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,256,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,12288,0.02124799953566657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,10240,0.04786044359207153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,10240,0.025978666212823655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,10240,0.019115555617544387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,7168,128,0.0052737775776121355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,8192,0.04225866662131416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,8192,0.022181333767043218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,8192,0.016495111915800307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,7168,0.039164443810780845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,7168,0.019903999235894945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,7168,0.014876445134480795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,65536,0.07175999879837036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,6144,0.03567733367284139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,6144,0.017909333109855652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,6144,0.013927999469969006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,16384,0.02251200046804216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,5120,0.03289066751797994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,5120,0.015840000576443143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,12288,0.01851288808716668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,5120,0.013056889176368713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,10240,0.015784889459609985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,4096,0.03127200073666043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,4096,0.011739555332395764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,8192,0.014331556028789945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,4096,0.01090666651725769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,3584,0.02869866622818841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,3584,0.01035733355416192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,3584,0.01036711119943195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,7168,0.012695111334323883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,3072,0.026456000076399908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,3072,0.009501333038012186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,3072,0.009677333136399588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,6144,0.011742221812407175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,2560,0.0243848893377516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,2560,0.00867022242810991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,2560,0.008392000363932716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,5120,0.01068800025516086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,2048,0.021772444248199463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,7168,0.04169866773817274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,2048,0.007253333098358578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,2048,0.006952889263629913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,1536,0.01995022263791826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,4096,0.008967111508051554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,1536,0.0063893331421746155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,1536,0.006562666760550604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,1024,0.017837334010336135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,1024,0.0041662222809261745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,1024,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,3584,0.007997333175606197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,768,0.018192888961897958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,768,0.003540444291300244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,768,0.005620444400442972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,512,0.016125332978036668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,512,0.0035048888789282907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,512,0.0052328887912962176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,3072,0.007294221884674496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,256,0.015840000576443143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,256,0.0031928889867332247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,256,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,2560,0.007256888680987888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,6144,128,0.014833778142929077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,128,0.0028311111446883944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,128,0.005277333160241445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,7168,4096,0.03118844495879279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,64,0.0029226665695508323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,64,0.0052062223354975385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,2048,0.006768888897365994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,6144,32,0.0028453332682450614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,6144,32,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,1536,0.006619555668698416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,65536,0.10672977897855972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,65536,0.1140693293677436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,1024,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,65536,0.0771626697646247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,768,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,16384,0.03267644511328803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,16384,0.06695022185643514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,16384,0.02367288867632548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,12288,0.027064889669418335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,12288,0.048877331945631236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,512,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,12288,0.01975644462638431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,10240,0.022826666633288067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,10240,0.043904887305365674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,10240,0.01793866687350803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,256,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,8192,0.020238222347365487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,8192,0.03849066628350152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,8192,0.015541333291265698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,7168,0.03747466537687514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,7168,0.018173333671357896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,6144,128,0.005222222043408289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,7168,0.014108444253603617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,6144,0.03251288996802436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,6144,0.016385777129067313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,6144,0.013423110875818463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,5120,0.029873778422673542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,5120,0.014489778214030795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,5120,0.01219288922018475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,65536,0.06953333483801948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,4096,0.027323555615213182
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,16384,0.021872000561820135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,4096,0.011180444724029966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,4096,0.010499555203649731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,3584,0.026428444517983332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,12288,0.017892445127169292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,3584,0.009656888743241629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,3584,0.009667555491129557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,10240,0.015431111057599386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,3072,0.008811555802822113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,3072,0.02534488836924235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,3072,0.00853777759604984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,8192,0.013945778210957846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,2560,0.0229777776532703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,2560,0.007923555870850882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,2560,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,2048,0.020964443683624268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,7168,0.012442666623327466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,2048,0.006846222198671765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,2048,0.0069715554515520734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,6144,0.011014222270912595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,1536,0.0191920002301534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,5120,0.009526222116417354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,1536,0.005244444641802046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,1536,0.00656088896923595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,1024,0.017154667112562392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,1024,0.003957333250178231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,1024,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,4096,0.00813422186507119
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,768,0.017866666118303936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,3584,0.007617777420414819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,768,0.0034862222770849862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,768,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,512,0.015640000502268474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,3072,0.007336888876226213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,512,0.0032346666687064697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,512,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,2560,0.007189333438873291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,256,0.01482222146458096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,256,0.003030222323205736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,256,0.005292444593376583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,2048,0.006565333240562015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,128,0.0028453332682450614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,128,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,1536,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,64,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,64,0.005187555733654234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,1024,0.0059075554211934405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,5120,32,0.0028515555378463534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,5120,32,0.005219555563396878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,65536,0.09123644563886855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,768,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,65536,0.07089689042833117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,512,0.005304000029961268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,16384,0.027719999353090923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,16384,0.02209866709179348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,16384,0.06634133391910128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,256,0.005271111097600725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,12288,0.021201777789327834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,12288,0.05111733410093519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,12288,0.01882844501071506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,10240,0.018739556272824604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,5120,128,0.0052764444715446895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,10240,0.01686044368478987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,8192,0.04122222132152981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,8192,0.016689777374267578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,8192,0.015185778339703878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,65536,0.06712622112698026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,7168,0.03901066713862949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,16384,0.021180444293551978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,7168,0.014871110518773397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,7168,0.01350933313369751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,12288,0.017332444588343304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,6144,0.033879998657438494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,6144,0.013735111388895245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,6144,0.012629333469602795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,10240,0.014869333969222175
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,5120,0.030792001220915053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,5120,0.01236533373594284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,5120,0.011679111255539788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,8192,0.013455111119482251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,4096,0.011001778145631155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,4096,0.00944622192117903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,7168,0.011707555916574268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,3584,0.026547556122144062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,6144,0.010037333601050908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,3584,0.011723555624485016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,3584,0.00869333330127928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,3072,0.024345777100986902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,5120,0.008692444198661381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,3072,0.01054844425784217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,3072,0.007305777735180325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,2560,0.02163644466135237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,2560,0.009678222239017487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,2560,0.007264888948864407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,4096,0.008006222546100616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,2048,0.020620443754725985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,2048,0.008288000192907121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,2048,0.0069644442862934535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,3584,0.007612444460391998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,1536,0.01962311069170634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,1536,0.006535999890830781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,3072,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,1536,0.006260444306664997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,1024,0.017492444978819955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,1024,0.004928888960017098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,1024,0.005965333431959152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,2560,0.0069644442862934535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,768,0.01788444485929277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,768,0.004568888909286923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,768,0.005721777677536011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,2048,0.006631111105283101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,512,0.015583111180199517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,512,0.003916444463862313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,512,0.005628444254398346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,1536,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,256,0.015504888362354703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,256,0.0029519999192820657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,256,0.005564444594913059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,1024,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,768,0.005557333429654439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,128,0.014488889111412896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,128,0.0028684443483750024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,128,0.005239999956554837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,512,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,64,0.0025928889711697898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,64,0.00544088871942626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,4096,32,0.002513777878549364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,4096,32,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,256,0.005225777626037598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,65536,0.08835733599132961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,65536,0.08091733190748426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,65536,0.068031112353007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,16384,0.06500977940029569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,4096,128,0.005261333452330695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,16384,0.02555199960867564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,16384,0.021348444951905146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,12288,0.04832355512513054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,12288,0.019840000404251944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,12288,0.01796355512407091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,10240,0.04549511273701986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,65536,0.06645600001017253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,10240,0.017318222257826064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,10240,0.016351110405392118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,8192,0.04059911105367873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,16384,0.020953777763578627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,8192,0.015006222658687167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,8192,0.014671110444598727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,7168,0.03701600101259019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,7168,0.014021333720948962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,7168,0.013110222087966071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,12288,0.016873778568373788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,6144,0.03352533446417914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,6144,0.012768000364303589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,6144,0.012404444317022959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,10240,0.014892444014549255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,5120,0.03150399857097202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,5120,0.011608888705571493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,5120,0.011376889215575324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,8192,0.013025778035322825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,4096,0.02747022277779049
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,4096,0.01036711119943195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,4096,0.00833511104186376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,5120,128,0.014105778601434497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,7168,0.01073155552148819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,3584,0.011358222199810876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,65536,0.09093066718843247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,3584,0.025063110722435847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,3584,0.00817599975400501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,3072,0.02275200022591485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,3072,0.01036711119943195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,6144,0.009360888765917884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,3072,0.0073315559162033936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,2560,0.020959110723601446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,2560,0.008332444561852349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,2560,0.007300444775157505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,2048,0.019239111079110038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,5120,0.00869333330127928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,2048,0.007237333390447829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,2048,0.006612444503439798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,1536,0.019309333629078336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,1536,0.006228444476922353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,1536,0.006298666612969504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,4096,0.007954667011896769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,10240,0.04533777634302775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,1024,0.017159110969967313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,1024,0.004592000196377437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,1024,0.005882666756709416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,3584,0.007644444704055786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,768,0.01824711097611321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,768,0.004202666795916027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,768,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,3072,0.007325333025720384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,512,0.01512977812025282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,512,0.003903999924659729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,512,0.0055440002017551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,2560,0.006918222126033571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,256,0.015468445089128284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,256,0.0031946665710873077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,256,0.0052168890833854675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,2048,0.006311111152172089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3584,128,0.014816888504558139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,128,0.0028115556471877625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,128,0.005460444423887465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,64,0.0025253333151340485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,1536,0.006256000035338932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,64,0.0052684446175893145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3584,32,0.002504000026318762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3584,32,0.0052782222628593445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,4096,4096,0.028103109863069322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,1024,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,65536,0.06930310858620538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,65536,0.08674222230911255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,16384,0.02156977852185567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,768,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,16384,0.05455199877421061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,65536,0.06543644269307454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,16384,0.02102577851878272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,512,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,12288,0.045965333779652916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,12288,0.01758577757411533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,12288,0.01755199995305803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,256,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,10240,0.04546844296985202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,10240,0.01570577753914727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3584,128,0.0052284445199701525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,10240,0.015856888559129503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,8192,0.036679112248950534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,8192,0.013759111364682516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,8192,0.013907555076811047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,7168,0.03494844502872891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,7168,0.013125333521101209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,65536,0.06538666619194879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,7168,0.01275199982855055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,16384,0.020654223031467862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,6144,0.03346933258904351
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,6144,0.011876444849703046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,6144,0.011909333368142446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,5120,0.02884711159600152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,5120,0.010899555351999072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,12288,0.016879111528396606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,5120,0.0103493332862854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,4096,0.009780444204807281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,4096,0.025352888637118872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,10240,0.01442399952146742
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,4096,0.007990222010347579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,8192,0.011726222104496427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,3584,0.024329778220918443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,3584,0.009273777405420939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,3584,0.007962666451931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,3072,0.009298666483826106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,3072,0.022276444567574397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,3072,0.0076044441925154785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,7168,0.010547555155224271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,2560,0.007660444411966536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,2560,0.02029244436158074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,2560,0.007287999822033777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,6144,0.009259555902745988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,2048,0.006657777975002925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,2048,0.0188719994492001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,2048,0.006625777731339137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,5120,0.008603555460770925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,1536,0.018187556001875136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,1536,0.006468444648716185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,1536,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,1024,0.01648977730009291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,1024,0.004912000149488449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,4096,0.007947555846638149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,1024,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,3584,0.007664889097213745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,768,0.01752800080511305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,768,0.004368888835112254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,768,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,512,0.004222222086456087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,512,0.014474666780895658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,3072,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,512,0.00553688903649648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,256,0.014814222852389017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,256,0.003491555651028951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,2560,0.0069671107663048645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,256,0.005241777747869492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,2048,0.006294222341643439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,128,0.0032151111712058387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,3072,128,0.014820444915029736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,1536,0.006251555350091722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,128,0.005230222311284807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,64,0.002540444541308615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,64,0.00554666668176651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,3072,32,0.0028124445428450904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,3072,32,0.005249777601824866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,1024,0.005872000008821487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,65536,0.059451553556654185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,768,0.005577777822812398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,65536,0.08495110935635036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,16384,0.019446222318543326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,65536,0.06275111436843872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,16384,0.05922489033805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,512,0.0052764444715446895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,16384,0.02056711084312863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,12288,0.04383377896414863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,12288,0.016420443852742512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,256,0.0052151112920708126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,12288,0.017204445269372728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,3072,128,0.0052355556852287715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,10240,0.013980444934633044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,10240,0.015279110934999255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,8192,0.03636799918280707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,8192,0.01276622215906779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,65536,0.06466488705741034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,8192,0.013667555318938361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,16384,0.020397333635224234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,7168,0.03405066662364536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,7168,0.012431999875439538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,7168,0.011705777711338468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,6144,0.010999110837777456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,6144,0.030856887499491375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,12288,0.016547555724779766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,6144,0.011352889239788055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,5120,0.027449778384632532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,5120,0.010013333625263637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,5120,0.009649777577983009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,10240,0.013230221966902414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,4096,0.023020444644822016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,4096,0.009300444689061906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,4096,0.007988444632954067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,8192,0.011067555182509951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,3584,0.02127555509408315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,3584,0.008928000099129146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,3584,0.00776622196038564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,7168,0.009827555881606208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,3072,0.020626667473051283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,3072,0.00794933322403166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,3072,0.0074133334888352295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,6144,0.009325332939624786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,2560,0.0069022224181228215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,2560,0.019374221563339233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,2560,0.006952889263629913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,5120,0.00867022242810991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,2048,0.006221333311663733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,2048,0.018574222922325134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,2048,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,4096,0.007631999750932057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,1536,0.01717155509524875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,1536,0.005618666609128316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,1536,0.006298666612969504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,3584,0.007299555672539606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,1024,0.004568888909286923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,3072,0.007261333366235097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,1024,0.005938666562239329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,2560,0.0069439998931354946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,768,0.0042364444169733255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,768,0.005600000007285013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,2048,0.0064444442590077715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,512,0.013734222286277346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,512,0.0038382220599386427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,1536,0.006231110956933763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,512,0.005208888815508948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,256,0.01442488862408532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,256,0.0031973332580592898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,1024,0.0058906666106647914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,256,0.005200888961553574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,128,0.013767110804716745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,128,0.0031822222388452957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,128,0.00526311124364535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,768,0.005587555468082428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,64,0.0025439999169773525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,64,0.005579555614127054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2560,32,0.002801777794957161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,512,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2560,32,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,65536,0.051727109485202365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,256,0.005219555563396878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,65536,0.06114844481150309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,65536,0.07772799995210436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,16384,0.015944000747468736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,16384,0.06054311328464084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,16384,0.019926221834288705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2560,128,0.0052062223354975385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,12288,0.013220444321632385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,12288,0.05089066757096184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,12288,0.016711110870043438
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,10240,0.01202222208182017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,10240,0.043508443567487926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,10240,0.015320888823933072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,65536,0.06361066632800631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,8192,0.03808710972468058
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,8192,0.011239999698268043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,8192,0.012993777791659037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,16384,0.02001066671477424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,7168,0.03363555669784546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,7168,0.010653333531485664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,7168,0.011022222538789114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,12288,0.01420088940196567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,6144,0.03025955624050564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,6144,0.009668444593747457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,6144,0.009986666341622671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,10240,0.012074666718641916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,5120,0.026769777139027912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,5120,0.00903555585278405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,5120,0.009010666774378883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,8192,0.010675555302037133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,4096,0.0248364441924625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,4096,0.010012444522645738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,4096,0.008004444340864817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,7168,0.009682666924264696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,3584,0.022982221510675218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,3584,0.009343110852771336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,3584,0.007684444387753804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,3072,0.021879111727078755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,6144,0.00903200027015474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,3072,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,3072,0.007284444239404466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,5120,0.008339555727110969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,2560,0.020615999897321064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,2560,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,2560,0.006966222491529252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,4096,0.007643555601437886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,2048,0.01882222294807434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,2048,0.006959999601046245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,2048,0.006592000110281839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,3584,0.007585778004593319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,1536,0.018576888574494254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,1536,0.0058666666348775225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,3072,0.007244444555706448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,1536,0.006269333263238271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,1024,0.016164445214801364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,2560,0.0069271110826068455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,1024,0.004569777597983678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,1024,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,2048,0.006322666588756774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,768,0.018193778064515855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,768,0.004584889031118817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,768,0.005570666657553778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,512,0.0144533332851198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,1536,0.006227555374304454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,512,0.003855111284388436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,512,0.0052675555149714155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,256,0.0147724449634552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,256,0.0031537777847713898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,1024,0.0058711109062035876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,256,0.0052764444715446895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2048,128,0.013792888985739814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,128,0.003149333306484752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,128,0.0052391112678580815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,768,0.00544799988468488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,64,0.002854222224818336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,64,0.0052666668262746595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,512,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,2048,32,0.0028417776856157514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,2048,32,0.005388444496525659
