framework,version,device,op_name,kernel_source,gemm_dtype,m,n,k,latency
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,10240,30.880836995442706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,12288,39.05600992838542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,12288,45.172428385416666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,16384,50.66553955078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,16384,58.99861246744791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,10240,56.05509033203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,10240,28.813210042317706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,12288,67.9503173828125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,8192,23.47171834309896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,8192,24.124859619140626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,7168,20.354695638020836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,16384,94.3943359375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,7168,18.98079630533854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,6144,17.82005818684896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,8192,45.63534342447917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,6144,18.612530517578126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,7168,39.59780680338542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,5120,14.653166707356771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,4096,11.878741455078124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,6144,33.23617350260417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,5120,13.932612101236979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,4096,11.50542500813802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,3584,10.512042236328124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,5120,27.70954182942708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,4096,22.044364420572915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,3072,8.900777180989582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,3584,10.227848307291667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,3072,9.497698974609374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,3584,19.3727783203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,2560,7.607159423828125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,3072,16.42427775065104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,2048,5.9748692830403645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,2560,7.499093119303386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,2048,6.121947733561198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,2560,13.67364298502604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,1536,4.604961140950521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,2048,10.66455078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,1536,7.983103942871094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,1024,3.1606783548990887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,1536,4.7140864054361975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,1024,3.3129814147949217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,768,2.490026601155599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,1024,6.796629333496094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,768,2.9100372314453127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,768,5.036100260416666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,512,1.8725866953531902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,256,1.6680960337320965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,512,3.275195821126302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,512,2.1373611450195313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,256,2.228019205729167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,256,2.0649983723958334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,128,1.742847951253255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,128,1.9300010681152344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,128,2.286079915364583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,64,1.995844268798828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,64,2.1962069193522136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,64,2.133367411295573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,32,2.0069674173990886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,32,2.3957504272460937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,32,2.1691734313964846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,12288,14.100991821289062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,16384,15.407308959960938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,16384,18.57136433919271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,12288,16.457010904947914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,16384,22.225339762369792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,10240,6.736008707682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,12288,10.74718729654948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,10240,11.829794311523438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,8192,9.363319905598958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,10240,13.510041300455729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,8192,5.916365051269532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,7168,4.918681844075521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,7168,8.210670979817708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,8192,10.63581034342448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,7168,9.368883260091145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,65536,61.74627685546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,6144,7.008324178059896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,6144,4.210787200927735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,5120,5.757201131184896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,6144,7.9158274332682295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,5120,6.425702412923177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,5120,3.4514261881510415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,4096,4.626397705078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,4096,2.743671417236328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,4096,5.14522450764974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,65536,76.30155436197916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,3584,4.073472086588541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,3584,4.363332112630209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,3584,2.4858965555826824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,3072,3.4457600911458335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,3072,2.14640630086263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,3072,3.7939542134602866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,2560,1.765614954630534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,2560,3.0314122517903646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,2560,2.9747540791829428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,2048,2.027212778727214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,65536,89.94829915364583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,2048,1.460087458292643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,1536,1.2911605834960938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,2048,2.36591796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,1536,1.139029312133789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,1536,1.787938181559245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,1024,0.9224533081054688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,1024,1.3635242462158204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,1024,0.8630271911621094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,768,0.7186431884765625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,768,0.7307946523030598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,768,1.0778624216715493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,512,0.5431978861490886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,512,0.7490528106689454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,512,0.5217280069986979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,256,0.456221866607666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,256,0.506333859761556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,256,0.5046613375345866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,128,0.45127681096394856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,128,0.41936213175455733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,128,0.4658495903015137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,64,0.4817237218221028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,64,0.38655999501546223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,64,0.5208405176798503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,32,0.49486506779988604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,32,0.5320021311442058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,32,0.5424469629923503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,16384,11.625130208333333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,12288,12.052138264973959
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,16384,16.362495930989585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,12288,12.486007690429688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,16384,16.457898966471355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,12288,8.463326009114583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,10240,5.6360621134440105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,8192,8.293580627441406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,10240,10.30871073404948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,10240,10.077388509114583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,8192,4.173926289876302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,8192,7.991398620605469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,7168,3.6229120890299478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,7168,7.2005971272786455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,65536,48.261531575520834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,7168,6.952652994791667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,6144,6.133794148763021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,6144,3.269051615397135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,6144,6.032179260253907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,5120,2.7558570861816407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,5120,5.11658681233724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,5120,5.067195638020833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,4096,4.045823923746744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,65536,65.9068603515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,4096,3.5208875020345056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,4096,2.0602879842122395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,65536,66.30638834635417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,3584,3.5482966105143228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,3584,1.786572774251302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,3584,3.3886891682942712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,3072,1.6124586741129558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,3072,2.9916842142740885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,3072,2.7534334818522135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,2560,2.5876480102539063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,2560,1.2883968353271484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,2048,1.6827392578125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,2048,1.7618602752685546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,2560,2.3285077412923174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,2048,1.075780232747396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,1536,1.080285898844401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,1536,1.290990956624349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,1536,0.8825855890909831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,1024,0.7362890879313151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,1024,1.0502485275268554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,1024,0.6573738733927409
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,768,0.5898239771525066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,768,0.8145919799804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,512,0.43547306060791013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,768,0.5703285217285157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,512,0.5657941182454427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,512,0.4320597330729167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,256,0.38120107650756835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,256,0.35437227884928385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,256,0.3971754709879557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,128,0.3736234664916992