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,65536,0.037920888927247785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,256,0.005211555709441503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,65536,0.0767422252231174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,65536,0.05946933560901218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,16384,0.05983199675877889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,16384,0.013735111388895245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,2048,128,0.005208000126812193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,16384,0.019536000159051683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,12288,0.048439999421437584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,12288,0.01146399974822998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,12288,0.016308445069524977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,10240,0.04307644565900167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,65536,0.063573333952162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,10240,0.010318222145239512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,10240,0.01440000037352244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,8192,0.01036622209681405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,8192,0.011076444553004371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,16384,0.017845332622528076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,7168,0.032411555449167885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,7168,0.009813333551088968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,12288,0.013593778014183044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,7168,0.010041777458455827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,10240,0.012044444680213928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,6144,0.027797334724002417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,6144,0.008943999807039896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,6144,0.009351111120647853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,8192,0.010665777656767102
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,5120,0.02535022298494975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,5120,0.007717333734035492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,5120,0.008980444735950893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,4096,0.02370133333735996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,7168,0.009720000127951304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,4096,0.007589333587222629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,4096,0.008182221816645728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,3584,0.022778666681713525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,6144,0.009071110851234859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,3584,0.007316444483068254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,3584,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,5120,0.008344888687133789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,3072,0.020594666401545208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,3072,0.006241777704821692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,3072,0.007611555357774098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,2560,0.01955111159218682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,4096,0.00794400026400884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,2560,0.005883555445406172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,2560,0.006909333169460297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,2048,0.018231999542978074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,3584,0.007293333609898885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,2048,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,2048,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,1536,0.017535110314687092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,3072,0.007276444799370236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,1536,0.0052720002002186244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,1536,0.005933333188295364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,1024,0.015463110473420886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,2560,0.006912888752089606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,1024,0.004230222354332606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,1024,0.005922666854328579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,768,0.016970665918456185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,2048,0.006409777949253718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,768,0.0038808890514903595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,768,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,512,0.013763555222087435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,1536,0.006231110956933763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,512,0.003532444437344869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,512,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,256,0.014498665928840637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,1024,0.005649777750174205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,256,0.0032071111102898917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,256,0.005377777748637729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,768,0.005587555468082428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,128,0.014454222387737699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,128,0.0028400001012616684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,10240,0.03758755657407973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,512,0.00526577772365676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,128,0.005244444641802046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,64,0.0028275555620590844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,64,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1536,32,0.002529777793420686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1536,32,0.0052373334765434265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,256,0.0052151112920708126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,65536,0.027082666754722595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,65536,0.07619466384251912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,65536,0.057413332992129855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,16384,0.01516177753607432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,16384,0.04470666580730014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1536,128,0.0052737775776121355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,16384,0.018339556124475267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,12288,0.012747555143303342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,12288,0.013905778527259827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,65536,0.06295822064081828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,10240,0.03230399886767069
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,16384,0.01649688846535153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,10240,0.011708444191349877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,10240,0.012106666962305704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,8192,0.02869422237078349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,12288,0.013411555853154926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,8192,0.010671999719407825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,8192,0.010664000279373594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,7168,0.009715555442704095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,7168,0.025342222717073228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,7168,0.009992000129487779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,10240,0.012034667034943899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,1024,0.015508444772826301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,6144,0.008648888932334052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,6144,0.023990222149425085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,6144,0.009698666632175446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,8192,0.010664888554149203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,5120,0.00795555528667238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,5120,0.00889066689544254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,7168,0.009675555759006077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,4096,0.021203555994563635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,4096,0.007297777467303806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,6144,0.009016888837019602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,4096,0.007739555504586961
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,3584,0.01957244508796268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,3584,0.0069546666410234236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,3584,0.007647111184067196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,5120,0.008628444539176093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,3072,0.019221333993805777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,3072,0.006246222390068903
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,3072,0.007296889192528195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,4096,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,2560,768,0.017551110850440133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,2560,0.017878222796652053
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,3584,0.00737955586777793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,2560,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,2560,0.006976889239417181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,2048,0.01681422193845113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,2048,0.007249777515729268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,3072,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,2048,0.006351111249791251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,2560,0.006897777732875612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,1536,0.015116444892353482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,1536,0.00592533333433999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,1536,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,1024,0.013791999883121915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,1024,0.004282666577233208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,2048,0.006312888943486744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,1024,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,768,0.016452444924248587
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,768,0.0041991112132867174
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,1536,0.006217777729034424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,768,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,512,0.013071999781661563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,512,0.005404444618357553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,512,0.003839111162556542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,1024,0.005537777725193236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,256,0.0148062225845125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,256,0.0032044444233179092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,768,0.005565333283609814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,256,0.0052977779673205475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,128,0.014474666780895658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,128,0.002826666666401757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,512,0.005383111122581694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,64,0.002801777794957161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,128,0.005203555441564984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,256,0.004922666483455234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,64,0.005223999834722943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,1024,32,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,1024,32,0.0052471111218134565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,1024,128,0.005199111170238919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,65536,0.02235377828280131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,65536,0.07547110981411405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,65536,0.05715555614895291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,16384,0.03418133325046963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,16384,0.010508444574144153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,16384,0.015826667348543804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,65536,0.062216891182793506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,12288,0.028805332051383123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,12288,0.011354666617181567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,12288,0.01332444449265798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,16384,0.01649599936273363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,10240,0.025369778275489807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,10240,0.009787555370065901
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,10240,0.012183999849690331
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,12288,0.013519110778967539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,8192,0.009332444104883406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,8192,0.01070844464831882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,7168,0.021976000732845728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,10240,0.012089778151777057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,7168,0.008932444784376357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,7168,0.010047111246320937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,8192,0.010823999842007955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,6144,0.020580444071027968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,6144,0.008005333443482717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,6144,0.009365333451165093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,7168,0.009679111341635386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,5120,0.01904444396495819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,5120,0.006949333681000604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,5120,0.008747555315494537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,6144,0.009304888546466827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,4096,0.01796000036928389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,4096,0.007268444531493717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,4096,0.00794577764140235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,5120,0.008329778081840938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,3584,0.007080888582600488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,3584,0.007631111476156447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,4096,0.007931555310885111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,3072,0.016453334026866488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,3072,0.006963555183675554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,3584,0.007575111256705389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,3072,0.007261333366235097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,2560,0.015863110621770222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,2560,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,3072,0.007246221933099959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,2560,0.007247111035717859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,2048,0.015115555789735583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,2048,0.005203555441564984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,2048,0.006596444381607904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,2560,0.006675555474228329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,1536,0.014475555883513557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,1536,0.004941333499219683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,1536,0.006256888724035687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,1024,0.013450666434235044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,2048,0.00657155571712388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,1024,0.004190222256713443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,1024,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,1536,0.0058453331391016645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,768,0.015801777442296345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,768,0.0035457776652442086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,768,0.0058746664888328975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,1024,0.00592533333433999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,512,0.013071999781661563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,512,0.0034942221310403612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,512,0.005565333283609814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,256,0.014820444915029736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,768,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,256,0.003162666741344664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,256,0.005197333378924264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,512,0.0053173332578606075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,128,0.01311644415060679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,128,0.0028648889727062653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,128,0.0052062223354975385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,256,0.005222222043408289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,64,0.002526222210791376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,64,0.0052720002002186244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,768,32,0.0028320000403457214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,768,32,0.005569777968857024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1536,8192,0.03836089041497972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,65536,0.016819554898473952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,768,128,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,65536,0.07484622134102716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,65536,0.056112892097897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,16384,0.026871111657884385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,16384,0.010943111446168689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,16384,0.015848888291252982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,65536,0.05477422144677904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,12288,0.008641777767075432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,12288,0.02813777658674452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,12288,0.013395555317401886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,16384,0.01651822196112739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,10240,0.02567200031545427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,10240,0.007802666889296637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,10240,0.012124444047609964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,8192,0.008010666403505538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,8192,0.019332443674405415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,12288,0.013415999710559845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,8192,0.010476444330480365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,10240,0.011483555866612328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,7168,0.02148444453875224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,7168,0.00792622235086229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,7168,0.009883555273214975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,8192,0.010729778144094678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,6144,0.0069208890199661255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,6144,0.017551110850440133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,6144,0.009672000176376766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,7168,0.009415110780133141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,5120,0.017144888639450073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,5120,0.006631111105283101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,5120,0.00867377801073922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,6144,0.008741333252853816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,4096,0.017834666702482436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,4096,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,5120,0.008472888833946651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,4096,0.007652444144090016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,3584,0.015463110473420886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,3584,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,3584,0.00794933322403166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,4096,0.007644444704055786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,3072,0.014775110615624322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,3072,0.006513777706358168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,3584,0.007271111011505127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,3072,0.007643555601437886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,2560,0.01550044450494978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,2560,0.00600355573826366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,2560,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,3072,0.0069102222720781965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,2048,0.005888888819350137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,2048,0.01479555500878228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,2048,0.0063822223908371395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,2560,0.0069191112286514705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,1536,0.00453955555955569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,1536,0.006253333141406377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,2048,0.006302222195598815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,1024,0.012749333348539142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,1024,0.0038515557017591265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,1536,0.006257777826653586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,1024,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,768,0.014785778191354541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,768,0.003565333369705412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,1024,0.005564444594913059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,768,0.005487999982304043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,768,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,512,0.0035111109415690103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,512,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,512,0.005277333160241445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,256,0.014107555150985718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,256,0.002891555635465516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,256,0.0052737775776121355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,12288,0.03664888938268026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,128,0.013282666603724161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,128,0.002833777831660377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,256,0.005249777601824866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,128,0.0052737775776121355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,64,0.002833777831660377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,64,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,512,32,0.0024888888001441956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,512,32,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,512,128,0.005219555563396878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,65536,0.016629333297411602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,65536,0.0723173353407118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,65536,0.049676444795396596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,16384,0.02626666592227088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,16384,0.008953777452309927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,65536,0.051348444488313466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,16384,0.015666666958067153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,12288,0.02270577847957611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,12288,0.008264888491895463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,1024,5120,0.021730666359265644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,12288,0.013430222041077085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,16384,0.01651022169325087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,10240,0.007315555380450354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,10240,0.020976000361972388
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,10240,0.012083555261294046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,12288,0.013468444347381592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,8192,0.01924800044960446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,8192,0.006959111326270633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,10240,0.011735111474990845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,8192,0.010729778144094678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,7168,0.018031110366185505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,7168,0.010198222266303169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,8192,0.010409778190983666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,7168,0.006633777585294511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,6144,0.01688088807794783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,7168,0.009824000298976898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,6144,0.006292444550328785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,6144,0.009691555466916826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,5120,0.016590222716331482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,6144,0.009076444639099969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,5120,0.006280000011126201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,5120,0.009013333254390294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,4096,0.01516266663869222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,5120,0.00851466672288047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,4096,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,4096,0.008103110724025303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,3584,0.014775110615624322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,3584,0.005950222412745158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,4096,0.00795733349190818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,3584,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,3072,0.014121777481502958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,3072,0.005625777774386936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,3584,0.007439999944633908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,3072,0.0074888889988263445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,2560,0.014089778065681458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,2560,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,3072,0.006934222247865465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,2560,0.007088888850477006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,2560,0.006934222247865465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,2048,0.013029333617952136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,2048,0.0058373332851462895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,2048,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,1536,0.012160888976520963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,2048,0.006272888845867581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,1536,0.004264889078007804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,1536,0.005963555640644497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,1536,0.006285333385070165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,1024,0.0038728887836138406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,1024,0.0058711109062035876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,768,0.01480977733929952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,1024,0.005617777920431561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,768,0.003489777859714296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,768,0.00534400012758043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,512,0.011046222514576383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,768,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,512,0.0032053333189752367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,512,0.005305777821275923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,512,0.005241777747869492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,256,0.0028924445311228433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,256,0.005261333452330695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,8192,0.023687111006842718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,128,0.010453333457310995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,256,0.004925333377387789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,128,0.0028728888266616394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,128,0.0052471111218134565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,64,0.003028444531891081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,256,128,0.00490044429898262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,64,0.004937777916590373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,256,32,0.0028115556471877625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,256,32,0.004918222212129169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,65536,0.07082577546437581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,65536,0.014105778601434497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,768,3584,0.01751733322938283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,65536,0.04725955592261421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,16384,0.0262444449795617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,65536,0.0499582224422031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,16384,0.007849777738253275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,16384,0.015847111741701763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,12288,0.021544888615608215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,16384,0.016187555260128446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,12288,0.007275555696752336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,12288,0.013095999757448832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,10240,0.020255999432669747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,12288,0.013409777647919126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,10240,0.007128000259399414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,10240,0.012070222033394707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,8192,0.006629333313968446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,8192,0.01789777808719211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,10240,0.011763555308183035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,8192,0.010684444672531553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,7168,0.016605334149466622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,8192,0.010384000009960597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,7168,0.0064026667839950984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,7168,0.010040889183680216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,6144,0.006224888894293044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,7168,0.009646221995353699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,6144,0.015844444433848064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,6144,0.009654222263230218
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,6144,0.00902844468752543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,5120,0.014908444550302295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,5120,0.005920889063013925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,5120,0.008672888908121321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,5120,0.008347555167145198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,4096,0.014130666851997375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,4096,0.005601777798599667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,4096,0.007948444121413762
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,3584,0.013429332938459186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,4096,0.007974222302436829
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,3584,0.00629688882165485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,3584,0.00759733302725686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,3584,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,3072,0.013763555222087435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,3072,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,3072,0.007325333025720384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,3072,0.006959999601046245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,2560,0.005595555735958948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,2560,0.013811555173661975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,2560,0.0069306666652361555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,2560,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,2048,0.012304888831244575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,2048,0.005896889087226655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,2048,0.006879999819729064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,2048,0.005927111125654644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,1536,0.004517333375083076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,1536,0.012129777835475074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,1536,0.005964444329341252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,1536,0.00620888877246115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,1024,0.012078222301271228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,1024,0.003540444291300244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,1024,0.005912000106440649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,1024,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,768,0.005612444546487596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,768,0.013415111435784234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,768,0.003306666596068276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,512,0.011710222396585675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,512,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,768,0.005316444569163852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,512,0.0031840000301599503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,256,0.010386666489972008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,512,0.005398222141795688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,256,0.0028773333049482773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,256,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,256,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,128,128,0.010724444356229572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,128,0.0028159999185138275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,128,0.005145777844720417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,64,0.002529777793420686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,4,128,128,0.004911111046870549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,64,0.005208888815508948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,128,32,0.0024871110088295406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,128,32,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,65536,0.010716444088353051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,16384,0.006623999940024481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,65536,0.046872887346479625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,16384,0.015796444482273526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,12288,0.005961777849329843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,12288,0.013403555585278405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,10240,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,10240,0.011911999848153857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,8192,0.006238222122192383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,8192,0.010680889089902243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,7168,0.0059368887709246735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,7168,0.009967111051082611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,6144,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,1536,0.013808888693650564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,6144,0.009023110899660323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,4096,0.005295111073387994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,5120,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,5120,0.00869333330127928
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,4096,0.007654222349325816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,3584,0.005544888890451855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,3584,0.007653333246707916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,3072,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,3072,0.007311999797821045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,2560,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,2560,0.0069306666652361555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,2048,0.0059057776298787855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,2048,0.006265777680608962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,1536,0.0041893331540955436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,1536,0.006253333141406377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,768,0.0034835553831524322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,1024,0.0038524443904558816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,512,512,0.01204355557759603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,1024,0.005311111195219888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,768,0.005562666803598404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,512,0.003154666680428717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,512,0.005366222312053044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,256,0.0028275555620590844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,256,0.005283555636803309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,128,0.0028231110837724474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,32,0.002516444358560774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,128,0.004905777672926585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,32,0.005202666752868229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,64,64,0.0024826667375034755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,64,64,0.004873777843183941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,65536,0.008671999805503422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,16384,0.006239999913507038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,65536,0.046794666184319385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,16384,0.015683554940753512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,12288,0.00647555540005366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,12288,0.013417777915795645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,10240,0.005966222120655908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,7168,0.010009778042634329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,10240,0.011693333586057028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,8192,0.0052266667286554975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,8192,0.010703110860453712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,7168,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,4096,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,6144,0.005947555518812603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,6144,0.009308444129096137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,5120,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,5120,0.00868266655339135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,4096,0.007963555554548899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,3584,0.005903111149867375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,2560,0.0069368887278768755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,3584,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,3072,0.00519466648499171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,3072,0.0069884442620807225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,2560,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,2048,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,2048,0.006569777925809224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,1536,0.004226666771703296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,1536,0.006237333433495627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,1024,0.003867555409669876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,1024,0.005600000007285013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,768,0.003552888830502828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,768,0.00555377784702513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,512,0.0032044444233179092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,512,0.004911111046870549
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,256,0.0028151110228565005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,256,0.004913777940803104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,128,0.0028275555620590844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,128,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,64,0.0025235555238193935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,64,0.00510044437315729
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,4,32,32,0.002464888824356927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,4,32,32,0.0052666668262746595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,16384,0.3373564349280463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,12288,0.2554071214463976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,16384,0.17763200071122912
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,16384,0.26027290026346844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,12288,0.21192088392045763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,12288,0.13270310560862222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,10240,0.21390133433871797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,256,0.012437333663304647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,4,256,1024,0.01238933371173011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,10240,0.112519105275472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,8192,0.17596266004774305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,8192,0.1761528915829129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,8192,0.09609066777759129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,7168,0.145797332127889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,7168,0.1532471047507392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,7168,0.0816355546315511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,6144,0.1269342237048679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,6144,0.13056355052524143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,6144,0.06770488950941297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,5120,0.11375199423895942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,5120,0.10883377657996283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,5120,0.05787999762429131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,10240,0.18145510885450575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,4096,0.10795911153157552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,4096,0.09053333600362141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,4096,0.04768266611629062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,12288,0.1171262264251709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,16384,0.13584089279174805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,3584,0.07287822167078654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,3584,0.042908443344963916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,10240,0.08613155285517375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,3072,0.08944444523917304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,3072,0.06311021910773383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,8192,0.07165333297517565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,3072,0.03745422098371718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,2560,0.08187555604510836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,2560,0.05277155505286323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,7168,0.06449066930347018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,2560,0.0328453348742591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,6144,0.05954222546683418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,2048,0.04331288735071818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,2048,0.027772444817754958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,5120,0.05547822184032864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,1536,0.06995289193259345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,1536,0.0337137778600057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,1536,0.022188444932301838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,3584,0.10043199857076009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,1024,0.06303111049864027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,1024,0.02420444455411699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,1024,0.017856889300876193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,4096,0.04825866553518507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,768,0.030124445756276447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,768,0.018930666976504855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,2048,0.07692977454927233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,768,0.016376889414257474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,512,0.0557804438802931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,512,0.013812444276279874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,512,0.014560888210932413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,3584,0.045781334241231285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,3072,0.03902933332655165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,256,0.008965333302815756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,256,0.013280889226330651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,2560,0.03387377659479777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,128,0.019316444794336956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,128,0.006511999915043513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,128,0.010354667074150508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,2048,0.027544000082545813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,64,0.005903111149867375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,64,0.010325333310498131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,65536,32,0.005577777822812398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,65536,32,0.010049777726332346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,1536,0.023436443673239812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,65536,0.31799999872843426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,65536,0.2197751071718004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,65536,0.17765422662099203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,16384,0.09187377823723687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,16384,0.08516977892981635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,16384,0.05486577749252319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,65536,0.15840444299909803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,12288,0.07794221904542711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,12288,0.06580711073345609
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,12288,0.041308446062935725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,16384,0.05354666709899902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,10240,0.0660239987903171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,65536,256,0.02348088887002733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,10240,0.0566177765528361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,1024,0.017907554904619854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,10240,0.03605511122279697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,8192,0.04445599847369724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,8192,0.058712886439429395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,8192,0.030177778667873804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,7168,0.05934933159086439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,7168,0.03831733266512553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,7168,0.027070222629441157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,768,0.015146666102939181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,6144,0.0491795539855957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,6144,0.03328533305062188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,512,0.012997333374288348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,6144,0.024263999528355066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,5120,0.04405688908365038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,5120,0.029003553920321997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,5120,0.0214346663819419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,12288,0.03401066528426276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,10240,0.03407466742727492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,8192,0.033215999603271484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,4096,0.024719999896155462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,4096,0.017503110898865592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,7168,0.030046221282747056
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,3584,0.02094399929046631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,3584,0.01644888851377699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,6144,0.022089777721299067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,3072,0.03645955522855123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,3072,0.018931556079122756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,3072,0.01516711049609714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,5120,0.020849777592553034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,128,0.010016889207892949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,65536,256,0.010399999717871347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,2560,0.033420443534851074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,2560,0.016156444946924847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,2560,0.013774221969975365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,2048,0.013437333206335703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,2048,0.031256887647840716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,2048,0.012519111235936483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,1536,0.029612445169025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,1536,0.01092711091041565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,1536,0.01071911139620675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,4096,0.020223110914230347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,1024,0.02721244427892897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,1024,0.008359111017651027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,1024,0.008856000171767341
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,3584,0.018395556343926322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,768,0.019253333409627277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,768,0.00720177756415473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,768,0.0069351109365622205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,3072,0.016525333126386006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,512,0.02233422299226125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,512,0.005063111169470681
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,512,0.006346666564544042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,2560,0.014485332700941296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,256,0.017498667041460674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,2048,0.013008889224794177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,256,0.003519999898142285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,256,0.00629066675901413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,128,0.015454222758611044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,128,0.003168888803985384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,128,0.006140444427728653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,1536,0.01072355525361167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,64,0.0028844444702068963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,64,0.005948444621430502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,16384,32,0.0029066666546795103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,16384,32,0.006284444282452266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,1024,0.008672888908121321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,65536,0.24386400646633574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,65536,0.1637093358569675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,65536,0.15151111284891763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,4096,0.04092533389727274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,768,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,16384,0.06654400295681424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,16384,0.0795982215139601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,16384,0.045144889089796275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,12288,0.06507911284764607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,12288,0.05126755436261495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,16384,3584,0.03986933496263292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,12288,0.03628355595800612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,10240,0.04416266745991177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,10240,0.05919022030300564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,10240,0.03141955534617106
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,512,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,8192,0.03856888744566176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,8192,0.02572088936964671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,256,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,7168,0.05053777827156914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,16384,128,0.006303999986913469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,7168,0.0344195564587911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,7168,0.02330400049686432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,65536,0.14426044623057047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,6144,0.029840889904234145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,6144,0.02146044373512268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,16384,0.04558577802446154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,5120,0.03762133253945245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,5120,0.026735999517970618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,12288,0.03201333350605435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,5120,0.01917777789963616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,10240,0.03108888864517212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,4096,0.022970666488011677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,4096,0.015799111790127225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,8192,0.022634666826989915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,3584,0.03457955519358317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,3584,0.02027466727627648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,3584,0.014797333214018079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,7168,0.020484444167878892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,3072,0.03260711166593764
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,3072,0.0184897780418396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,3072,0.013776889277829064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,6144,0.01928622192806668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,2560,0.029262221521801416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,2560,0.016192888220151264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,2560,0.01239111108912362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,5120,0.018163555198245578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,2048,0.028091556496090356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,2048,0.014074666632546319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,2048,0.011104889214038849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,4096,0.017160000072585214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,3584,0.01610844499535031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,1536,0.026922666364245947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,1536,0.008821333448092142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,1536,0.010090666512648264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,1024,0.02288355595535702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,1024,0.007035555938879649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,1024,0.006979555719428592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,3072,0.014418666561444601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,768,0.0192284451590644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,768,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,768,0.006953777538405524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,2560,0.012916444076432122
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,512,0.020278222031063504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,512,0.0038737778862317405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,512,0.006623111251327727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,2048,0.011063999599880643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,256,0.01683288812637329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,256,0.0032088889016045462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,256,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,8192,0.05026222268740336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,128,0.014835556348164877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,128,0.003030222323205736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,128,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,1536,0.009194666312800514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,64,0.002546666603949335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,64,0.005950222412745158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,12288,32,0.0028373334142896864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,12288,32,0.006227555374304454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,1024,0.007992888490358988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,65536,0.2054471174875895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,65536,0.14812089337242976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,6144,0.04312710960706075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,768,0.007267555428875818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,16384,0.05710222323735555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,16384,0.08184178007973565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,16384,0.04254844453599718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,512,0.006634666687912411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,65536,0.14355377356211343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,12288,0.044588443305757314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,12288,0.06485244300630358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,12288,0.03177333209249709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,10240,0.057518223921457924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,10240,0.03837688763936361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,10240,0.029496888319651287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,8192,0.05264888869391548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,8192,0.031889776388804116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,256,0.006000888844331105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,8192,0.02424000038041009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,7168,0.04942755566702949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,12288,4096,0.03493155704604255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,7168,0.029567999972237483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,7168,0.021224000387721594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,6144,0.04028888874583774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,6144,0.02604266670015123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,6144,0.01924533314175076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,12288,128,0.006286222073766921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,5120,0.037138667371537946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,5120,0.022622222701708477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,5120,0.017238222890430026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,65536,0.14079377386305067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,16384,0.038070221741994224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,4096,0.020051555501090158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,4096,0.015150222513410779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,12288,0.029789331886503432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,3584,0.033203555477990046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,3584,0.018246221873495314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,10240,0.025212445192866858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,3584,0.014020444618331062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,3072,0.03083555565940009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,3072,0.01657511128319634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,3072,0.013117333253224691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,8192,0.026434666580624048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,2560,0.028772443532943726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,2560,0.014939554863505893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,2560,0.011690666278203329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,7168,0.019875556230545044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,2048,0.027094221777386133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,2048,0.010695999695195092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,2048,0.010662222074137794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,6144,0.018796443939208984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,1536,0.024357333779335022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,1536,0.009229333036475712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,1536,0.008298666940795051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,5120,0.01890222231547038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,1024,0.020980444219377305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,1024,0.00646577775478363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,1024,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,4096,0.01566133399804433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,768,0.01904088921017117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,768,0.004511111312442356
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,768,0.0069208890199661255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,3584,0.014866666661368476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,512,0.01869155632125007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,512,0.003491555651028951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,512,0.006227555374304454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,3072,0.013441777891582914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,256,0.01720533271630605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,256,0.0031911111954185697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,256,0.005934222290913264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,2560,0.011968889170222811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,2048,0.00979288915793101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,128,0.015491555134455362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,128,0.0028604444944196274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,128,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,1536,0.008657777474986183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,64,0.0028444443725877335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,64,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,10240,32,0.0028497777465316984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,10240,32,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,1024,0.007651555869314406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,65536,0.13076089488135442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,65536,0.17016800244649252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,768,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,65536,0.09783733553356594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,16384,0.046559111939536206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,16384,0.07648533582687378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,16384,0.030787554052140977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,512,0.006545777950021956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,12288,0.0558204452196757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,12288,0.035471111536026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,12288,0.025040000677108765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,10240,0.03017688790957133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,256,0.005912000106440649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,10240,0.05230399966239929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,10240,0.02151022189193302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,8192,0.02504355543189579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,8192,0.04634044567743937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,8192,0.019305777218606737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,7168,0.03995199998219808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,7168,0.02204977803760105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,10240,128,0.006211555666393704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,7168,0.01756977703836229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,6144,0.03792622354295518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,6144,0.01926577753490872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,6144,0.016031111280123394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,65536,0.07833955685297649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,5120,0.034978667894999184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,5120,0.017090666625234816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,5120,0.014454222387737699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,16384,0.02404888967672984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,12288,0.020265777905782063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,4096,0.014112000664075216
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,10240,0.01846311158604092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,4096,0.01236977759334776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,3584,0.03091199861632453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,8192,0.015368888775507608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,3584,0.012881778180599213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,3584,0.011510222322411008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,3072,0.02948088778389825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,3072,0.011743110915025076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,3072,0.011038222246699862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,7168,0.01349422252840466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,2560,0.026698667142126296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,2560,0.00998488896422916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,10240,4096,0.03457155492570665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,2560,0.01036888857682546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,6144,0.013108444710572561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,2048,0.025106666816605464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,2048,0.008650666309727563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,2048,0.008998221821255153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,1536,0.0219217770629459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,1536,0.007277333074145847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,1536,0.006962666908899943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,5120,0.01333333303531011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,1024,0.01960533360640208
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,1024,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,1024,0.006275555739800136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,4096,0.011417778001891242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,768,0.018177777528762817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,768,0.004204444587230682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,768,0.006240889016124938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,3584,0.011003555523024665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,512,0.017479111750920612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,512,0.003555555724435382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,512,0.005543111099137201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,3072,0.009520000053776635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,256,0.016767111089494493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,256,0.0031999999450312722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,256,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,2560,0.008637333081828224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,128,0.01516533394654592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,2048,0.007995555798212687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,128,0.0028106667515304354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,128,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,64,0.0028640000770489373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,64,0.005583111196756363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,8192,32,0.0025822222232818604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,1536,0.007130666739410824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,8192,32,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,65536,0.14686577849917942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,1024,0.00630044440428416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,65536,0.09099733167224461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,16384,0.06861244307623969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,16384,0.04241066508822971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,768,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,16384,0.028098666005664404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,512,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,12288,0.03385511040687561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,256,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,12288,0.022986667023764715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,8192,128,0.00554666668176651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,10240,0.05106755428844028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,10240,0.028764443265067205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,10240,0.0210844443904029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,65536,0.07391466697057088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,8192,0.04478933413823446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,8192,0.02535466684235467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,8192,0.017881777551439073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,16384,0.023706666297382776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,7168,0.04470755656560262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,7168,0.01939288940694597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,7168,0.016194666425387066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,6144,0.039000001218583845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,6144,0.017493334081437852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,6144,0.014888000157144336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,12288,0.01884977850649092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,5120,0.033928887711630926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,5120,0.015106666419241162
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,5120,0.013515555196338229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,10240,0.016536000702116225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,4096,0.03184355629814996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,4096,0.01330133361948861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,4096,0.01146577795346578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,8192,0.014510222607188754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,3584,0.029726223813162908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,3584,0.01146666705608368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,7168,0.01310222182008955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,3584,0.011046222514576383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,3072,0.027786665492587622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,3072,0.010715555813577441
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,3072,0.010372444159454769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,6144,0.011805333197116852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,2560,0.026376888155937195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,2560,0.009350222018029954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,2560,0.009526222116417354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,5120,0.011015999648306104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,2048,0.023637332850032385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,2048,0.008084444536103142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,8192,4096,0.03291733397377862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,2048,0.007633777956167857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,1536,0.02092355489730835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,1536,0.006718222051858902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,4096,0.010024000373151567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,1536,0.006598222172922558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,1024,0.01885955532391866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,1024,0.004562666846646203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,1024,0.006227555374304454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,3584,0.008782222039169734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,768,0.018596443865034316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,768,0.003906666818592283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,768,0.005937777873542573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,3072,0.00795999997191959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,512,0.01666755477587382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,512,0.0035217776894569397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,512,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,2560,0.007340444458855524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,256,0.015824889143308003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,256,0.003162666741344664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,256,0.0052639999323421055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,2048,0.0070462218589252895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,128,0.015062222878138224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,1536,0.006627555522653792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,128,0.0028248888750871024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,128,0.005557333429654439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,64,0.002809777855873108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,64,0.0052675555149714155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,7168,32,0.0028213332924577924
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,7168,32,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,1024,0.0058871110280354815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,65536,0.12378044923146565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,65536,0.11750843789842393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,65536,0.08430311414930557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,768,0.00591466658645206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,16384,0.07005066341824003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,16384,0.03811733259095086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,16384,0.025796444879637823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,512,0.005551110953092575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,12288,0.05647110939025879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,12288,0.03052177694108751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,12288,0.021291555629836187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,256,0.0052177777720822235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,10240,0.04938933253288269
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,10240,0.025702221526039973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,10240,0.018961777289708454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,8192,0.04384088847372267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,7168,128,0.005280888742870755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,8192,0.0221377776728736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,65536,0.12581777572631836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,8192,0.01645599967903561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,7168,0.0196426659822464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,7168,0.04258222381273905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,7168,0.015216888652907478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,65536,0.07176355520884196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,6144,0.03564533260133531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,6144,0.017831999394628737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,6144,0.01425155500570933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,5120,0.031983110639784075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,5120,0.01593866613176134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,5120,0.01272266689274046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,16384,0.022781333989567224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,12288,0.01827022267712487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,10240,0.015798222687509324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,4096,0.012056888805495368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,4096,0.010952889091438718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,7168,12288,0.05487111210823059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,3584,0.028561777538723413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,3584,0.010712888505723743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,8192,0.014169777433077494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,3584,0.010244444840484196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,7168,0.012748444245921241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,3072,0.026878222823143005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,3072,0.009388444324334463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,3072,0.009515555368529426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,6144,0.011757333245542316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,2560,0.023323555787404377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,2560,0.008838222258620793