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,128,0.2909184137980143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,128,0.3468970616658529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,64,0.3606186548868815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,64,0.3992234547932943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,64,0.28504746754964194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,32,0.36792214711507165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,32,0.28651520411173503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,32,0.41175041198730467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,16384,9.908462524414062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,12288,11.63871561686198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,16384,14.137616984049478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,16384,15.356211344401041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,12288,10.185216267903645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,12288,6.846464029947917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,10240,4.81955820719401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,10240,8.410108947753907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,10240,9.67720947265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,8192,7.621392822265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,8192,3.595195770263672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,8192,6.761847432454426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,7168,3.061074066162109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,65536,40.52718912760417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,7168,5.747814432779948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,7168,6.7126953125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,6144,4.985480753580729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,6144,2.775347137451172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,6144,5.7268224080403645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,5120,2.168251800537109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,5120,4.949845377604166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,5120,4.192870330810547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,65536,55.644327799479164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,4096,1.7544874827067058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,4096,3.2375094095865884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,4096,3.7696853637695313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,3584,3.2689834594726563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,3584,1.5595860799153647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,65536,61.273189290364584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,3584,2.659839884440104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,3072,2.9440681457519533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,3072,1.2947797139485677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,3072,2.3333206176757812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,2560,2.256827799479167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,2560,1.07236696879069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,2560,1.8994815826416016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,2048,1.499135971069336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,2048,1.5974058787027994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,2048,0.9095850626627604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,1536,0.9642666498819986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,1536,1.1156138102213542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,1024,0.6456661224365234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,1536,0.7309312184651693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,1024,0.5464746475219726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,1024,0.8783530553181966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,768,0.5155839920043945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,768,0.6475093205769856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,768,0.44257278442382814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,512,0.37833385467529296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,512,0.4495360056559245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,512,0.3505152066548666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,256,0.3026602745056152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,256,0.26760533650716145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,256,0.3089066823323568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,128,0.28798185984293617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,128,0.23886507352193198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,128,0.28757333755493164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,64,0.2348031997680664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,64,0.30665388107299807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,64,0.32453972498575845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,32,0.3068586667378744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,32,0.23661227226257325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,32,0.34177707036336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,16384,8.163771565755209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,12288,8.14223378499349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,16384,11.096917724609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,12288,10.60802510579427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,16384,14.15243123372396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,12288,5.065489196777344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,10240,3.6401494344075522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,10240,6.516394551595051
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,10240,8.8560302734375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,8192,7.153834533691406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,65536,31.58234456380208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,8192,2.7480745951334637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,8192,5.188164265950521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,7168,2.390459696451823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,7168,4.577791849772135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,7168,6.138606770833333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,6144,2.1635414123535157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,6144,3.7314900716145836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,6144,5.485670471191407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,65536,43.85321044921875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,5120,1.775172297159831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,5120,3.1827967325846354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,5120,4.382276407877604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,4096,2.4400896708170574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,4096,3.4777086893717444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,4096,1.411447525024414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,3584,3.190032958984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,3584,2.029803721110026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,3584,1.216034189860026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,65536,56.79629720052084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,3072,2.576520538330078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,3072,1.0484383900960288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,3072,1.818623987833659
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,2560,2.158831024169922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,2560,0.8518314361572266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,2560,1.454523722330729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,2048,1.3687466939290365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,2048,0.7533909479777019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,2048,1.174630355834961
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,1536,0.8234325408935547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,1536,0.5887658437093098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,1536,0.9954645156860351
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,1024,0.6877525329589844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,1024,0.5573290506998698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,1024,0.41130666732788085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,768,0.43509759902954104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,768,0.524561055501302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,768,0.35184640884399415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,512,0.3219061215718587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,512,0.3609600067138672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,512,0.26388479868570963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,256,0.2556191921234131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,256,0.21142187118530273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,256,0.24715946515401205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,128,0.22562133471171059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,128,0.18923519452412924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,64,0.2299562613169352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,128,0.22272000312805176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,64,0.18595840136210123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,64,0.2401279926300049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,32,0.23196372985839844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,32,0.18633386294047039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,32,0.2507402737935384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,16384,7.1115773518880205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,12288,7.077341715494792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,16384,9.734041341145833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,12288,10.276351928710938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,16384,13.550967407226562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,12288,4.5697021484375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,10240,3.4858324686686197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,10240,5.866288248697916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,10240,8.585727945963542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,65536,26.475587972005208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,8192,6.831104024251301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,8192,4.593698120117187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,8192,2.6577578226725262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,7168,2.172143046061198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,7168,5.934796651204428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,7168,3.965064493815104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,65536,38.04436442057292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,6144,1.83722661336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,6144,5.126587931315104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,6144,3.359027099609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,5120,1.6026283264160157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,5120,2.5040214538574217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,5120,4.200379689534506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,4096,1.2175360361735026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,4096,2.1094058990478515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,4096,3.5286356608072915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,3584,2.902801005045573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,3584,1.0994346618652344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,3584,1.8111146291097004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,3072,2.446643320719401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,3072,0.8604330698649088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,65536,54.809122721354164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,3072,1.5014570871988933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,2560,1.2537515004475912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,2560,2.028578186035156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,2560,0.8011775970458984
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,2048,1.357141367594401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,2048,0.6175743738810222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,2048,0.9823914845784506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,1536,0.7390879948933919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,1536,0.8285525639851888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,1536,0.5157205263773601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,1024,0.5273599942525228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,1024,0.6171648025512695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,1024,0.3728384017944336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,768,0.4037290573120117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,768,0.45455360412597656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,768,0.31334400177001953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,512,0.3047082583109538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,512,0.32020371754964194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,512,0.23213440577189126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,256,0.23425706227620444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,256,0.18486612637837727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,256,0.22231040000915528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,128,0.2061994711558024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,128,0.17148586908976238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,128,0.19718186060587567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,64,0.20746240615844727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,64,0.16240533192952472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,64,0.2219007968902588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,32,0.210534397761027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,32,0.16448853810628256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,32,0.21975040435791016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,16384,6.099080403645834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,12288,5.703133646647135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,16384,8.057036844889323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,12288,9.823743693033855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,16384,13.147648111979166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,12288,3.9726420084635414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,10240,4.905710856119791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,10240,2.9682688395182293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,65536,23.27794392903646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,10240,8.17425079345703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,8192,6.631697082519532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,8192,3.6547584533691406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,8192,2.2351529439290365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,65536,32.34542439778646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,7168,2.023799514770508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,7168,3.300829823811849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,7168,5.706410725911459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,6144,1.6632831573486329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,6144,2.561570231119792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,6144,4.919330342610677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,5120,2.226210021972656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,5120,1.4603946685791016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,5120,4.244138590494791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,4096,1.0873172760009766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,4096,3.222118377685547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,4096,1.7017173767089844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,3584,2.782173919677734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,3584,0.9100287