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,2560,0.008083555433485243
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,2048,0.02136444383197361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,2048,0.007640889121426477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,2048,0.006909333169460297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,5120,0.01072622256146537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,1536,0.0199608885579639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,1536,0.006268444574541516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,1536,0.006577777779764599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,4096,0.009014222357008193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,1024,0.004176889028814104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,1024,0.005950222412745158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,3584,0.008040000167157915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,768,0.017874666386180453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,768,0.003826666623353958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,3072,0.007666666474607255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,768,0.005879111174080107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,512,0.01611199975013733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,512,0.0034871109657817413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,2560,0.007608888877762689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,512,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,256,0.015827554795477126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,256,0.003160888950030009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,2048,0.006643555644485686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,256,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,128,0.01441422270403968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,128,0.002858666703104973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,128,0.005215999980767568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,1536,0.006538666784763336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,64,0.002893333426780171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,64,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,6144,32,0.0028373334142896864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,6144,32,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,1024,0.0058959999846087555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,65536,0.10725066396925186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,768,0.005922666854328579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,512,0.005724444571468566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,65536,0.07684533463584052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,16384,0.059621334075927734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,16384,0.0328400002585517
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,256,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,16384,0.02402577797571818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,12288,0.026737777723206416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,12288,0.05167644553714328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,12288,0.019864888654814828
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,6144,128,0.0052177777720822235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,10240,0.023002665903833177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,10240,0.04281511240535312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,10240,0.017670222454600863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,8192,0.042111112011803516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,8192,0.01998133295112186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,8192,0.015765332513385348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,65536,0.0694133308198717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,7168,0.037590222226248846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,7168,0.017882666654056974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,16384,0.021939555803934734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,7168,0.01423377792040507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,6144,0.03337688909636603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,6144,0.01647111111217075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,12288,0.017823111679818895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,6144,0.013384000294738345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,10240,0.015471999843915304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,5120,0.029607110553317603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,5120,0.01458577811717987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,5120,0.01201333353916804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,4096,0.027832001447677612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,4096,0.01075644459989336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,4096,0.010668444136778513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,8192,0.013639999760521783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,3584,0.028417779339684382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,3584,0.009385777844323052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,3584,0.009679111341635386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,7168,0.012085333466529846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,3072,0.025568000144428674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,3072,0.008634666601816813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,6144,0.011352000137170156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,3072,0.008265777594513362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,5120,0.009689778089523315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,2560,0.023542222049501207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,2560,0.007924444145626491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,2560,0.007661333514584436
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,2048,0.02086666723092397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,2048,0.006894222150246303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,4096,0.008276444342401292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,2048,0.006730666591061487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,3584,0.007622222105662028
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,1536,0.01920355525281694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,1536,0.005196444276306364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,1536,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,3072,0.007369777394665613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,1024,0.01755199995305803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,1024,0.0039013334446483185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,2560,0.007249777515729268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,1024,0.005906666732496685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,768,0.017868444323539734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,768,0.003532444437344869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,768,0.005883555445406172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,2048,0.006624889042642381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,512,0.015507555670208402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,512,0.0031679999083280563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,512,0.005616888817813661
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,1536,0.006243555496136348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,256,0.014760888285107084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,256,0.0031582222630580268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,256,0.005386666705211003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,128,0.015117333994971381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,1024,0.0058986664646201665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,128,0.00282577777074443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,128,0.005191111316283544
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,4096,0.03094489044613308
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,64,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,64,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,5120,32,0.0028106667515304354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,768,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,5120,32,0.005594666633341048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,65536,0.08912266625298394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,65536,0.09047377771801418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,65536,0.07081600030263265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,512,0.005562666803598404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,16384,0.028035554620954726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,16384,0.06490221950742933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,16384,0.02193333374129401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,12288,0.05235377616352505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,12288,0.02150844368669722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,12288,0.018363555272420246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,10240,0.04571111003557841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,256,0.005234666582610872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,10240,0.018900444110234577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,10240,0.016511110795868766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,6144,1024,0.018187556001875136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,5120,128,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,8192,0.016819554898473952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,8192,0.014811555544535318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,65536,0.06705777512656318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,7168,0.041095998552110456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,7168,0.014946666028764514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,7168,0.01367911116944419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,16384,0.021011556188265484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,6144,0.013973333769374423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,6144,0.035016000270843506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,6144,0.012415111064910889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,12288,0.017172444197866652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,5120,0.012378666963842181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,5120,0.030447999636332195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,5120,0.011370666325092316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,10240,0.015034667319721647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,4096,0.01127911110719045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,4096,0.009682666924264696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,8192,0.013336000343163809
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,3584,0.026072000463803608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,3584,0.01182755579551061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,7168,0.011694221860832639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,3584,0.008107555409272512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,6144,0.010339555641015371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,3072,0.024695111645592585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,3072,0.010538666612572141
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,3072,0.007993777592976889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,2560,0.02161688937081231
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,5120,0.008549333446555668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,2560,0.009290667043791877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,2560,0.006960888703664144
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,2048,0.020260444945759244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,2048,0.009140444298585257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,2048,0.006613333192136552
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,4096,0.007995555798212687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,1536,0.019503111640612285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,1536,0.006302222195598815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,1536,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,5120,65536,0.11366755432552761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,1024,0.017614222235149808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,1024,0.005809777726729711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,1024,0.006224000205596288
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,3584,0.007654222349325816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,768,0.017550221747822232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,3072,0.007305777735180325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,768,0.004924444274769889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,768,0.005778666585683823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,2560,0.006980444822046492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,512,0.015186667442321777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,512,0.0040755557517210645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,512,0.005519999812046687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,256,0.015471999843915304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,2048,0.0063493334584765965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,256,0.002879111096262932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,256,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,1536,0.006228444476922353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,128,0.014767999450365702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,128,0.0028364445186323593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,128,0.005218666460778978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,1024,0.005962666538026597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,64,0.0026862221873468826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,64,0.005582222094138463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,4096,32,0.0025333333760499954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,768,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,4096,32,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,65536,0.08110044399897258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,512,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,65536,0.06793777810202704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,256,0.005095110999213325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,16384,0.025428444147109985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,16384,0.06611822048823039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,16384,0.021262221866183813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,4096,128,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,12288,0.019966221517986722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,12288,0.018214222457673814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,10240,0.04536799920929802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,10240,0.017768889665603638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,10240,0.016152888536453247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,65536,0.06615377797020806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,8192,0.041336887412601046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,8192,0.01500888831085629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,8192,0.01459200017982059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,16384,0.020968000094095867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,7168,0.03702044486999512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,7168,0.014027555783589682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,7168,0.013078221844302284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,12288,0.017058667209413316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,6144,0.03489777776930068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,6144,0.013044444223244986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,6144,0.01239466667175293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,10240,0.014859555496109856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,5120,0.03046222196684943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,5120,0.011691555380821228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,5120,0.011016888750924004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,8192,0.012840000291665396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,4096,0.027748444014125403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,4096,0.010671999719407825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,4096,0.008575999902354347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,7168,0.010712888505723743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,3584,0.025843555728594463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,3584,0.011388444238238864
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,3584,0.008261333737108443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,6144,0.009359999663299983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,3072,0.02386666668785943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,3072,0.010379555324713389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,3072,0.007553777760929531
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,5120,0.008660444782839881
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,2560,0.020967110991477966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,2560,0.009017777939637503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,2560,0.007254222200976477
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,4096,0.00796088907453749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,3584,0.007640000018808577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,2048,0.007310222420427535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,2048,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,3072,0.007329777710967594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,1536,0.019221333993805777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,1536,0.006038222048017714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,1536,0.006344888773229387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,2560,0.0069520001610120135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,1024,0.01754755609565311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,1024,0.0052639999323421055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,1024,0.005946666830115848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,2048,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,768,0.018238221605618794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,768,0.004595555365085602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,768,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,8192,0.04018400112787882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,1536,0.006253333141406377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,512,0.004187555362780889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,512,0.01442311041884952
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,512,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,1024,0.005882666756709416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,256,0.0031066667288541794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,256,0.005528000079923206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,768,0.005600000007285013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,512,0.0053004444473319584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,4096,4096,0.02906755606333415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,128,0.0028168888141711554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,128,0.0053582220441765254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,64,0.0025351111673646504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,64,0.005584888988071018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3584,32,0.0026319999661710528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3584,32,0.005221333354711533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,256,0.00526311124364535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,65536,0.06839822398291694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,65536,0.08637777964274089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,65536,0.0653031137254503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,16384,0.058879110548231334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,16384,0.021611554755104914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3584,128,0.005241777747869492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,16384,0.02086666723092397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,12288,0.04529777831501431
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,12288,0.017995556195576985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,12288,0.017695999807781644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,65536,0.06527733140521579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,16384,0.020430222153663635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,10240,0.04452622267935011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,10240,0.016019556257459853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,10240,0.01577422188387977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,12288,0.01678400072786543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,8192,0.041238221857282854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,8192,0.013747555514176687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,8192,0.013752889302041797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,7168,0.03822133276197646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,7168,0.012816000315878125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,10240,0.014167111780908374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,7168,0.012832889126406776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,6144,0.03219555483924018
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,6144,0.01201333353916804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,6144,0.011732444167137146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,8192,0.012265777422322167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,5120,0.02938844429122077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,5120,0.011023110813564725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,65536,0.0879431102010939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,5120,0.009823111196359
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,4096,0.027433777848879497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,4096,0.00980444418059455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,7168,0.00999466660949919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,4096,0.008788444101810455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,3584,0.025421332981851365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,3584,0.009318222602208456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,6144,0.009332444104883406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,3584,0.007646222081449296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,12288,0.051515556044048734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,3072,0.00904266701804267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,3072,0.007074666519959767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,5120,0.00866844422287411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,2560,0.01993955506218804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,2560,0.007800000409285228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,2560,0.0069644442862934535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,4096,0.00794222205877304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,2048,0.01925066610177358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,2048,0.006594666590293248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,2048,0.006573333508438534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,3584,0.007606222397751278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,1536,0.01789866718981001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,1536,0.006226666685607698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,3072,0.007305777735180325
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,1536,0.006228444476922353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,1024,0.016542222764756944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,1024,0.005241777747869492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,1024,0.006118222243256039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,2560,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,768,0.01754577789041731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,768,0.004833777745564778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,768,0.005570666657553778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,2048,0.006615110983451207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,512,0.014128888646761576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,512,0.004239999999602635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,512,0.0052151112920708126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,1536,0.005947555518812603
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,256,0.014813333749771118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,256,0.0035573335157500375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,256,0.0054551110499435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,1024,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,128,0.01477866702609592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,128,0.0031395554542541504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,128,0.0052382221652401825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,768,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,64,0.0028595555987623003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,64,0.005254222287072076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,3072,32,0.002846222163902389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,512,0.005261333452330695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,3072,32,0.005330666485759947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,2048,0.02058844433890449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,65536,0.060405333836873375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,65536,0.08458933565351699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,256,0.0052684446175893145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,65536,0.0629040002822876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,16384,0.0196426659822464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,16384,0.06348622507519193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,16384,0.020237333244747586
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,12288,0.016319110989570618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,3072,128,0.005236444373925527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,12288,0.04814666509628296
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,12288,0.017265778448846605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,10240,0.042302221059799194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,10240,0.01423822177780999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,10240,0.015094222293959724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,8192,0.03809955716133118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,65536,0.0641768905851576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,8192,0.012664889295895895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,8192,0.013473778135246701
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,16384,0.02032355633046892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,7168,0.03188000122706095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,7168,0.01205599970287747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,7168,0.01239822225438224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,6144,0.03018310997221205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,12288,0.01629155543115404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,6144,0.011031111081441244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,6144,0.011176889141400656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,5120,0.02425777746571435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,10240,0.013680000272062091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,5120,0.010016000105275048
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,5120,0.00999111102686988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,8192,0.010943111446168689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,256,0.015463110473420886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,4096,0.025157334076033697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,4096,0.009212444225947062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,4096,0.00775022225247489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,7168,0.010025777750545079
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,3584,0.008727110922336578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,3584,0.02342044479317135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,3584,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,6144,0.00906133320596483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,3072,0.022300443715519373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,3072,0.0079644446571668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,3072,0.007329777710967594
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,2560,0.020968000094095867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,5120,0.008338666624493068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,2560,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,2560,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,2048,0.01795288920402527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,4096,0.008007999923494127
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,2048,0.005948444621430502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,2048,0.006639110959238476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,1536,0.017154667112562392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,3584,0.007518221934636434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,1536,0.005917333480384614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,1536,0.006241777704821692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,3072,0.007303111255168915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,1024,0.015519110692871941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,1024,0.004932444542646408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,1024,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,768,0.004230222354332606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,768,0.01752622259987725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3584,128,0.01477688882086012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,768,0.005528000079923206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,2560,0.006610666712125142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,512,0.013819555441538492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,512,0.0036640001667870414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,512,0.0052595556610160405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,2048,0.006693333387374878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,256,0.0032177778581778207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,256,0.014472888575659858
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,1536,0.006317333214812809
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,256,0.005205333232879639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2560,128,0.013741333451535968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,128,0.003165333428316646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,128,0.005559999909665849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,1024,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,64,0.002838222309947014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,64,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,768,0.005585777676767773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2560,32,0.0028533333291610083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2560,32,0.0052782222628593445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,512,0.005576000031497743
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,65536,0.0510426660378774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,65536,0.07788088586595324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,256,0.005205333232879639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,65536,0.060248891512552895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,16384,0.016509332590632968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,16384,0.06299110915925768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,16384,0.019714666737450492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2560,128,0.005187555733654234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,12288,0.016693333784739178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,12288,0.013624000052611033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,12288,0.04478311207559374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,10240,0.042709334029091724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,65536,0.0640044477250841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,10240,0.01236355553070704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,10240,0.015103111664454142
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,16384,0.01980177727010515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,8192,0.03805155555407206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,8192,0.011013333168294696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,7168,0.01036533299419615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,8192,0.012784889174832238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,12288,0.014491554763582019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,7168,0.011990221838156382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,6144,0.009763555394278632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,6144,0.00999022192425198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,10240,0.012124444047609964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,5120,0.028648889727062646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,5120,0.008984888593355814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,5120,0.00868800034125646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,8192,0.010988444089889526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,4096,0.026384888423813715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,4096,0.010392888552612728
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,4096,0.007710222568776872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,7168,0.009673777553770278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,3584,0.024382222029897902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,3584,0.00906133320596483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,3584,0.007616000043021307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,6144,0.008982222113344405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,3072,0.02279377811484867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,3072,0.006576889091067844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,3072,0.007618666523032718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,5120,0.008576889004972246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,2560,0.020229332976871066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,2560,0.006051555689838197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,2560,0.0069715554515520734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,2048,0.0196106665664249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,4096,0.008003555238246918
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,2048,0.006634666687912411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,2048,0.00656355544924736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,3584,0.007585778004593319
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,1536,0.017208000024159748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,3072,0.007277333074145847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,1536,0.006238222122192383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,1536,0.006586666736337874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,2560,0.006952889263629913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,1024,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,1024,0.005970666805903117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,2048,0.006625777731339137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,768,0.01792888840039571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,768,0.0042275554604000514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,768,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,1536,0.006215999937719769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,3072,3072,0.024001777172088623
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,512,0.003917333152559069
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,512,0.013385777672131857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,512,0.005564444594913059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,256,0.014848000473446317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,1024,0.0059368887709246735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,256,0.0035013332962989807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,256,0.0052533331844541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,128,0.01419911119672987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,128,0.0031991110493739447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,128,0.005208888815508948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,768,0.005583111196756363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,64,0.002850666642189026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,64,0.005420444326268301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,512,0.00526311124364535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,2048,32,0.0026364444444576898
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,2048,32,0.005035555611054103
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,65536,0.037272890408833824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,256,0.00535111129283905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,65536,0.0770346654786004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,65536,0.058936887317233615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,16384,0.05526933405134413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,16384,0.013719999955760108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,16384,0.019368888603316415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,2048,128,0.005245333330498801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,12288,0.011386666860845355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,12288,0.0425217780801985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,12288,0.01650311052799225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,10240,0.010791999598344168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,65536,0.06327999962700738
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,10240,0.014467555615637036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,16384,0.017208000024159748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,8192,0.03320088982582092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,8192,0.010423111418883005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,8192,0.011066666907734342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,12288,0.013800000150998434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,7168,0.00979555563794242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,7168,0.010384889112578498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,10240,0.012337778177526263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,6144,0.03128977616628011
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,6144,0.008843555218643611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,8192,0.010644444160991244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,6144,0.009381333159075843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,5120,0.02763644523090786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,7168,0.010011555420027839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,5120,0.007506666911972894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,5120,0.008996444443861643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,4096,0.023364444573720295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,4096,0.007583111524581909
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,6144,0.00903555585278405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,4096,0.007972444097201029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,3584,0.022659555077552795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,3584,0.007152888509962294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,3584,0.007696000238259633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,5120,0.008294222255547842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,3072,0.022962666220135156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,4096,0.007748444047239091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,3072,0.006590222318967183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,3072,0.007261333366235097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,2560,0.018954666124449838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,2560,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,3584,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,2560,0.006976889239417181
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,3072,0.007205333146784041
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,2048,0.005888888819350137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,2048,0.006360888895061281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,2560,0.006952889263629913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,1536,0.01685422162214915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,1536,0.005578666511509154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,1536,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,2048,0.006566222343179915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,1024,0.016159999701711867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,1024,0.004516444272465176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,1024,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,1536,0.006272000157170826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,768,0.01750577820671929
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,768,0.0042151109211974675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,768,0.005599110904667113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,1024,0.005590222362014983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,512,0.013755555782053204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,512,0.0034871109657817413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,768,0.005591111050711737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,512,0.005219555563396878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,256,0.014895111322402954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,256,0.0028968888024489083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,256,0.00554666668176651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,512,0.005425777700212266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,256,0.005203555441564984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,128,0.014583110809326172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,128,0.002809777855873108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,128,0.005535111245181825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1536,128,0.0052435555391841466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,64,0.0028648889727062653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,64,0.0052764444715446895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1536,32,0.002520888836847411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1536,32,0.0057502223385704895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,65536,0.07594666878382365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,65536,0.027316444449954565
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,65536,0.05781777699788412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,16384,0.04691822330156962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,65536,0.06309511264165242
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,16384,0.0151582227812873
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,16384,0.017960889471901786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,12288,0.0359866652223799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,12288,0.012636444634861417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,7168,0.03286044465170966
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,12288,0.01367911116944419
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,16384,0.01649777756796943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,10240,0.033556444777382746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,6144,0.031195554468366835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,10240,0.011732444167137146
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,10240,0.012376888758606382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,12288,0.01346666696998808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,8192,0.030943999687830608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,8192,0.010679110884666443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,8192,0.010388444695207808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,7168,0.028821332587136164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,7168,0.009383111364311641
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,10240,0.012102222277058495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,7168,0.010041777458455827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,6144,0.024710221423043147
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,6144,0.008664000365469191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,8192,0.010653333531485664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,6144,0.009713778065310584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,5120,0.008021333151393467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,5120,0.022629333866967097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,5120,0.008811555802822113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,7168,0.009504888620641496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,4096,0.021291555629836187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,4096,0.00720177756415473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,4096,0.007979555262459649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,6144,0.009242666264375051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,3584,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,3584,0.01993333299954732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,3584,0.007637333538797166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,2048,1024,0.016215999921162922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,5120,0.008277333445019191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,3072,0.018912000788582694
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,3072,0.006527110934257507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,4096,0.0076453329788313965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,3072,0.007314666277832455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,3584,0.007638221813572778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,2560,0.01681777834892273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,2560,0.0058088890380329555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,2560,0.006931555353932911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,2048,0.01682133310370975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,2048,0.0063164445261160536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,2048,0.008344888687133789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,3072,0.007270221908887227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,1536,0.0063075555695427795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,2560,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,1536,0.015474667151769003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,1536,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,1024,0.014826666977670459
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,1024,0.005212444398138259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,2048,0.006569777925809224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,1024,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,768,0.016128889388508268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,1536,0.005936000082227919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,768,0.004214222232500712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,768,0.005597333527273602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,512,0.0038684445122877755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,1024,0.005946666830115848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,512,0.012759110993809171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,512,0.00553688903649648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,256,0.01482399966981676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,768,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,256,0.0031715554909573663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,256,0.0052426668504873914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,512,0.005333333379692501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1024,128,0.014101333088344999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,128,0.0028924445311228433
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,128,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,64,0.002537777854336633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,256,0.005244444641802046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,64,0.005454222361246745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,1024,32,0.0028577778074476454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,1024,32,0.005218666460778978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,7168,0.033612443341149226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,65536,0.057258665561676025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,1024,128,0.005202666752868229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,10240,0.03864355550871955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,65536,0.021948445174429152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,65536,0.07544977797402276
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,16384,0.03516799873775906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,16384,0.010370666782061258
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,16384,0.01752711170249515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,65536,0.06249511241912842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,12288,0.02944799926545885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,12288,0.011420444481902652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,12288,0.013390222357379066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,16384,0.0162471118900511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,10240,0.009649777577983009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,10240,0.012083555261294046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,12288,0.013409777647919126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,8192,0.023698666029506262
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,8192,0.009142222503821055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,8192,0.010695999695195092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,10240,0.011807999677128263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,7168,0.02132355504565769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,7168,0.008323555191357931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,7168,0.010384889112578498
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,8192,0.010689777632554373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,6144,0.02130488885773553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,6144,0.007663110891977946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,6144,0.009466666314336989
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,7168,0.00964355551534229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,5120,0.019352000620630052
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,5120,0.006647111227114995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,5120,0.009023110899660323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,6144,0.008976000050703684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,1536,2048,0.0185244447655148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,4096,0.018578666779730055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,4096,0.007672888537247975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,4096,0.007981333467695449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,5120,0.008344888687133789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,3584,0.01772622267405192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,3584,0.0069822221994400024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,3072,0.0069324444565508105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,3584,0.007370666497283512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,4096,0.007726222276687622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,3072,0.016872000363137987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,3072,0.007019555403126611
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,3584,0.007609777980380588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,2560,0.0166320006052653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,2560,0.00628355559375551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,2560,0.0069057775868309875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,3072,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,2560,0.006633777585294511
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,2048,0.015863999724388123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,1536,0.004972444640265571
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,2048,0.004943999979231092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,2048,0.006304889089531369
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,1536,0.01444355481200748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,1536,0.005984000033802456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,2048,0.006587555425034628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,1024,0.00453244439429707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,1024,0.005534222142563925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,1536,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,768,0.015807999504937064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,768,0.0038444445365005066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,768,0.005568888866239124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,512,0.01242844429281023
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,1024,0.005952888892756567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,768,0.005241777747869492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,512,0.0034977777136696708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,512,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,256,0.014171555638313293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,256,0.0031679999083280563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,256,0.005245333330498801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,512,0.005386666705211003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,128,0.013790222505728403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,64,0.0026497778793176017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,128,0.0028000000036425064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,128,0.005600000007285013
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,32,0.0052675555149714155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,128,0.004902222090297275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,768,256,0.005210666606823604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,65536,0.07464444637298584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,768,64,0.005605333381228977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,768,32,0.0024888888001441956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,65536,0.017311111092567444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,65536,0.05592888593673706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,16384,0.027063111464182537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,16384,0.011309333145618439
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,16384,0.015821332732836407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,16384,0.016176000237464905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,65536,0.05504444572660658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,12288,0.02905155552758111
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,12288,0.008323555191357931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,12288,0.013428444663683573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,10240,0.007662222617202335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,10240,0.0120319997270902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,8192,0.01959377858373854
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,12288,0.01350933313369751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,8192,0.007954667011896769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,8192,0.010416000253624385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,10240,0.012089778151777057
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,7168,0.018412444326612685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,7168,0.007607111500369177
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,7168,0.009662222531106737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,8192,0.010739555789364709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,6144,0.017525333497259352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,6144,0.0069751110341813825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,6144,0.009297777381208207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,7168,0.009721777505344814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,5120,0.016491555505328707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,5120,0.006956444846259222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,6144,0.00907733374171787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,5120,0.008619555168681675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,4096,0.015582222077581616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,4096,0.007298666569921706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,4096,0.007592000067234039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,5120,0.008295111358165741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,3584,0.015132443772421943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,3584,0.006631111105283101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,4096,0.007648000286685095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,3584,0.007623111208279927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,3072,0.01590844492117564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,3072,0.006569777925809224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,3072,0.007267555428875818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,3584,0.007648888561460707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,2560,0.01443377799457974
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,2560,0.0062773335311147906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,2560,0.006917333437336816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,3072,0.006937777830494775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,2048,0.013863110707865821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,2048,0.00591555568906996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,2560,0.006909333169460297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,2048,0.006287999865081575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,1536,0.013388444152143268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,1536,0.00452533322903845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,1536,0.005879999862776862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,2048,0.006590222318967183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,1024,0.013083555632167392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,1024,0.003915555361244413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,1024,0.005912000106440649
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,1536,0.006099555641412735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,768,0.003522666792074839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,768,0.014478221535682678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,1024,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,768,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,512,0.011700444751315646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,512,0.0031519999934567344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,512,0.005622222191757626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,768,0.005630222045713001
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,256,0.013184000220563678
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,256,0.002858666703104973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,512,0.005255110975768831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,256,0.005294222384691238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,128,0.011686222420798408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,128,0.0028648889727062653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,256,0.005209777918126848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,64,0.005395555661784277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,128,0.0052684446175893145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,64,0.002814222127199173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,512,128,0.0052915554907586836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,512,32,0.002502222235004107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,512,32,0.005230222311284807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,65536,0.01521511044767168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,65536,0.0726426641146342
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,16384,0.008641777767075432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,65536,0.04980977707438999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,16384,0.026392888691690233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,65536,0.051310221354166664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,16384,0.016182222300105624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,16384,0.015777778294351365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,12288,0.02265600032276577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,12288,0.007983110845088959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,12288,0.013416888813177744
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,10240,0.020617778102556866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,10240,0.007657777931955125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,12288,0.013397333522637686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,10240,0.012368888490729861
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,8192,0.019318222999572754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,10240,0.012042666474978128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,8192,0.0069715554515520734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,7168,0.006601777755551868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,8192,0.01033333357837465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,7168,0.018181333939234417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,8192,0.010686222049925061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,6144,0.017281777328915067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,7168,0.010042666561073726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,7168,0.009711999860074785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,6144,0.006296000132958095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,6144,0.009371555513805812
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,5120,0.016172443826993305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,6144,0.008960000342792934
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,5120,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,5120,0.00868000007337994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,4096,0.015196444259749519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,5120,0.008292444050312042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,4096,0.0058764442801475525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,4096,0.00795377790927887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,3584,0.014799111419253878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,4096,0.007914666500356462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,3584,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,3072,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,3584,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,3584,0.007424888511498769
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,3072,0.007600000335110559
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,2560,0.006587555425034628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,2560,0.013410666750537025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,3072,0.007296889192528195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,2560,0.006968888971540663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,2048,0.01276622215906779
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,2048,0.005598222215970357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,2560,0.0069253332912921906
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,2048,0.006586666736337874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,1536,0.01273155543539259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,2048,0.006441777778996362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,1536,0.004214222232500712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,1536,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,1024,0.012582222620646158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,1536,0.0059262220230367445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,1024,0.0038622220357259116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,10240,0.03218400147226121
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,1024,0.005902222047249476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,768,0.014483556151390076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,1024,0.005575110928879843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,768,0.003555555724435382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,768,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,512,0.011391110718250275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,768,0.005597333527273602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,512,0.0031848889258172777
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,512,0.0052062223354975385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,256,0.01274400038851632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,512,0.0052586665583981415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,256,0.002871111035346985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,256,0.0052737775776121355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,128,0.011394666300879585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,256,0.005203555441564984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,128,0.0028151110228565005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,128,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,64,0.002475555572244856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,64,0.004931555440028508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,256,128,0.0052044445441828836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,256,32,0.002528000002106031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,256,32,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,65536,0.07125777668423124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,65536,0.014113777213626437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,65536,0.04707644383112589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,16384,0.00796622203456031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,65536,0.05056711037953695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,16384,0.015798222687509324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,16384,0.01646488904953003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,12288,0.022327999273935955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,12288,0.013397333522637686
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,12288,0.007335999773608313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,10240,0.019887111253208585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,10240,0.00693866651919153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,12288,0.013515555196338229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,10240,0.0123831108212471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,10240,0.011752888560295105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,8192,0.018255111243989732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,8192,0.006529777828190062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,8192,0.01072533345884747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,8192,0.010686222049925061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,7168,0.017491555876202054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,7168,0.006574222197135289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,7168,0.00997155573632982
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,6144,0.006244444598754247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,6144,0.01585244470172458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,6144,0.009650666680600908
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,7168,0.009699555734793345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,5120,0.005958222266700532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,5120,0.014811555544535318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,6144,0.008975110948085785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,5120,0.008980444735950893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,5120,0.008299555215570662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,4096,0.007937777373525832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,4096,0.01408088869518704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,4096,0.005879999862776862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,3584,0.014319111903508505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,4096,0.007652444144090016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,3584,0.005943999936183293
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,3584,0.007657777931955125
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,768,1024,0.013585777746306526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,3584,0.0075902218619982404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,3072,0.013927110367351107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,3072,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,2560,0.006271111054552927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,3072,0.007277333074145847
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,3072,0.006980444822046492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,2560,0.013421333498424955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,2560,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,2560,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,2048,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,2048,0.006603555546866522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,2048,0.006222222414281633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,1536,0.006232888748248418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,1536,0.012056888805495368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,1536,0.004359111189842224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,1536,0.005727111051479976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,1024,0.0038399998512532976
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,1024,0.005587555468082428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,1024,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,768,0.013439999686347114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,768,0.0035706667436493766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,768,0.005543111099137201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,768,0.005617777920431561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,512,0.010052444206343757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,512,0.0031999999450312722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,512,0.005221333354711533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,512,0.005249777601824866
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,256,0.0028364445186323593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,256,0.010015111002657149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,256,0.004912000149488449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,256,0.004913777940803104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,128,0.00978133330742518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,128,0.0025235555238193935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,512,10240,0.025073778298166063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,128,0.004924444274769889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,64,0.002566222308410539
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,2,128,128,0.004919111314747068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,64,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,128,32,0.002847111059559716
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,128,32,0.004956444518433677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,65536,0.010036444498433007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,65536,0.046520888805389404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,16384,0.006890666567616993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,16384,0.015497777197096082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,12288,0.005942222144868638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,12288,0.013055110971132914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,10240,0.005941333456171884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,10240,0.011723555624485016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,8192,0.006232000059551663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,7168,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,8192,0.010621333287821876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,7168,0.009703999592198266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,6144,0.005587555468082428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,6144,0.009000889129108852
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,5120,0.005567111074924469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,4096,0.005296000176005893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,4096,0.0075333333677715724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,5120,0.008639111287064022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,3584,0.005618666609128316
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,3584,0.007295110987292395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,3072,0.005271111097600725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,3072,0.007293333609898885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,2560,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,2560,0.0069875551594628235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,2048,0.005254222287072076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,2048,0.006281777802440856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,1536,0.004219555606444676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,1536,0.006052444378534953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,1024,0.00350577798154619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,1024,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,768,0.003286222202910317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,512,0.0031759999692440033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,768,0.005572444448868434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,512,0.005214222189452913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,256,0.0028560000161329904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,128,0.002519999941190084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,256,0.004917333523432414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,64,0.002502222235004107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,128,0.0052622221410274506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,64,0.00519466648499171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,64,32,0.0028248888750871024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,64,32,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,65536,0.009004444711738164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,65536,0.04628800021277534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,16384,0.006588444527652528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,16384,0.015470221638679504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,12288,0.005818666683302985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,12288,0.013267555170589022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,8192,0.005241777747869492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,10240,0.0062871111763848206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,8192,0.010375111467308467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,10240,0.011874666644467248
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,7168,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,7168,0.010006222460005019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,5120,0.004971555537647671
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,6144,0.00590133335855272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,6144,0.009315555294354757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,5120,0.008979555633332994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,4096,0.005244444641802046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,4096,0.007638221813572778
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,3584,0.005617777920431561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,3584,0.007603555917739868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,3072,0.005240889059172736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,3072,0.007256000406212277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,2048,0.005254222287072076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,2560,0.005591111050711737
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,2560,0.0069368887278768755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,2048,0.00599911105301645
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,1024,0.0035706667436493766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,1536,0.004230222354332606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,1536,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,768,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,1024,0.005583111196756363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,512,0.005064888960785336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,768,0.0035031110876136353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,512,0.0031759999692440033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,256,0.0028640000770489373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,256,0.004962666581074397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,128,0.002826666666401757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,128,0.004933333231343163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,64,0.0024684444069862366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,64,0.004958222309748332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,2,32,32,0.002492444382773505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,2,32,32,0.004928888960017098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,16384,0.3339386781056722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,16384,0.2599662144978841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,16384,0.18155999978383383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,12288,0.21158844894833037