755330404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,3584,1.514257049560547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,3072,2.375372823079427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,3072,0.8190975824991862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,3072,1.29628054300944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,2560,1.8979840596516926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,65536,51.879390462239584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,2560,0.654643185933431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,2560,1.0694656372070312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,2048,1.2022443135579428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,2048,0.5761344273885091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,2048,0.8553130467732748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,1536,0.716970698038737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,1536,0.6364501317342122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,1024,0.4357461293538411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,1024,0.4970837275187175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,1536,0.44407040278116866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,1024,0.3240960121154785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,768,0.37000532150268556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,768,0.2773674647013346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,768,0.38963200251261393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,512,0.26978880564371743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,512,0.2751829465230306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,512,0.2061994711558024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,256,0.15172266960144043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,256,0.20718933741251627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,256,0.18445653915405275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,128,0.17561599413553874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,128,0.1404245376586914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,128,0.16732160250345868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,64,0.1390933354695638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,64,0.17783466974894208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,64,0.1779370625813802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,32,0.1400490601857503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,32,0.18732372919718426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,32,0.17902933756510417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,16384,5.361288452148438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,16384,6.7126953125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,12288,5.104230244954428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,12288,9.462954711914062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,16384,12.547618611653645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,12288,3.3883135477701822
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,65536,20.630936686197916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,10240,2.6487808227539062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,10240,4.159795125325521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,10240,7.842167663574219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,8192,6.336749776204427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,65536,27.265604654947918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,8192,3.243485768636068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,8192,2.0634923299153645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,7168,2.8317354838053386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,7168,1.6095914204915363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,7168,5.486318969726563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,6144,2.380492909749349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,6144,1.5419391632080077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,6144,4.7512873331705725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,5120,3.895807902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,5120,1.1699541727701823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,5120,1.947852834065755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,4096,0.9346719741821289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,4096,1.5450453440348306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,4096,3.0937769571940104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,3584,2.6688512166341143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,3584,0.7937365214029948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,3584,1.3496320088704428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,65536,50.18343912760417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,3072,0.6578858693440754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,3072,1.1544235229492188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,2560,1.7694709777832032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,3072,2.2041600545247393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,2560,0.5796522776285807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,2560,0.9530709584554037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,2048,1.1460885365804037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,2048,0.7689215977986653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,2048,0.4721664110819499
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,1536,0.6777173360188802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,1536,0.5849429448445638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,1536,0.378163210550944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,1024,0.43707733154296874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,1024,0.43793067932128904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,1024,0.2815317471822103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,768,0.3452245394388834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,768,0.22555306752522788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,768,0.3317760149637858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,512,0.2453162670135498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,512,0.22555306752522788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,256,0.18049707412719726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,256,0.12912639776865642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,512,0.1597098668416341
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,256,0.1392639954884847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,128,0.15209813117980958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,128,0.11765759785970051
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,128,0.12462080319722493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,64,0.14510080019632976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,64,0.11537066300710042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,64,0.14520319302876789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,32,0.1152341365814209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,32,0.13544106483459473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,32,0.1344853401184082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,16384,4.457983907063802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,16384,5.375794982910156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,12288,3.7236043294270837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,16384,11.91239725748698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,12288,2.9332822163899737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,12288,8.915489705403646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,65536,17.10694376627604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,65536,21.298722330729166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,10240,2.13387934366862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,10240,7.467826334635417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,10240,3.1333653767903646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,8192,1.634508768717448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,8192,2.5853953043619793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,8192,5.977699279785156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,7168,1.435818608601888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,7168,2.180810546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,7168,5.461026000976562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,6144,1.8581483205159504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,6144,1.2641279856363932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,6144,4.460371398925782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,5120,1.523473103841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,5120,0.9942015965779623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,5120,3.7414911905924475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,4096,0.7307605107625326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,4096,1.2090367635091146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,4096,2.925806935628255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,3584,0.6608896255493164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,3584,1.0590890884399413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,3584,2.5303723653157553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,3072,2.1525845845540363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,3072,0.5441535949707031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,3072,0.9090047836303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,2560,1.661474100748698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,65536,47.43345540364583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,2560,0.4875946680704753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,2560,0.7469397226969401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,2048,0.5958656311035156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,2048,0.39021228154500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,1536,0.6015967686971029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,2048,1.0309290568033853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,1536,0.4549973487854004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,1536,0.30839465459187826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,1024,0.3179861386617025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,1024,0.3997354825337728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,1024,0.2171541372934977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,768,0.30037333170572916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,768,0.16117760340372722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,768,0.2596522649129232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,512,0.21152426401774088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,512,0.12076373100280761
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,512,0.18565120697021484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,256,0.1528831958770752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,256,0.1043455998102824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,256,0.07693653106689453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,128,0.11407360235850017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,128,0.07000746726989746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,128,0.0589141329129537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,64,0.0741376002629598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,64,0.09007786909739177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,64,0.048537600040435794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,32,0.06881279945373535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,32,0.04092586835225423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,32,0.04986879825592041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,16384,4.684356180826823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,12288,3.4197845458984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,16384,3.994999440511068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,12288,8.759978230794271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,16384,11.607210286458333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,12288,2.68223139444987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,10240,2.841156260172526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,65536,15.220326741536459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,65536,18.588604736328126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,10240,1.942801030476888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,10240,7.288524881998699
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,8192,1.4072149912516276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,8192,2.14828364054362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,8192,5.876565551757812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,7168,1.285529581705729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,7168,5.302032979329427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,7168,1.9337557474772136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,6144,1.1114495595296225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,6144,4.41156260172526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,6144,1.6100351969401043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,5120,0.8712191899617514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,5120,1.3137237548828125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,5120,3.656294504801432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,4096,2.8460713704427083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,4096,0.6664533615112305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,4096,1.066325314839681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,3584,2.479684193929036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,3584,0.5755562464396159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,3584,0.910370127360026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,3072,0.7810389200846355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,3072,2.0869461059570313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,3072,0.4760575930277507
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,2560,1.660791524251302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,2560,0.6531413396199544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,2560,0.43055680592854817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,65536,46.21841227213542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,2048,0.9914709091186523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,2048,0.5103594779968261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,2048,0.3547135988871256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,1536,0.5782869338989258