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,256,3072,0.014459555347760519
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,12288,0.2541084554460314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,12288,0.13066666656070286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,10240,0.214227557182312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,10240,0.1785422166188558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,10240,0.11215021875169541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,16384,0.025701334079106648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,8192,0.15791822804345026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,8192,0.17492089006635878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,8192,0.09695111380683051
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,7168,0.14125688870747885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,7168,0.14500533209906682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,7168,0.07907822397020128
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,2048,0.0127368892232577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,6144,0.12792978021833631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,2,128,1024,0.011758222348160215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,6144,0.12593955463833278
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,6144,0.06927111413743761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,5120,0.12151466475592719
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,5120,0.10763200124104817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,5120,0.05860444572236803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,4096,0.10701333151923285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,4096,0.08998133076561822
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,4096,0.04747555653254191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,16384,0.1604026688469781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,3584,0.10078222221798366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,3584,0.07340977589289348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,3584,0.04343200061056349
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,12288,0.12278132968478733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,3072,0.09232978026072185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,3072,0.0636791123284234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,3072,0.03757155603832669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,10240,0.10912533601125081
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,2560,0.08567377593782212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,2560,0.0533537765343984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,2560,0.03280444608794318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,8192,0.0942888855934143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,2048,0.07990488741132948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,2048,0.043792890177832715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,2048,0.02806666824552748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,7168,0.08550310797161526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,1536,0.0695564415719774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,1536,0.033824000093672014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,1536,0.02233244478702545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,6144,0.06867644521925184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,1024,0.06276000208324857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,5120,0.061518225404951304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,1024,0.024156444602542456
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,1024,0.017895110779338412
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,768,0.030458665556377832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,768,0.019087110956509907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,768,0.016162667009565566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,4096,0.05016444457901848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,512,0.0559119979540507
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,512,0.013724444641007317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,512,0.01459022197458479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,3584,0.045239110787709556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,256,0.02299733294381036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,256,0.009016888837019602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,256,0.01275111072593265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,3072,0.03875022133191427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,65536,128,0.01958933307064904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,128,0.00628266649113761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,128,0.010391111175219217
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,2560,0.03415822320514255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,64,0.005550222264395819
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,64,0.010089777410030365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,65536,32,0.005600888695981767
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,65536,32,0.010272888673676385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,2048,0.02933688958485921
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,65536,0.31812887721591526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,65536,0.29475288920932347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,1536,0.02327555583582984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,16384,0.09135466814041138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,16384,0.08431022034751044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,65536,0.19655466079711914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,16384,0.05343022280269199
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,12288,0.07463377714157104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,12288,0.0648568868637085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,12288,0.041646222273508705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,1024,0.017136888371573556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,768,0.01482311056719886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,10240,0.06702666812472872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,10240,0.05638666947682699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,10240,0.036185777849621244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,512,0.012764444781674279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,8192,0.06032622522777981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,8192,0.044072889619403414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,8192,0.031022220849990845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,7168,0.05451288819313049
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,7168,0.03863199883037143
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,7168,0.027104889353116352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,65536,0.16234044233957926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,6144,0.048419555028279625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,6144,0.0343137780825297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,6144,0.02465244465404087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,16384,0.06008266740375095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,5120,0.04577066832118564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,5120,0.028375999795065984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,5120,0.021512000097168818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,12288,0.04330577784114414
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,10240,0.037230223417282104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,4096,0.04050755500793457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,8192,0.03267644511328803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,4096,0.023556444380018447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,4096,0.018178666631380718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,3584,0.03873599900139703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,3584,0.021012443635198805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,3584,0.01682488951418135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,7168,0.02896266513400608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,3072,0.03729155659675598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,3072,0.018512000640233357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,3072,0.015223110715548197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,6144,0.026544888814290363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,2560,0.01605244477589925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,2560,0.013384000294738345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,5120,0.0236435549126731
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,2048,0.03152799937460158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,2048,0.013408888545301227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,2048,0.01275288893116845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,4096,0.019690665933820937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,1536,0.02938666608598497
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,1536,0.011013333168294696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,1536,0.010985777609878115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,3584,0.01798577772246467
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,1024,0.026307556364271376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,1024,0.008380444513426887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,1024,0.00869422240389718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,3072,0.01601777805222405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,768,0.01981511049800449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,768,0.0069751110341813825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,768,0.0069191112286514705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,256,0.010065777434243096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,512,0.021937777598698933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,512,0.004892444445027245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,512,0.006647999915811751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,2560,0.014862222803963555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,256,0.01660977800687154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,256,0.003540444291300244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,256,0.00601244428091579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,65536,128,0.010033778018421596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,2048,0.013007111019558378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,128,0.0029004443850782183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,128,0.005940444353553984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,1536,0.010256888965765635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,64,0.0028284444577164124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,64,0.006245333287451003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,16384,32,0.0028453332682450614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,1024,0.008343111309740279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,16384,32,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,65536,0.23863111601935494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,65536,0.2880604532029894
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,768,0.007592889169851939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,65536,0.1532640059789022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,16384,0.08640622430377537
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,16384,0.06636622216966417
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,512,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,16384,0.04843911197450426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,12288,0.05159911182191637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,12288,0.06880799929300944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,12288,0.03625511129697164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,10240,0.05766666597790188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,10240,0.04371111260520088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,10240,0.030836443106333416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,2560,0.03463110989994473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,8192,0.05168266759978401
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,8192,0.03870844509866502
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,8192,0.02630222174856398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,256,0.006276444428496891
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,7168,0.04610133171081543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,7168,0.035030222601360746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,7168,0.023736000061035156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,16384,128,0.006310222049554189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,6144,0.0410755541589525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,6144,0.030671109755833943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,6144,0.021014221840434607
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,65536,0.15022133456336126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,5120,0.04019822345839606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,5120,0.02677333354949951
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,5120,0.019288889235920377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,16384,0.04463555415471395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,12288,0.036488887336519026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,4096,0.02331377731429206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,4096,0.035616000493367515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,4096,0.015980444020695157
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,3584,0.034296890099843345
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,3584,0.020530665914217632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,3584,0.014576888746685453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,10240,0.03204977843496535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,3072,0.03320444292492337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,3072,0.01850488947497474
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,3072,0.01402933398882548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,8192,0.027782221635182697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,2560,0.030768887864218816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,2560,0.015807999504937064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,2560,0.012423111332787408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,7168,0.02544622288809882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,2048,0.028354667954974707
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,2048,0.013993778162532382
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,2048,0.011341333389282227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,6144,0.022501332892311945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,5120,0.020420443680551317
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,1536,0.025918222135967676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,1536,0.008934222161769867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,1536,0.010046222143703038
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,4096,0.017303110824690927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,1024,0.022621333599090576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,1024,0.007153777612580194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,3584,0.015736889508035447
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,1024,0.007675555845101674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,768,0.019208000765906442
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,768,0.005858666780922148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,768,0.00693866651919153
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,3072,0.014352889524565803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,512,0.019932443896929424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,512,0.003875555677546395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,512,0.006600889066855113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,16384,128,0.015837333268589444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,256,0.016185777054892648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,256,0.0032195556494924757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,256,0.006207110981146495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,2560,0.012992888689041138
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,12288,128,0.0148053334818946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,128,0.003175111073586676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,128,0.005950222412745158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,64,0.002597333242495855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,64,0.005940444353553984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,12288,32,0.002874666617976295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,12288,32,0.005939555664857228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,2048,0.010479999913109673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,65536,0.24922577540079752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,65536,0.20240533351898193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,1536,0.009076444639099969
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,65536,0.14366755220625135
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,1024,0.00797333319981893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,16384,0.08553244670232137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,16384,0.05675644344753689
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,16384,0.04005422194798788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,12288,0.06356355879041883
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,12288,0.044059554735819496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,768,0.00720177756415473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,12288,0.031814220878813006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,10240,0.03837688763936361
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,10240,0.055656890074412026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,10240,0.028059555424584284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,8192,0.04894311229387919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,8192,0.03187111020088196
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,8192,0.02310933338271247
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,512,0.006768888897365994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,7168,0.04588355620702108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,7168,0.02897777822282579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,256,0.00627377794848548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,7168,0.021939555803934734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,6144,0.041535112592909075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,12288,128,0.006058666855096817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,6144,0.026347556047969397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,6144,0.01886311173439026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,5120,0.038559112283918597
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,5120,0.022680888573328655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,5120,0.017120000388887193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,65536,0.1462764475080702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,4096,0.033702221181657575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,4096,0.019701333509551156
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,16384,0.04554933309555054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,4096,0.014442667365074158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,12288,0.03471111257870992
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,3584,0.03249866763750712
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,3584,0.018046221799320646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,3584,0.013448889056841532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,3072,0.030880000856187608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,3072,0.016524444023768108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,10240,0.030014223522610132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,3072,0.012412444584899478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,2560,0.029151111841201782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,2560,0.014910222755538093
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,2560,0.011612444288200803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,8192,0.025941333836979334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,2048,0.027129777603679236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,7168,0.023341332872708637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,2048,0.011603555745548673
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,2048,0.010976889067225985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,6144,0.020425778296258714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,1536,0.009186666872766282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,1536,0.008981333010726506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,5120,0.018848000301255118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,4096,0.01589866644806332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,1024,0.006555555595291986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,3584,0.014808000789748298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,1024,0.007255111303594377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,768,0.01893422173129188
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,768,0.00479644454187817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,768,0.006587555425034628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,3072,0.013414222333166333
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,512,0.01891911029815674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,512,0.003548444559176763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,512,0.006255110932721033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,2560,0.011648000114493899
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,256,0.016283555163277518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,256,0.0031804444475306403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,256,0.005633777628342311
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,2048,0.009847111172146266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,1536,0.008621333373917473
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,128,0.015808888607554965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,128,0.0028106667515304354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,128,0.006243555496136348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,1024,0.007294221884674496
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,64,0.0028568889117903183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,64,0.005950222412745158
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,10240,32,0.0028728888266616394
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,10240,32,0.005782222168313132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,768,0.0069475554757648045
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,65536,0.16933511363135445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,65536,0.20575288931528726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,65536,0.11031822363535564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,512,0.006595555692911148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,16384,0.06800088617536756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,16384,0.04596977763705783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,16384,0.0315324432320065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,12288,0.06146577994028727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,12288,0.035940445131725736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,12288,0.025713778204388086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,256,0.005920889063013925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,10240,0.05317599905861748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,10240,0.03049600124359131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,10240,0.022729777627521094
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,8192,0.04787466592258877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,8192,0.025087111526065405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,8192,0.019547555181715224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,10240,128,0.006220444622966979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,7168,0.042797333664364286
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,7168,0.02194044490655263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,7168,0.017536888519922893
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,65536,0.07907466755972968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,6144,0.038316445218192205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,6144,0.020311999652120803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,6144,0.015850666496488783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,16384,0.02588622272014618
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,5120,0.033407999409569636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,5120,0.017111111018392775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,12288,0.021363554729355708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,5120,0.014741332994567024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,10240,0.01923111081123352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,4096,0.01423822177780999
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,4096,0.012538666526476542
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,8192,0.01681777834892273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,3584,0.03165510959095425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,3584,0.013034666577974955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,3584,0.012071111136012606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,7168,0.014859555496109856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,3072,0.029882666137483384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,3072,0.011680000358157687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,6144,0.014488889111412896
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,3072,0.011040888726711273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,2560,0.028575999869240656
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,2560,0.010120000276300643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,2560,0.010215111076831818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,5120,0.01311911145846049
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,2048,0.024853333830833435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,2048,0.008643555144468943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,2048,0.008965333302815756
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,4096,0.011744000017642975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,1536,0.02103288968404134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,1536,0.0075839997993575195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,1536,0.0069644442862934535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,1536,0.024470221665170457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,3584,0.010701333483060202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,1024,0.019935111204783123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,1024,0.005578666511509154
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,1024,0.006622222148709827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,10240,1024,0.020709334148301017
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,768,0.018583999739752877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,768,0.004576888763242298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,768,0.005964444329341252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,512,0.01756177842617035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,512,0.0039013334446483185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,512,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,3072,0.009369778136412302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,256,0.016181333197487723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,256,0.003216888962520493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,256,0.0052639999323421055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,2560,0.008540444076061249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,2048,0.008057778080304464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,128,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,128,0.005199111170238919
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,1536,0.006992888947327931
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,64,0.00282222218811512
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,64,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,8192,32,0.0028400001012616684
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,1024,0.006638222270541721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,8192,32,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,768,0.005940444353553984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,65536,0.14696889453464085
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,512,0.005930666708283954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,65536,0.2684640089670817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,16384,0.07376711236106025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,65536,0.09383733405007257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,16384,0.04217777649561564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,256,0.005394666559166378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,16384,0.02883466747072008
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,12288,0.03348533312479655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,12288,0.060340444246927895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,12288,0.023680888944202002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,10240,0.05134488807784187
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,10240,0.028961777687072754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,10240,0.020607999629444547
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,8192,128,0.005561777700980504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,8192,0.045427554183536105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,8192,0.025408888856569927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,8192,0.017449777987268236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,65536,0.07387110922071669
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,7168,0.0409457782904307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,7168,0.01957066688272688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,16384,0.02360800074206458
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,7168,0.01588711142539978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,12288,0.018963555494944256
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,6144,0.037300444311565824
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,6144,0.018627555833922494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,6144,0.014743111199802823
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,5120,0.033301333586374916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,5120,0.014940443966123792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,5120,0.013406222065289816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,10240,0.016530666086408827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,4096,0.0310444434483846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,4096,0.01309511148267322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,4096,0.011720889144473605
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,3584,0.03029688861634996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,3584,0.011643555429246692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,3584,0.010994666980372535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,8192,0.014783110883500842
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,3072,0.02920888861020406
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,3072,0.010648888846238455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,3072,0.010780444575680627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,7168,0.012895999683274163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,6144,0.01258666647805108
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,2560,0.026390223039521113
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,2560,0.00979644474056032
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,2560,0.00963199966483646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,2048,0.02296977738539378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,5120,0.010991999672518836
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,2048,0.008227555288208855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,2048,0.007896888587209914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,4096,0.03154844376775954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,1536,0.020576889316240948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,4096,0.009703999592198266
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,1536,0.006948444578382704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,1536,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,1024,0.004873777843183941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,1024,0.018408889571825664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,1024,0.005919999960396025
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,768,0.018252443936136033
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,768,0.0039022221333450745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,3584,0.008381333616044786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,768,0.005954666684071223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,512,0.016480889585283067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,512,0.0038853333228164246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,512,0.005315555466545953
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,3072,0.008290666672918532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,256,0.01677955521477593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,256,0.0029333333174387612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,256,0.005560889012283749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,2560,0.007664889097213745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,7168,128,0.015104888214005364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,128,0.002834666727317704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,2048,0.0069306666652361555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,128,0.005271111097600725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,64,0.0028248888750871024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,1536,0.00654666663871871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,64,0.005554666535721884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,7168,32,0.0028782222006056043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,7168,32,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,1024,0.005952000204059813
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,65536,0.12462488810221355
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,768,0.0058062221441004015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,65536,0.18455199400583902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,65536,0.08326755629645453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,512,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,16384,0.03868888815244039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,16384,0.07221244441138373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,16384,0.025598223010698955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,12288,0.02986933456526862
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,12288,0.05250844359397888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,12288,0.020652444826232064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,10240,0.026343999637497797
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,10240,0.04619644416703118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,10240,0.018909333480728995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,256,0.005336888962321811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,8192,0.043150222963756986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,8192,0.02198133369286855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,8192,0.016396444704797532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,7168,0.0388026667965783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,7168,0.019837333096398246
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,8192,128,0.015089778436554803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,7168,0.01499644418557485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,6144,0.03526844580968221
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,7168,128,0.0052133335007561584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,6144,0.01830311119556427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,6144,0.01407911048995124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,5120,0.03457511133617825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,5120,0.01625155574745602
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,65536,0.07188622156778972
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,5120,0.012432888978057437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,16384,0.022329777479171753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,4096,0.030639999442630347
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,4096,0.012086222569147745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,12288,0.018200000127156574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,4096,0.010808000134097205
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,10240,0.015850666496488783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,3584,0.029828445778952703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,3584,0.01072533345884747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,3584,0.010664888554149203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,3072,0.02739199995994568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,3072,0.009716444545321995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,8192,0.014203555054134794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,3072,0.009361777868535783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,2560,0.025238222546047632
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,2560,0.00868888861603207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,2560,0.008309333688682979
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,7168,0.012642666697502136
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,2048,0.022280888424979314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,2048,0.007824888659848107
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,6144,0.011738667057620155
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,2048,0.0069511110583941145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,1536,0.019983111156357657
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,1536,0.006340444501903322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,1536,0.006561777657932705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,5120,0.010506666368908353
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,1024,0.018016889691352844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,1024,0.004523555437723796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,1024,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,4096,0.008994667066468133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,768,0.018162666095627677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,3584,0.008095111283991072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,768,0.0038737778862317405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,768,0.005606222069925732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,512,0.016335111525323655