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,1536,0.39512745539347327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,1536,0.24541865984598793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,1024,0.37823146184285483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,1024,0.2665130615234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,768,0.2820437431335449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,1024,0.1784490744272868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,768,0.23002452850341798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,768,0.14411093393961588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,512,0.1999498685201009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,512,0.16397652626037598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,512,0.10018133322397868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,256,0.13786452611287434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,256,0.0679253339767456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,128,0.08573653697967529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,128,0.051575465997060144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,256,0.08850560188293458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,128,0.052084267139434814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,64,0.06628586848576865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,64,0.040755200386047366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,64,0.0425983985265096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,32,0.06198613246281942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,32,0.029286400477091475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,32,0.042871467272440594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,16384,3.9682388305664062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,16384,3.558024597167969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,12288,2.8498944600423175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,12288,8.522069295247395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,16384,11.448865763346355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,12288,2.3322964986165364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,65536,13.476966349283853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,10240,2.2070271809895834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,65536,15.795745849609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,10240,7.096524556477864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,10240,1.6894634246826172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,8192,1.83722661336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,8192,1.3130069732666017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,7168,1.6299349466959636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,8192,5.72405751546224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,7168,1.19825070699056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,7168,4.974284871419271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,6144,0.9667925516764322
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,6144,1.3323263804117838
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,6144,4.2569727579752605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,5120,1.1270484924316406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,5120,3.5208875020345056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,5120,0.797320556640625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,4096,2.7874987284342447
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,4096,0.6098602930704753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,4096,0.8929621378580729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,3584,2.40906244913737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,3584,0.7667711893717448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,3584,0.5053781191507976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,3072,2.0195327758789063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,3072,0.4436309178670247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,3072,0.6511957168579101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,2560,1.6079530080159503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,65536,44.98951009114583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,2560,0.5374975840250651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,2560,0.3632810592651367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,2048,1.0199381510416665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,2048,0.4221952120463054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,2048,0.2811903953552246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,1536,0.5152085304260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,1536,0.32709973653157554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,1024,0.22790826161702477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,1024,0.3499690691630045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,1536,0.20493653615315757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,1024,0.14885546366373698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,768,0.26699094772338866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,768,0.19503787358601887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,768,0.12448000113169352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,512,0.18817706108093263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,512,0.13680639266967773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,512,0.08566933472951253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,256,0.07782399654388428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,256,0.060279464721679686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,256,0.11864746411641439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,128,0.07167680263519287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,128,0.04403200149536133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,128,0.04379200140635173
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,64,0.059562667210896814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,64,0.03502080043156942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,64,0.03689813216527303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,32,0.056081068515777585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,32,0.03775146802266439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,32,0.025941334168116253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,16384,3.18480962117513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,16384,3.348070526123047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,12288,2.348236846923828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,16384,11.09415283203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,65536,12.215296427408855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,12288,2.1435391743977865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,12288,8.31102294921875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,65536,16.060279337565103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,10240,6.96603291829427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,10240,1.9447466532389324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,10240,1.5413589477539062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,8192,1.5599274953206381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,8192,1.2466859181722005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,8192,5.75614267985026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,7168,1.2972373962402344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,7168,5.062792460123697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,7168,1.0093226750691733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,6144,1.108514149983724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,6144,4.1762133280436196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,6144,0.8846335728963217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,5120,0.9132373174031576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,5120,0.7710037231445312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,5120,3.5272705078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,4096,0.7235925038655598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,4096,0.5376000086466471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,4096,2.714794667561849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,3584,2.3491925557454425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,3584,0.4638378779093425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,3584,0.6188714981079102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,3072,1.9657044728597004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,3072,0.5325482686360676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,3072,0.3879253387451172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,2560,0.43223040898640946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,2560,1.568221918741862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,2560,0.3207167943318685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,65536,44.49897867838542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,2048,0.9534805297851563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,2048,0.3482965469360352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,2048,0.23476907412211098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,1536,0.4789226531982422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,1536,0.2668544133504232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,1024,0.3300693194071452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,1536,0.18920106887817384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,1024,0.12612266540527345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,1024,0.18599039713541668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,768,0.24562346140543617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,768,0.14332586924235027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,768,0.10427733262379964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,512,0.1721002737681071
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,512,0.12830719947814942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,256,0.09898666540781656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,256,0.07485439777374267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,512,0.0757045348485311
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,256,0.05369173288345337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,128,0.06515520016352336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,128,0.0370688001314799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,128,0.03942186832427978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,64,0.054306133588155114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,64,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,64,0.03321173389752706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,32,0.050653866926829015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,32,0.022323199113210044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,32,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,16384,2.537164815266927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,16384,2.688580322265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,12288,1.8964820861816407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,12288,8.121002705891927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,65536,10.449647013346354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,16384,10.875698852539063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,12288,1.8304682413736981
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,65536,12.820512898763022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,10240,1.5303989410400392
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,10240,6.768127950032552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,10240,1.425237274169922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,8192,1.1910826365152993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,8192,1.0722304026285807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,7168,1.0085365295410156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,8192,5.50106455485026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,7168,4.93851318359375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,7168,0.9207808176676432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,6144,0.8591701507568359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,6144,0.760149319966634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,6144,4.065757751464844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,5120,0.7187114715576172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,5120,3.3575594584147135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,5120,0.6552576065063477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,4096,0.5696512222290039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,4096,0.4694698651631673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,4096,2.6317822774251303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,3584,0.5084501266479492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,3584,2.2992212931315104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,3584,0.381166934967041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,3072,0.3294549306233724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,3072,0.409497610727946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,3072,1.9058347066243488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,65536,42.98468831380208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,2560,1.5274667104085287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,2560,0.3403434753417969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,2560,0.22941013971964516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,2048,0.9109845479329428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,2048,0.28463786443074546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,2048,0.1826133410135905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,1536,0.4815189361572266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,1536,0.1423701286315918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,1536,0.2141866683959961