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,512,0.003256888853179084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,512,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,3072,0.00794488853878445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,256,0.0031679999083280563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,256,0.016136889656384785
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,256,0.005421333428886201
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,2560,0.006979555719428592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,128,0.0028648889727062653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,6144,128,0.01482311056719886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,128,0.0053004444473319584
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,64,0.0028604444944196274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,64,0.005223999834722943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,2048,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,6144,32,0.0028444443725877335
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,6144,32,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,1536,0.006575999988449945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,65536,0.10573777887556289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,65536,0.20795378420088026
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,1024,0.0058782220714622075
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,65536,0.07655377520455255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,16384,0.03235200047492981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,16384,0.0709644423590766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,16384,0.02404088940885332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,768,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,12288,0.026299556096394856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,12288,0.059402664502461754
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,12288,0.019338667392730713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,512,0.005625777774386936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,10240,0.023240889112154644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,10240,0.05047288868162367
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,10240,0.01757244434621599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,256,0.005327111317051782
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,8192,0.020579554968410067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,8192,0.04235644472969902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,8192,0.015458666616015963
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,6144,128,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,7168,0.03831733266512553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,7168,0.017909333109855652
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,7168,0.014084445105658637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,6144,0.032200889454947576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,6144,0.016777777009540133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,6144,0.013234666652149625
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,5120,0.028973334365420874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,65536,0.06907021999359131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,5120,0.014662222729788886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,5120,0.011963555382357704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,4096,0.02920177910063002
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,16384,0.02164266672399309
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,4096,0.011195555329322815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,4096,0.01036800030204985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,3584,0.02757866680622101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,3584,0.009630222287442949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,12288,0.017880888448821176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,3584,0.009296000003814697
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,3072,0.025285333395004272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,10240,0.015268445014953613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,3072,0.00902844468752543
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,3072,0.008600888980759515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,8192,0.013785777820481194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,2560,0.023641778363121882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,2560,0.008007110820876228
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,2560,0.007239999870459239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,7168,0.012144889268610211
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,2048,0.02108977735042572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,2048,0.0070053330726093715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,2048,0.006687110910813014
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,6144,0.011141333315107556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,1536,0.018584888842370775
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,1536,0.005220444252093633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,1536,0.006233777850866318
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,5120,0.009923555784755284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,1024,0.01683288812637329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,1024,0.0041786668201287585
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,4096,0.007982222570313348
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,1024,0.006035555568006303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,768,0.01752622259987725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,768,0.0038622220357259116
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,768,0.005576888720194499
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,3584,0.007918222082985772
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,512,0.015503111812803479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,512,0.003265777809752358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,3072,0.007315555380450354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,512,0.00554933316177792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,2560,0.00702133360836241
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,256,0.016189333465364244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,256,0.0031066667288541794
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,256,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,5120,128,0.01477866702609592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,128,0.00282577777074443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,2048,0.006598222172922558
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,128,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,64,0.0025253333151340485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,64,0.005567111074924469
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,5120,32,0.0028044444819291434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,5120,32,0.005532444351249271
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,1536,0.006302222195598815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,65536,0.08810755279329087
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,65536,0.07033244768778484
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,65536,0.19468888971540663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,1024,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,16384,0.02782400117980109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,16384,0.02169422143035465
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,16384,0.06972977850172254
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,12288,0.04842577709092034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,12288,0.022014222211307947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,768,0.005604444278611078
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,12288,0.018216888109842937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,512,0.005430222385459476
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,10240,0.046816888782713145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,10240,0.01938311093383365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,10240,0.01646133263905843
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,256,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,8192,0.016262221667501662
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,8192,0.03897688786188761
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,8192,0.015096889601813423
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,5120,128,0.00526577772365676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,7168,0.015172445111804538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,7168,0.03572266631656223
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,7168,0.013162666724787818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,6144,0.03245866629812453
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,6144,0.014402666025691561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,6144,0.012523555921183692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,65536,0.06735022200478448
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,5120,0.029291556941138372
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,5120,0.01239733315176434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,5120,0.011345778074529437
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,16384,0.020941333638297185
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,4096,0.02777155571513706
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,4096,0.011350221931934357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,4096,0.009543110926946005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,12288,0.017281777328915067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,3584,0.012025777664449481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,3584,0.026221333278550044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,3584,0.008550222549173567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,10240,0.014947555131382413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,3072,0.024726221958796184
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,3072,0.011349333657158745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,3072,0.00795111142926746
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,8192,0.013238222234778933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,2560,0.01034755590889189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,2560,0.023316444622145757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,2560,0.007275555696752336
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,7168,0.011352889239788055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,2048,0.020586666133668687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,2048,0.009015999734401703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,6144,0.00997688869635264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,2048,0.006630222416586346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,1536,0.018561777141359117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,1536,0.006983111302057902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,1536,0.006098666538794835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,1024,0.016533333394262526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,5120,0.008656000097592672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,1024,0.00590844452381134
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,1024,0.005903999838564131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,768,0.018223111828168232
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,4096,0.007987555530336168
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,768,0.00462755560874939
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,768,0.005552000055710475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,512,0.014477334088749357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,3584,0.007659555309348636
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,512,0.0041751112374994485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,512,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,256,0.014831110835075378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,3072,0.007278222176763747
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,256,0.003216000066863166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,256,0.005428444594144821
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,2560,0.0069653333889113525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,4096,128,0.01477777792347802
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,2048,0.006649777707126405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,128,0.0028328889360030494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,128,0.005412444472312927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,64,0.0028880000528362063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,64,0.005285333428117964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,1536,0.006302222195598815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,4096,32,0.0025902222841978073
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,4096,32,0.005564444594913059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,1024,0.005968889014588461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,65536,0.07933421929677327
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,65536,0.06771822108162774
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,768,0.005593777944644292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,65536,0.19178933567470977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,16384,0.02534399926662445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,512,0.005366222312053044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,16384,0.021181333396169875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,12288,0.02032977839310964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,12288,0.051757332351472646
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,12288,0.01776266594727834
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,256,0.005234666582610872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,10240,0.04609510964817471
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,10240,0.017481777403089736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,4096,128,0.0052791109515561005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,10240,0.016003555721706815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,8192,0.015789333317014907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,8192,0.042430222034454346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,8192,0.014582221706708273
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,7168,0.037674665451049805
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,65536,0.0660711129506429
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,7168,0.0140346669488483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,7168,0.012990222209029727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,16384,0.020968000094095867
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,6144,0.03627644313706292
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,6144,0.012401777837011548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,6144,0.012063999970753988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,12288,0.01687111126052009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,5120,0.011635555161370171
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,5120,0.010767999622556897
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,10240,0.014765333798196582
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,4096,0.0273626661962933
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,4096,0.010650667051474253
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,4096,0.00867733359336853
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,8192,0.012484444512261285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,3584,0.025591111845440332
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,3584,0.012025777664449481
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,3584,0.00830488900343577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,3072,0.02330933345688714
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,3072,0.009977777798970541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,7168,0.010770666930410596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,3072,0.007325333025720384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,2560,0.02198666665289137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,2560,0.008953777452309927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,2560,0.0069217777086628815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,6144,0.009551111194822524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,2048,0.021194666624069214
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,5120,0.008664000365469191
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,2048,0.008284444610277811
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,2048,0.00656444455186526
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,1536,0.01794488893614875
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,1536,0.006242666807439592
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,4096,0.008192000289758047
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,1536,0.006232000059551663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,3584,0.007664889097213745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,1024,0.0052764444715446895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,1024,0.006250666661394968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,3072,0.007377777662542131
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,768,0.018069333500332303
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,768,0.004592888885074192
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,768,0.0059573331640826324
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,2560,0.006919999917348226
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,512,0.01442488862408532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,512,0.003891555385457145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,2048,0.006603555546866522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,512,0.0052355556852287715
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,256,0.014671110444598727
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,256,0.0028657778683635923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,1536,0.006301333506902059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,256,0.0052533331844541765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,1024,0.005909333212508096
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,128,0.014073777529928418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,128,0.002833777831660377
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,128,0.005294222384691238
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,64,0.0025128889828920364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,64,0.005203555441564984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3584,32,0.002801777794957161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3584,32,0.005214222189452913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,768,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,65536,0.06897599829567803
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,65536,0.19400621785057914
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,512,0.005537777725193236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,65536,0.06496622165044148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,16384,0.06191555658976237
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,256,0.005256888767083486
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,16384,0.022277333670192297
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,16384,0.020852444900406733
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3584,128,0.0052595556610160405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,12288,0.04647377795643277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,12288,0.018236445056067575
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,12288,0.017529777354664274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,10240,0.04330844349331326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,10240,0.016135111451148987
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,10240,0.015541333291265698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,65536,0.0647875534163581
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,8192,0.03983911209636264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,16384,0.020624889267815482
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,8192,0.013833777772055732
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,8192,0.013785777820481194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,7168,0.034203555848863386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,12288,0.01681600014368693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,7168,0.012691555751694573
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,7168,0.01298577752378252
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,6144,0.031140443351533677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,6144,0.01184622198343277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,10240,0.01420088940196567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,6144,0.011741333537631564
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,5120,0.027335999740494624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,5120,0.011020444333553314
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,5120,0.010237333675225576
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,8192,0.011753777662913004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,4096,0.02605777813328637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,4096,0.010117332968446944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,7168,0.00999022192425198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,4096,0.007786666353543599
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,3584,0.025012445118692186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,3584,0.009106666677527958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,6144,0.009317333499590555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,3584,0.0076657773719893555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,3072,0.023043556345833674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,3072,0.009337777892748514
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,3072,0.007304888632562425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,5120,0.00868622213602066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,2560,0.021293333835071985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,2560,0.008002666963471307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,2560,0.006976000136799282
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,4096,0.008110221889283922
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,2048,0.019648000597953796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,2048,0.0069617778062820435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,2048,0.006599999964237213
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,3584,0.007713778151406183
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,1536,0.017709333035680983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,1536,0.006576889091067844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,1536,0.006272000157170826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,1024,0.01612622208065457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,1024,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,3072,0.007256888680987888
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,1024,0.005968889014588461
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,768,0.017642666896184284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,768,0.004884444591071871
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,2560,0.006968888971540663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,768,0.00564444437623024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,512,0.014426666829321118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,2048,0.0069306666652361555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,512,0.004182222402758068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,512,0.00553955551650789
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,256,0.01516266663869222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,1536,0.006595555692911148
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,256,0.00351200004418691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,256,0.005221333354711533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3072,128,0.014820444915029736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,1024,0.005850666513045629
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,128,0.0028595555987623003
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,128,0.005208888815508948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,64,0.0028000000036425064
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,64,0.005215999980767568
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,3072,32,0.0027946666296985415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,768,0.005632888939645555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,3072,32,0.004927999857399198
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,65536,0.05922310882144504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,512,0.005390222287840313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,65536,0.06258933411704169
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,256,0.00526577772365676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,16384,0.020054222808943856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,16384,0.06320622232225206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,3072,128,0.0052168890833854675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,16384,0.020217777954207528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,12288,0.04424266682730781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,12288,0.016192000773217943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,12288,0.017208000024159748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,10240,0.03866666555404663
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,65536,0.06468266910976835
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,10240,0.014394667413499622
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,10240,0.015424888994958667
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,8192,0.034259554412629865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,8192,0.012483555409643384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,16384,0.020232889387342665
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,8192,0.013502221968438892
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,16384,0.06565688716040717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,7168,0.031529777579837374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,7168,0.011743110915025076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,12288,0.016308445069524977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,7168,0.01258488910065757
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,6144,0.010718222293588849
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,6144,0.011048888994587792
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,10240,0.013740444348918067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,5120,0.025701334079106648
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,5120,0.010011555420027839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,5120,0.009405333134863112
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,8192,0.011023999916182624
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,5120,0.03206844462288751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,7168,0.010024000373151567
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,4096,0.00926399976015091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,4096,0.007983110845088959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,6144,0.009349333743254343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,3584,0.022310222188631695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,3584,0.008651555412345463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,3584,0.007670222057236566
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,5120,0.00831288927131229
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,3072,0.022618666291236877
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,3072,0.008433777425024245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,4096,0.00793688909875022
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,3072,0.007304000357786815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,2560,0.02090311050415039
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,2560,0.007228444847795699
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,2560,0.0069102222720781965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,3584,0.007687999970383114
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,2048,0.018566222654448617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,2048,0.006555555595291986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,2048,0.006630222416586346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,1536,0.01687466601530711
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,3072,0.007272000114123027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,1536,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,1536,0.005943111247486538
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,3584,1024,0.016522667474216886
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,1024,0.015485333071814643
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,1024,0.0051893335249688895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,1024,0.0058462222417195635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,2560,0.006899555524190267
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,768,0.017179555363125272
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,768,0.00453599997692638
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,768,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,512,0.01348088930050532
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,512,0.003891555385457145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,2048,0.006621333460013072
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,512,0.00527022240890397
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,256,0.014502222339312235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,1536,0.006520888871616787
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,256,0.0032355555643637977
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,256,0.005260444349712796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,128,0.01478666729397244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,1024,0.005921777751710679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,128,0.00318755561278926
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,128,0.005271111097600725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,64,0.002532444480392668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,768,0.005542222410440445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,64,0.0052595556610160405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2560,32,0.002522666628162066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2560,32,0.005250666704442766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,512,0.005347555710209741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,65536,0.049880888726976186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,65536,0.18445244100358751
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,256,0.004940444396601783
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,65536,0.06069688664542305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,16384,0.06186044216156006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,16384,0.016912000046836007
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,16384,0.019914666811625164
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2560,128,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,12288,0.013615999784734515
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,12288,0.050754666328430176
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,12288,0.016905777984195285
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,10240,0.012102222277058495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,10240,0.04494577646255493
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,10240,0.015207111835479736
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,65536,0.06353244516584608
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,8192,0.011367999845080905
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,8192,0.04082577758365207
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,8192,0.012827555338541666
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,16384,0.019700444406933255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,7168,0.010303110712104374
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,12288,0.014761777387724983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,7168,0.011350221931934357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,6144,0.030530667967266504
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,10240,0.012453333371215396
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,6144,0.009783999787436591
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,6144,0.010383110907342698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,5120,0.02612888813018799
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,8192,0.01074222226937612
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,5120,0.008982222113344405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,5120,0.008973333570692275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,4096,0.02643377747800615
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,7168,0.010019555687904358
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,4096,0.010671111444632212
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,4096,0.007923555870850882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,3584,0.025054223007626
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,6144,0.009018667042255402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,3584,0.00958933350112703
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,3584,0.007602666815121968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,3072,0.022188444932301838
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,3072,0.006974221931563483
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,5120,0.008427555362383524
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,3072,0.007262222468852997
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,2560,0.020928000410397846
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,4096,0.00793866647614373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,2560,0.00630044440428416
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,2560,0.006973333656787872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,2048,0.020026667250527274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,2048,0.00794400026400884
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,3584,0.007701333198282454
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,2048,0.006688000013430913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,3072,0.007267555428875818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,1536,0.017311111092567444
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,1536,0.006634666687912411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,2560,0.006913777854707506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,1536,0.0059368887709246735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,1024,0.015494222442309061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,1024,0.0052933332820733385
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,1024,0.005935110979610019
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,2048,0.006689777804745569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,768,0.017842666970358956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,768,0.004543999830881755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,768,0.005559111220969095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,1536,0.006223111102978389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,512,0.013459555804729462
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,512,0.004232888834344016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,512,0.005193777796294955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,1024,0.0059004442559348206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,256,0.014496000276671516
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,256,0.0032213332338465583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,256,0.0052284445199701525
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,768,0.005608888963858287
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,512,0.005210666606823604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,128,0.0031635556370019913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,128,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,256,0.0052177777720822235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,64,0.0028479999552170434
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,64,0.005251555393139522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,2048,32,0.00286666676402092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,2048,32,0.005261333452330695
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,2048,128,0.005091555416584015
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,65536,0.17134755187564424
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,65536,0.03758222195837233
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,65536,0.05899555815590752
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,16384,0.013847110999955071
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,16384,0.06053688791063097
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,16384,0.019315555691719055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,65536,0.15741777420043945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,12288,0.043304890394210815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,12288,0.01184088902340995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,12288,0.01646222174167633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,65536,0.06291377544403076
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,10240,0.03727555606100295
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,10240,0.010371555884679159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,10240,0.014184888866212634
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,16384,0.01824711097611321
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,8192,0.03289244572321574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,8192,0.01038755559258991
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,8192,0.010713777608341642
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,12288,0.013780444032616086
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,7168,0.030202666918436687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,7168,0.009862222605281407
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,7168,0.009993777506881291
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,10240,0.011867555479208628
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,6144,0.027442667219373915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,6144,0.008981333010726506
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,6144,0.009345778160625035
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,8192,0.010746666954623329
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,5120,0.024676443801985845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,5120,0.007928888830873702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,6144,0.027284445034133062
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,5120,0.008942222429646386
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,4096,0.007633777956167857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,7168,0.010033778018421596
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,4096,0.02369333306948344
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,4096,0.007640000018808577
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,3584,0.022839999861187402
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,3584,0.007300444775157505