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,1024,0.2699946721394857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,1024,0.1513802687327067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,1024,0.10219519933064777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,768,0.2317311922709147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,768,0.11878399848937989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,768,0.08471893469492595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,512,0.15503360430399576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,512,0.09499306678771972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,512,0.06253226598103842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,256,0.08069120248158773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,256,0.056695465246836343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,128,0.05836799939473471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,256,0.04389546712239583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,128,0.0318122665087382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,128,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,64,0.04829866488774617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,64,0.024576000372568765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,64,0.02723840077718099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,32,0.027948800722757978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,32,0.04440746704737346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,32,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,16384,1.896448008219401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,12288,1.3202772776285807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,16384,2.2993237813313803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,65536,8.97962646484375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,65536,9.264469401041667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,16384,10.476816813151043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,12288,7.869371541341145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,12288,1.5599274953206381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,10240,1.1248298645019532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,10240,1.172309366861979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,10240,6.568243408203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,8192,0.912110964457194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,8192,0.9197226842244467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,7168,0.7536256154378255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,8192,5.44549560546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,7168,4.612334696451823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,7168,0.778715705871582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,6144,0.6431061426798503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,6144,0.6626303990681965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,5120,0.5441194534301758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,6144,3.9889236450195313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,5120,3.2687103271484377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,5120,0.5516970952351887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,4096,0.4266954739888509
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,4096,0.3623594601949056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,4096,2.5556991577148436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,3584,0.36457811991373695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,3584,0.31747414271036783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,3584,2.209894307454427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,3072,1.860983403523763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,3072,0.2297856012980143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,3072,0.30579840342203773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,65536,41.670039876302084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,2560,1.4858922322591146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,2560,0.2596522649129232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,2560,0.1765002727508545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,2048,0.866644287109375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,2048,0.2061994711558024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,2048,0.14209705988566082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,1536,0.4709034601847331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,1536,0.16186025937398274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,1536,0.11042133172353108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,1024,0.11373226642608643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,1024,0.2798250516255697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,1024,0.07922346591949463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,768,0.1805994669596354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,768,0.10277333259582519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,768,0.06584320068359376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,512,0.1356800079345703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,512,0.07478613058725993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,512,0.04915200074513753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,256,0.07051946322123209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,256,0.03652266661326091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,256,0.03446933428446452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,128,0.050995198885599766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,128,0.02461013396581014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,128,0.025941334168116253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,64,0.04235946734746297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,64,0.019317332903544107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,64,0.02215253313382467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,32,0.03887786865234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,32,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,32,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,16384,1.1623082478841147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,16384,1.852962112426758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,12288,0.8886271794637045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,65536,5.670502217610677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,65536,7.296100362141926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,12288,1.278054428100586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,12288,7.650440470377605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,16384,10.298197428385418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,10240,0.7331839879353841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,10240,0.9900373458862305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,8192,0.5828266779581706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,10240,6.37119140625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,8192,0.759771728515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,7168,0.514355214436849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,8192,5.212398783365885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,7168,0.6217045466105143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,7168,4.555980936686198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,6144,0.43714561462402346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,6144,0.5255477269490559
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,5120,0.37133652369181314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,5120,0.4432895978291829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,6144,3.8275072733561197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,5120,3.1637504577636717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,4096,0.26009599367777503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,4096,2.5536458333333334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,4096,0.2734421412150065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,3584,0.1950719992319743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,3584,0.233949867884318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,3584,2.246826680501302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,3072,1.7983829498291015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,3072,0.2032298723856608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,3072,0.14626132647196452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,2560,1.4307669321695964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,65536,40.46329345703125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,2560,0.12107093334197998
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,2560,0.1690282662709554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,2048,0.8470154444376627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,2048,0.14025386174519855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,2048,0.10079039732615154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,1536,0.43854506810506183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,1536,0.10765653451283771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,1024,0.2654208024342855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,1536,0.07915519873301188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,1024,0.07526400089263915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,1024,0.055637331803639736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,768,0.17714986801147461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,768,0.06789120038350424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,768,0.04706986745198567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,512,0.09400320053100586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,512,0.050619733333587644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,512,0.03375786542892456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,256,0.06048426628112793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,256,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,256,0.02461013396581014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,128,0.04369066556294759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,128,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,128,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,64,0.03594239950180054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,64,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,64,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,32,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,32,0.03204693396886189
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,32,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,16384,0.9234090805053711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,12288,0.6912000020345052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,16384,1.6168960571289062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,65536,4.389683024088542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,65536,6.425600179036458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,12288,1.1182421366373698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,12288,7.530768839518229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,10240,0.594538688659668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,16384,10.022195434570312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,10240,0.8838144302368164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,8192,0.524458662668864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,10240,6.26705067952474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,8192,0.682700792948405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,7168,0.46523733139038087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,8192,5.261345926920573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,7168,0.5799936294555664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,7168,4.395144653320313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,6144,0.3986090660095215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,6144,0.4635306676228841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,5120,0.3287381490071615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,6144,3.760162099202474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,5120,3.11364262898763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,5120,0.3718485196431478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,4096,0.2463007926940918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,4096,0.22224213282267252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,4096,2.4703999837239583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,3584,0.22094507217407228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,3584,2.1077674865722655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,3584,0.13977600733439127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,3072,0.18838186264038087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,3072,0.11301546891530354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,3072,1.7594027201334637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,2560,1.3946868896484375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,65536,39.8881103515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,2560,0.16151893933614095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,2560,0.09830400149027506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,2048,0.8179712295532227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,2048,0.1313109318415324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,2048,0.07997439702351888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,1536,0.06345386505126953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,1536,0.4145834604899089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,1024,0.2401269276936849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,1536,0.09888426462809244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,1024,0.06976853211720785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,1024,0.04464639822642009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,768,0.17595733006795247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,768,0.048674134413401286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,768,0.03768320083618164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,512,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,512,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,256,0.05550080140431722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,512,0.08666453361511231
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,256,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,256,0.01966080069541931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,128,0.03993599812189738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,128,0.01437013347943624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,128,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,64,0.032529066006342575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,64,0.011332266529401143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,32,0.02935466567675273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,64,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,32,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,32,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,16384,0.6365866978963216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,65536,2.7856213887532553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,12288,0.4805973370869954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,16384,1.3810005187988281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,65536,5.535505167643229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,12288,0.9784320195515951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,12288,7.397270202636719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,10240,0.3992234547932943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,16384,9.901055908203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,10240,0.7454037348429362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,8192,0.31634772618611656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,10240,6.264934285481771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,8192,0.5963093439737956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,7168,0.27880105972290037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,8192,5.072384134928385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,7168,0.4814165433247884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,7168,4.344081115722656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,6144,0.24238079388936362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,6144,0.4068352063496907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,5120,0.17715199788411456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,5120,0.31358292897542317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,6144,3.715072123209635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,5120,3.0676310221354166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,4096,0.16909653345743816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,4096,0.13482133547465008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,4096,2.4297813415527343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,3584,0.11581439971923828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,3584,2.0754709879557294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,3584,0.10100053151448567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,3072,1.7404586791992187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,3072,0.08516266345977783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,3072,0.10031572977701823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,2560,1.3727403004964194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,2560,0.08434346516927084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,65536,39.29961344401042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,2560,0.07406933307647705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,2048,0.78919677734375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,2048,0.07058773040771485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,2048,0.060482132434844973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,1536,0.4133887926737468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,1536,0.05485226710637411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,1536,0.046830932299296066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,1024,0.04464639822642009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,1024,0.032767999172210696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,1024,0.2458965301513672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,768,0.18216959635416669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,768,0.03444053332010905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,768,0.028398933013280232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,512,0.07929173310597738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,512,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,512,0.0201365331808726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,256,0.0504149317741394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,256,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,256,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,128,0.03618133465449015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,128,0.010989866654078166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,128,0.012351999680201214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,64,0.029252266883850096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,64,0.007913599908351897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,64,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,32,0.025941334168116253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,32,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,32,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,16384,0.5000874519348144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,65536,1.809100850423177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,12288,0.3709952036539713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,16384,1.137629826863607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,65536,4.660497029622396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,12288,0.8243541081746419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,12288,7.359283447265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,16384,9.806916300455729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,10240,0.3082570711771647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,10240,0.6633130391438802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,10240,6.199466451009115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,8192,0.25245013236999514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,7168,0.1989290714263916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,8192,5.012781778971354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,8192,0.5225472132364909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,7168,0.4140032132466634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,7168,4.379374694824219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,6144,0.14100054105122883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,6144,0.3380565325419108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,5120,0.08523093064626058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,5120,0.2432000001271566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,6144,3.698892720540365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,5120,3.0434644063313803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,4096,0.0723626693089803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,4096,0.10209279855092365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,4096,2.397934977213542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,3584,0.06327999830245971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,3584,2.0334932963053385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,3584,0.067413330078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,3072,0.05403306484222412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,3072,1.6917504628499347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,3072,0.05665599902470907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,2560,0.0444757342338562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,65536,38.9545654296875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,2560,1.335977554321289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,2560,0.04894719918568929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,2048,0.7664981206258138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,2048,0.036380799611409505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,2048,0.03979946772257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,1536,0.3704832077026367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,1536,0.0289792001247406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,1536,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,1024,0.23603199323018392
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,1024,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,1024,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,768,0.16772905985514325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,768,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,768,0.019108267625172932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,512,0.07140693664550782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,512,0.014643200238545737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,512,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,256,0.04495360056559245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,256,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,256,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,128,0.032561065753300984
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,128,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,64,0.02645333409309387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,64,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,128,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,64,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,32,0.022801067431767783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,32,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,32,0.008566400408744812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,16384,0.4868778546651204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,65536,1.7785172780354817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,16384,1.1224746704101562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,12288,0.347545591990153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,65536,4.643766276041666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,12288,0.8126805623372396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,10240,0.26518185933430993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,16384,9.80811055501302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,12288,7.351398213704426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,10240,0.6572714487711588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,10240,6.193595886230469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,8192,0.21862400372823082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,8192,0.48288427988688154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,8192,5.005789693196615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,7168,0.2113493283589681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,7168,0.39335254033406575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,7168,4.373299153645833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,6144,0.08297812938690186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,6144,0.3162453333536784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,5120,0.0506880005200704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,5120,0.22173013687133789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,6144,3.693329111735026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,5120,3.1553504943847654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,4096,0.045499734083811444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,4096,0.08297812938690186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,4096,2.3884117126464846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,3584,0.04242773453394572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,3584,0.05659306844075521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,3584,2.022980244954427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,3072,0.037853864828745525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,3072,0.048093867301940915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,3072,1.6818859100341796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,2560,0.03136426607767741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,2560,0.04133546749750773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,65536,38.95408121744792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,2560,1.3259712219238282
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,2048,0.7604223887125652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,2048,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,2048,0.03389439980189006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,1536,0.3780949274698893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,1536,0.0264138658841451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,1536,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,1024,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,1024,0.22882986068725586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,1024,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,768,0.15960747400919598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,768,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,768,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,512,0.07120213508605958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,512,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,512,0.011332266529401143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,256,0.04423679908116658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,256,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,256,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,128,0.031948800881703696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,64,0.025600000222524004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,128,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,128,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,64,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,64,0.007743999858697255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,32,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,32,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,32,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,16384,0.4453333218892415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,65536,1.7508351643880207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,16384,1.117457071940104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,12288,0.3421866734822591
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,65536,4.630289204915365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,12288,0.8227797190348307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,10240,0.2581461270650228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,12288,7.3461761474609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,10240,0.667682139078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,16384,9.788177490234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,10240,6.194721984863281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,8192,0.21739519437154137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,8192,0.47916374206542967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,7168,0.182476806640625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,7168,0.41755307515462237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,8192,4.996983337402344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,6144,0.07321600119272867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,7168,4.3430231730143225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,6144,0.32570025126139324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,5120,0.050619733333587644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,6144,3.6859563191731772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,5120,3.0022656758626303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,5120,0.22821547190348307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,4096,0.038809601465861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,4096,0.06775466601053873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,4096,2.347997792561849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,3584,0.03607893387476603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,3584,0.055910400549570714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,3584,2.0140703837076823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,3072,0.0310261329015096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,3072,0.0478549321492513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,3072,1.682193120320638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,2560,0.02635093331336975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,2560,1.3345109303792317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,2560,0.04113066593805949