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,6144,0.008994667066468133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,3584,0.007668444679843054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,3072,0.02181511123975118
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,5120,0.008458666503429413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,3072,0.006569777925809224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,3072,0.007232888705200619
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2560,4096,0.024691555235120986
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,2560,0.019893333315849304
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,2560,0.005944889038801193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,4096,0.008043555749787224
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,2560,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,3584,0.007711999946170383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,2048,0.01889866590499878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,2048,0.006588444527652528
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,2048,0.0062977779242727495
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,3072,0.007307555940416124
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,1536,0.01681600014368693
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,1536,0.005342222336265776
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,1536,0.006235555642180973
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,2560,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,1024,0.015114666687117683
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,1024,0.004597333156400257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,1024,0.0058871110280354815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,2048,0.006603555546866522
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,768,0.01647200021478865
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,768,0.004242666479614046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,768,0.005558222118351195
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,512,0.0035955554081334006
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,1536,0.006248888870080312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,512,0.013389333254761167
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,512,0.005210666606823604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,256,0.0028968888024489083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,1024,0.005892444401979446
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,256,0.014411555396185981
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,256,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,768,0.005580444302823808
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1536,128,0.013818666338920593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,128,0.0029048888633648553
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,128,0.005371555685997009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,512,0.005271111097600725
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,64,0.0025048889219760895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1536,32,0.0028684443483750024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,64,0.005264889034960005
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1536,32,0.005280888742870755
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,256,0.005234666582610872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,65536,0.026524444421132404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,65536,0.13200978438059488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,65536,0.05748889181349012
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1536,128,0.005245333330498801
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,16384,0.015166223049163818
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,16384,0.017647110753589206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,16384,0.05047466688685947
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,12288,0.013050666285885705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,12288,0.03772888912094964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,12288,0.013728000223636627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,10240,0.011762667033407422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,10240,0.03304977880583869
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,10240,0.012358222570684222
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,65536,0.06291466951370239
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,16384,0.01721333298418257
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,8192,0.029770665698581274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,8192,0.010689777632554373
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,8192,0.010632000035709806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,12288,0.013406222065289816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,7168,0.009415999882751042
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,7168,0.026335999369621277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,10240,0.012076444096035428
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,7168,0.010004444254769219
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,6144,0.024305777417288885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,8192,0.010764444039927589
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,6144,0.008691555923885768
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,6144,0.009388444324334463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,5120,0.021988444858127173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,5120,0.007977777885066139
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,7168,0.009470221896966299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,5120,0.0086666668454806
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,4096,0.021292444732454088
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,4096,0.007947555846638149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,6144,0.009101333717505137
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,4096,0.007921777665615082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,5120,0.008612444831265343
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,3584,0.01976977785428365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,3584,0.0069271110826068455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,3584,0.007602666815121968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,3072,0.006250666661394968
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,4096,0.007972444097201029
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,3072,0.018575999471876357
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,3072,0.007269333634111617
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,2560,0.018241778016090393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,3584,0.007664889097213745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,2560,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,2560,0.007000889215204451
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,2048,0.017543110582563613
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,2048,0.0075866662793689305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,3072,0.007257777783605788
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,2048,0.006279110908508301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,2560,0.006743999818960826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,1536,0.015673778123325773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,1536,0.006036444256703059
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,2048,0.006288888967699475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,1536,0.006238222122192383
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,1024,0.01443733274936676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,7168,0.03208177619510227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,1024,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,1024,0.005889777921968036
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,1536,0.005613333235184352
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,768,0.016318221886952717
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,768,0.00452622233165635
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,768,0.005632888939645555
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,1024,0.005314666777849197
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,512,0.013053333593739403
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,512,0.0038968887594011095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,512,0.005442666510740916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,256,0.014109333356221518
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,768,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,256,0.0031822222388452957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,256,0.005487999982304043
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,512,0.005225777626037598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,1024,128,0.014497778481907315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,64,0.0052720002002186244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,128,0.00286666676402092
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,128,0.005434666656785541
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,128,0.004898666507667965
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,64,0.002801777794957161
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,1024,256,0.004892444445027245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,1024,32,0.0028311111446883944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,1024,32,0.0052639999323421055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,16384,0.036899556716283165
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,65536,0.021747555997636583
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,65536,0.09586311048931545
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,65536,0.05662755833731758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,16384,0.010407110883129967
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,16384,0.016529776983790927
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,65536,0.0626408921347724
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,12288,0.030015110969543457
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,12288,0.011072888970375061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,12288,0.013108444710572561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,16384,0.016179554992251925
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,10240,0.02573066618707445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,10240,0.009700444837411245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,10240,0.012056888805495368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,8192,0.01070577816830741
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,12288,0.013412444127930535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,8192,0.023932443724738225
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,8192,0.00904977818330129
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,10240,0.011780444118711682
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,7168,0.022662222385406494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,7168,0.008647999829716152
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,7168,0.00976533359951443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,8192,0.010404444403118556
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,6144,0.020870221985710993
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,6144,0.007984889050324758
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,6144,0.009306666751702627
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,7168,0.009689778089523315
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,5120,0.019428445233239066
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,5120,0.00666844430896971
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,5120,0.008784000244405534
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,2048,128,0.014789332946141561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,6144,0.008793777889675563
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,4096,0.01831466621822781
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,4096,0.007304888632562425
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,3584,0.007653333246707916
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,4096,0.008004444340864817
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,4096,0.008014221986134848
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,5120,0.008343111309740279
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,3584,0.017836444907718234
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,3584,0.0069146665434042616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,3072,0.017156443662113614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,3072,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,3072,0.007295110987292395
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,3584,0.007180444068378872
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,2560,0.016180444094869826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,2560,0.006239111224810283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,2560,0.007283555964628856
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,3072,0.00721066693464915
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,2048,0.016167110866970487
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,2048,0.004912888838185204
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,1536,0.005922666854328579
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,2048,0.006589333216349284
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,2560,0.006660444455014334
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,1536,0.01443733274936676
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,1536,0.005207111024194294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,2048,0.006196444647179709
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,1024,0.004256000121434529
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,1024,0.005566222386227713
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,1536,0.0057546666098965546
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,768,0.0158906661801868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,768,0.0038533334930737815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,512,0.005369777894682354
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,768,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,1024,0.005391110976537068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,512,0.013070222404268054
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,256,0.0053484443989064955
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,512,0.003551111039188173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,768,0.00555644432703654
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,128,0.0028773333049482773
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,256,0.014769777655601501
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,256,0.002887111157178879
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,512,0.005233777893914117
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,128,0.013140444126394061
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,128,0.005277333160241445
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,32,0.0052666668262746595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,256,0.005236444373925527
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,64,0.002830222249031067
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,768,64,0.005219555563396878
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,768,32,0.002521777732504739
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,768,128,0.004924444274769889
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,16384,0.03498311175240411
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,65536,0.01686844395266639
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,65536,0.07675466934839885
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,65536,0.05607199668884277
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,12288,0.008640888664457533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,16384,0.011119999819331698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,16384,0.0155404441886478
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,65536,0.056899554199642606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,10240,0.025264889001846313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,10240,0.01240799989965227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,12288,0.032652444309658475
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,16384,0.016171556380059984
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,8192,0.008001777860853408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,12288,0.013412444127930535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,10240,0.007722666694058313
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,12288,0.013427555561065674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,7168,0.007663999994595845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,8192,0.01940088967482249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,8192,0.010371555884679159
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,10240,0.011707555916574268
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,7168,0.018200000127156574
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,8192,0.01055466632048289
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,7168,0.010144888526863521
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,6144,0.01749955614407857
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,6144,0.007246221933099959
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,7168,0.009346666435400644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,6144,0.009527111219035255
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,5120,0.01920977731545766
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,5120,0.006923555499977536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,5120,0.008957333034939235
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,6144,0.009196444518036312
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,4096,0.017847110827763874
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,4096,0.007313778003056844
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,5120,0.008296888735559251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,3584,0.007310222420427535
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,4096,0.007831110722488826
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,4096,0.00796088907453749
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,3584,0.015132443772421943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,3584,0.006631111105283101
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,3072,0.016359999775886536
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,3072,0.00657066661450598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,3072,0.007639110916190677
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,3584,0.007259555160999298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,2560,0.015815110670195687
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,2560,0.006641777853171031
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,2560,0.007239110767841339
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,3072,0.007265778051482306
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,2048,0.01516888870133294
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,2048,0.005611555443869696
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,2560,0.0069146665434042616
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,2048,0.006265777680608962
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,1536,0.013436444103717804
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,1536,0.004570666700601578
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,1536,0.005978666659858491
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,1536,0.006105777704053455
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,2048,0.006362666686375936
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,1024,0.012795555922720166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,1024,0.003871110992299186
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,1024,0.0052586665583981415
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,768,0.015125332607163323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,768,0.0034906665484110513
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,1024,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,768,0.0052684446175893145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,512,0.011749332977665795
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,512,0.0031822222388452957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,512,0.00554666668176651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,768,0.005617777920431561
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,256,0.01346933344999949
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,256,0.0028657778683635923
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,512,0.005239999956554837
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,256,0.00554666668176651
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,512,128,0.01312622262371911
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,64,0.005385777602593104
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,32,0.0024862223201327855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,128,0.0028684443483750024
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,128,0.0052488889131281115
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,256,0.005208888815508948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,512,64,0.0025333333760499954
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,512,32,0.0052693333062860705
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,512,128,0.004943111290534337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,65536,0.01551555593808492
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,65536,0.07300711340374418
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,65536,0.04899377624193827
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,16384,0.008627555436558193
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,16384,0.026538666751649644
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,65536,0.050965332322650485
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,16384,0.015818667080667283
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,12288,0.0229751103454166
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,16384,0.016192888220151264
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,10240,0.012063999970753988
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,12288,0.007725333174069722
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,12288,0.013174222575293647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,12288,0.013384888569513956
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,10240,0.007279111279381647
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,10240,0.02056711084312863
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,8192,0.01886577738655938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,7168,0.006274666637182236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,8192,0.006638222270541721
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,10240,0.011753777662913004
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,8192,0.010910222099887
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,6144,0.006575111299753189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,7168,0.017853332890404593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,8192,0.01036800030204985
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,7168,0.01000711073478063
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,6144,0.016728000508414376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,7168,0.009638222555319468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,6144,0.009523555636405945
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,5120,0.01615111033121745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,5120,0.00830844458606508
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,6144,0.009024889104896123
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,5120,0.006223111102978389
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,5120,0.009030222064918941
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,4096,0.015388444066047668
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,4096,0.005711110929648082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,4096,0.007612444460391998
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,3584,0.014832889040311178
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,4096,0.00742133292886946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,3072,0.006991110742092133
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,3584,0.005960888746711943
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,3584,0.007609777980380588
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,3584,0.007423111134105259
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,3072,0.0058871110280354815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,2560,0.013791110780504016
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,3072,0.006912888752089606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,2560,0.006264888991912206
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,2560,0.0069324444565508105
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,2048,0.013239111337396832
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,2560,0.006895111252864202
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,2048,0.005610666755172942
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,2048,0.006263111200597551
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,1536,0.01239377756913503
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,2048,0.00664444433318244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,1536,0.004334222111437055
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,1024,0.0038373333712418876
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,1536,0.005991999887757831
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,1024,0.011705777711338468
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,1536,0.00591111100382275
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,1024,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,768,0.013125333521101209
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,1024,0.005877333382765452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,768,0.003548444559176763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,768,0.0052800000541739995
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,512,0.01070666644308302
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,768,0.005535111245181825
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,512,0.0031822222388452957
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,512,0.0052266667286554975
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,256,0.011380444798204633
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,256,0.002867555452717675
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,512,0.0052577778697013855
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,256,0.0052373334765434265
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,128,0.010339555641015371
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,256,0.004927111168702443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,128,0.0025111111915773815
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,128,0.004914666629499859
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,64,0.0025519999778932994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,256,128,0.005218666460778978
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,64,0.005208888815508948
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,256,32,0.0025271111064487034
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,256,32,0.0052435555391841466
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,65536,0.04695377747217814
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,65536,0.01408088869518704
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,65536,0.06970755259195964
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,16384,0.02642400066057841
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,65536,0.04994044370121426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,16384,0.007318221860461765
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,16384,0.015503111812803479
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,16384,0.016490666402710807
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,12288,0.021330666210916307
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,12288,0.006959999601046245
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,12288,0.013073777986897362
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,12288,0.013167111410035027
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,10240,0.0196106665664249
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,8192,0.00600355573826366
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,10240,0.0069715554515520734
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,10240,0.012056888805495368
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,10240,0.0120319997270902
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,8192,0.017829333742459614
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,8192,0.010337777435779572
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,8192,0.010383110907342698
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,7168,0.017343999611006845
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,7168,0.00629333323902554
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,7168,0.010089777410030365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,7168,0.009359999663299983
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,6144,0.015840889679061044
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,6144,0.005928888916969299
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,6144,0.009325332939624786
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,768,1024,0.014133334159851074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,6144,0.009002666506502364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,5120,0.015492444237073263
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,4096,0.007648000286685095
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,5120,0.005886222339338726
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,5120,0.00868000007337994
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,5120,0.008271111382378472
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,4096,0.01479822231663598
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,4096,0.0052728888889153796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,3584,0.013456888496875763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,4096,0.00775022225247489
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,3584,0.006225777582989798
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,3584,0.007594666547245449
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,3584,0.007153777612580194
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,3072,0.013799111048380533
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,3072,0.005717333406209946
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,3072,0.007270221908887227
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,3072,0.007251555720965068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,2560,0.012157333393891653
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,2560,0.006175999840100606
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,2560,0.006950221955776215
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,2560,0.006924444602595435
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,2048,0.012616888516479068
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,2048,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,2048,0.006568000134494569
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,2048,0.006229333165619109
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,1536,0.012049777640236748
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,1536,0.004560889055331548
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,1536,0.006256000035338932
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,1536,0.005997333261701796
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,1024,0.010748444332016839
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,768,0.005577777822812398
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,1024,0.003551111039188173
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,1024,0.00555288874440723
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,1024,0.005574222240183089
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,768,0.012092444631788464
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,768,0.00351200004418691
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,512,0.01071999967098236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,768,0.005244444641802046
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,512,0.0031902222997612427
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,512,0.004928888960017098
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,256,0.010603555374675326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,256,0.005246222019195557
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,512,0.005196444276306364
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,256,0.0028391112056043413
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,256,0.0052595556610160405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,64,0.002519999941190084
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,128,0.0028560000161329904
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,128,0.005229333208666907
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,nvfp4,1,128,128,0.004933333231343163
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,64,0.005596444424655702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,128,32,0.002534222271707323
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,128,32,0.004909333255555895
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,65536,0.010021333065297868
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,65536,0.04652533266279432
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,16384,0.006614222294754452
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,16384,0.01586577792962392
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,10240,0.012093333734406365
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,12288,0.0059368887709246735
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,12288,0.013433777623706393
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,10240,0.005782222168313132
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,8192,0.006630222416586346
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,8192,0.01070844464831882
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,7168,0.005592000153329637
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,5120,0.006037333359320958
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,7168,0.010063110954231685
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,6144,0.005614222337802251
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,4096,0.007748444047239091
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,6144,0.009319110876984065
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,5120,0.008658666577604082
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,4096,0.0052666668262746595
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,3584,0.0059075554211934405
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,3584,0.00777333312564426
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,3072,0.0059164443777667145
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,3072,0.0069777775141927935
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,2560,0.005951111101441913
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,2560,0.006928000185224745
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,2048,0.005603555589914322
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,1024,0.0035573335157500375
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,1024,0.005581333405441708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,2048,0.006582222051090664
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,1536,0.004562666846646203
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,1536,0.006231110956933763
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,768,0.0032151111712058387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,768,0.005588444570700328
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,512,0.003177777760558658
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,256,0.0028533333291610083
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,512,0.005199999858935674
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,256,0.0052720002002186244
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,128,0.0026568888376156488
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,128,0.005573333137565189
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,64,0.002556444456179937
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,64,0.0052151112920708126
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,64,32,0.0025315555847353404
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,16384,0.015411555767059326
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,64,32,0.005252444495757421
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,65536,0.00885511106914944
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,65536,0.04655199911859301
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,16384,0.00658044425977601
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,12288,0.00638755535085996
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,12288,0.013150222599506378
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,10240,0.005602666487296422
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,10240,0.012039999994966718
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,8192,0.005471999860472149
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,8192,0.010747555229398938
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,7168,0.005586666779385672
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,7168,0.00979733301533593
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,6144,0.00563466673096021
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,6144,0.00925333301226298
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,5120,0.005583111196756363
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,5120,0.008749333520730337
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,4096,0.0052560000783867305
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,4096,0.007654222349325816
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,3584,0.005718222094906702
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,3584,0.007325333025720384
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,3072,0.005582222094138463
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,3072,0.007258666886223688
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,2560,0.005623999983072281
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,2560,0.006620444357395172
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,2048,0.005607111172543631
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,2048,0.006241777704821692
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,1536,0.004190222256713443
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,1536,0.006230222268237009
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,1024,0.003847111016511917
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,1024,0.005607999861240387
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,768,0.003370666669474708
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,768,0.005537777725193236
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,512,0.0031946665710873077
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,512,0.005210666606823604
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,64,0.0025964443468385274
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,256,3072,0.014133334159851074
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,256,0.002526222210791376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,256,0.004932444542646408
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,128,0.0028764444092909494
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,128,0.00489155575633049
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,64,0.004930666751331753
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,float16,1,32,32,0.002526222210791376
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8,1,32,32,0.00527555536892679
SGLang,0.5.6.post2,NVIDIA B200,gemm,sglang,fp8_block,1,128,128,0.01000355515215132