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,65536,38.911930338541666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,2048,0.754688008626302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,2048,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,2048,0.033617067337036136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,1536,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,1536,0.025941334168116253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,1536,0.3876864115397135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,1024,0.21237759590148925
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,1024,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,1024,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,768,0.16868693033854168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,768,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,768,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,512,0.07092906634012858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,512,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,512,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,256,0.044202665487925216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,256,0.010545066992441813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,256,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,128,0.031778132915496825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,128,0.004402133325735727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,128,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,64,0.025497599442799883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,64,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,32,0.022562134265899658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,32,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,32,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,64,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,16384,0.4154026667277019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,65536,1.7471146901448567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,16384,1.121621322631836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,12288,0.34887679417928064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,65536,4.628104654947917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,12288,0.8194389343261719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,12288,7.320132446289063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,10240,0.25815040270487466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,16384,9.787493896484374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,10240,0.6522538503011067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,8192,0.21087573369344076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,10240,6.174481201171875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,8192,0.5105663935343425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,8192,4.995379130045572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,7168,0.1859242598215739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,7168,0.3889493306477865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,7168,4.266256968180338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,6144,0.0559445341428121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,6144,0.3101696014404297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,5120,0.03638613224029541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,6144,3.642333984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,5120,0.21452800432840982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,5120,3.1456255594889324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,4096,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,4096,0.067413330078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,4096,2.383257548014323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,3584,0.02645333409309387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,3584,2.0235925038655598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,3584,0.05577386617660522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,3072,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,3072,1.6792213439941406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,3072,0.04751360019048055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,2560,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,2560,1.3217450459798177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,2560,0.04116479953130086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,2048,0.014916266997655234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,65536,39.0789794921875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,2048,0.0335530678431193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,2048,0.7583744049072265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,1536,0.3684010823567708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,1536,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,1536,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,1024,0.22111573219299316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,1024,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,1024,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,768,0.1606613318125407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,768,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,512,0.07147520383199056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,768,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,512,0.0056981335083643595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,512,0.011260799566904704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,256,0.04392960071563721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,256,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,256,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,128,0.03181013266245524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,128,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,64,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,64,0.025497599442799883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,64,0.007605333129564922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,32,0.022425599892934165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,32,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,10240,14.825881958007812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,12288,18.428314208984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,12288,17.695369466145834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,16384,23.264017740885414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,16384,24.66153767903646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,10240,27.883998616536456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,12288,34.31106363932292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,10240,14.062762451171874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,8192,11.797470092773438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,7168,10.17204081217448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,8192,12.318924967447916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,16384,46.20045572916667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,8192,22.001664225260416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,7168,10.450841267903645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,6144,8.772402954101562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,7168,18.992913818359376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,6144,7.957913716634114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,5120,7.299515787760417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,6144,16.22469991048177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,5120,7.176977030436198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,5120,13.303910319010416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,4096,5.76013437906901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,3584,4.986606852213542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,4096,6.0834803263346355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,4096,10.368546549479166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,3584,5.17935791015625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,3072,4.1834154764811196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,3584,9.107968139648438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,3072,4.360191853841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,3072,7.637879435221355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,2560,3.5028650919596354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,2560,6.539297993977864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,2560,3.7877759297688804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,2048,2.8552182515462237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,1536,2.183063507080078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,2048,3.1355222066243487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,2048,5.200379943847656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,1536,2.2822911580403646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,1536,3.929770660400391
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,1024,1.458005396525065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,1024,1.6680960337320965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,768,1.1985919952392579
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,1024,2.933350372314453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,768,1.4449663798014323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,512,0.9351168314615885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,768,2.379332224527995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,512,1.5673333485921224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,512,1.1244202931722005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,256,0.9058303833007812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,256,0.9778517405192056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,256,1.1198463439941406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,128,0.8699221293131509
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,128,1.220130157470703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,128,0.9525930404663085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,64,1.0222250620524087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,64,1.0419530868530273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,64,1.0269013086954752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,32,1.0080938975016276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,32,1.2142932891845704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,32,1.0803541819254556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,16384,5.79590810139974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,12288,6.989277648925781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,16384,9.545693969726562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,12288,8.251318359375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,16384,11.280077107747395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,12288,4.177680969238281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,10240,5.79065195719401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,10240,3.67820790608724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,8192,4.604484049479167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,10240,6.825267028808594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,65536,28.034625244140624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,8192,5.110508728027344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,8192,2.557201131184896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,7168,2.437120056152344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,7168,3.944652811686198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,7168,4.610150146484375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,6144,3.5283968607584635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,65536,38.5800537109375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,6144,2.0770132700602213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,6144,3.7687637329101564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,65536,43.881709798177084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,5120,2.745953114827474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,5120,1.777015431722005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,5120,3.0473899841308594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,4096,1.880951436360677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,4096,1.4212437947591146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,3584,1.6002389272054036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,4096,2.3429120381673174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,3584,2.0729855855305988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,3584,1.2197205861409506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,3072,1.2533077239990233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,3072,0.9747104008992513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,2560,1.0457087834676106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,3072,1.7332906087239581
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,2560,0.8725845336914062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,2560,1.4273194630940755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,2048,0.8419669469197592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,2048,0.7522986729939778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,2048,1.1518922170003256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,1536,0.6415360132853191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,1024,0.4373504002888997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,1536,0.8847359975179037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,1536,0.5888682683308919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,1024,0.699185053507487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,1024,0.43724800745646164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,768,0.3569664001464844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,768,0.5443925221761068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,768,0.36433919270833337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,512,0.2765141487121582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,512,0.3807231903076172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,256,0.22913707097371422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,512,0.2846720059712728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,256,0.2527232011159261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,256,0.2260650634765625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,128,0.21862400372823082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,128,0.22036479314168295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,64,0.2401279926300049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,128,0.22695253690083822
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,64,0.1859242598215739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,64,0.2458965301513672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,32,0.24866132736206054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,32,0.18660693168640136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,32,0.2579114596048991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,16384,4.481365458170573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,12288,6.028151448567709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,16384,8.378368123372395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,12288,6.115806070963542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,16384,8.265318298339844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,12288,3.040563201904297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,10240,2.8086273193359377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,65536,19.973940022786458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,10240,4.946261088053385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,10240,5.106995137532552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,8192,3.9952725728352867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,8192,3.6095659891764322
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,8192,2.0135594685872396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,7168,3.6211369832356772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,7168,1.8524842580159504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,7168,3.323050689697266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,65536,32.54825032552083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,6144,2.948949432373047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,65536,33.19534912109375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,6144,2.5040214538574217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,6144,1.5793429056803385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,5120,1.3345792134602865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,5120,2.3843839009602865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,5120,2.2079488118489583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,4096,1.7145514170328777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,4096,1.0761898040771485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,4096,1.706598409016927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,3584,1.3302101135253905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,3584,1.4538752237955728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,3584,0.959112548828125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,3072,1.0409247716267904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,3072,0.823466682434082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,3072,1.2587008158365884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,2560,0.8295082728068033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,2560,1.0370047887166343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,2560,0.6649514516194661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,2048,0.7013376235961915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,2048,0.8545621236165365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,2048,0.5813589096069336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,1536,0.5297151883443196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,1536,0.660036277770996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,1536,0.4478293418884277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,1024,0.32812372843424475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,1024,0.44124161402384443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,1024,0.35140266418457033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,768,0.4031488100687663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,768,0.2863445281982422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,768,0.28651520411173503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,512,0.224563201268514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,512,0.2797226587931315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,256,0.18329599698384602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,512,0.21804374059041343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,256,0.1528490702311198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,256,0.18537813822428387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,128,0.16448853810628256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,128,0.17817600568135578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,128,0.1410048007965088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,64,0.1719637393951416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,64,0.1390933354695638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,64,0.1860607941945394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,32,0.14015146891276042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,32,0.17329492568969726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,32,0.1949013392130534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,16384,3.461563618977865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,16384,6.950430806477864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,16384,7.713007100423177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,12288,5.195537312825521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,12288,5.67708994547526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,12288,2.7831296284993488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,10240,4.066098022460937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,10240,4.757503763834635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,65536,18.352229817708334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,10240,2.3027712504069013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,8192,3.7554517110188805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,8192,3.2040618896484374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,65536,26.758860270182293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,8192,1.6861867268880206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,7168,1.405508295694987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,7168,3.2728747049967444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,7168,2.8843348185221354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,6144,2.760499318440755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,6144,2.255701446533203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,65536,31.208209228515624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,6144,1.2811946868896484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,5120,2.1866495768229166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,5120,1.9200682322184246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,5120,1.0642431894938151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,4096,1.4796459197998046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,4096,0.8956928253173828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,4096,1.4826496124267579
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,3584,1.2631722768147786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,3584,1.2933781941731772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,3584,0.7533568064371745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,3072,0.9285962422688803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,3072,1.113326899210612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,3072,0.6631765365600586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,2560,0.756326421101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,2560,0.5786965052286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,2560,0.9360725402832031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,2048,0.7539370854695637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,2048,0.6196565628051758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,2048,0.48766291936238604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,1536,0.4685824076334636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,1536,0.5911893208821615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,1536,0.3794602711995443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,1024,0.3082922617594401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,1024,0.3967658678690592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,768,0.24968533515930175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,768,0.2063360055287679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,768,0.3340970675150553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,1024,0.266376527150472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,512,0.19380906422932942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,512,0.23330133756001792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,256,0.15028907457987467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,512,0.14438400268554688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,256,0.12936533292134603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,128,0.13274453481038412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,256,0.13800106048583985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,128,0.12096853256225586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,128,0.11745279630025227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,64,0.1393664042154948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,64,0.11393706798553467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,64,0.13175466855367024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,32,0.11468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,32,0.1357141335805257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,32,0.13591893513997394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,16384,3.163374837239583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,12288,4.0815274556477865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,16384,5.4863530476888025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,12288,5.274555969238281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,16384,7.073587036132812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,12288,2.2960469563802084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,65536,14.94883015950521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,10240,1.7965396881103515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,10240,3.1507797241210938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,10240,4.386065165201822
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,8192,3.4995531717936195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,65536,21.582574462890626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,8192,1.4353023529052735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,8192,2.6134869893391928
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,7168,2.145041147867839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,7168,1.200332768758138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,7168,2.996326446533203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,6144,2.530713653564453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,65536,28.63329264322917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,6144,1.054037348429362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,6144,1.871121088663737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,5120,2.016426722208659
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,5120,0.9077760060628256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,5120,1.5067476908365884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,4096,1.3053610483805338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,4096,0.6978901545206706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,4096,1.2067157745361328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,3584,1.1050272623697917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,3584,1.015671475728353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,3584,0.6481237411499023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,3072,0.8216917037963867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,3072,0.5335381189982097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,3072,0.8635050455729166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,2560,0.6456661224365234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,2560,0.7367680231730144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,2560,0.4712064107259114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,2048,0.5927253087361654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,2048,0.5415594736735027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,2048,0.3794944127400716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,1536,0.40383148193359375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,1536,0.4514133453369141
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,1536,0.24907093048095702
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,1024,0.2657279968261719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,1024,0.3150848070780436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,1024,0.19288746515909833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,768,0.21278719902038573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,768,0.15636480649312337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,768,0.26579519907633464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,512,0.16554667154947916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,512,0.1842858632405599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,512,0.11919039885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,256,0.1148565371831258
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,256,0.10349226792653402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,128,0.06662613153457642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,128,0.08164693514506022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,128,0.05666133165359497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,256,0.07512746651967367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,64,0.06799360116322836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,64,0.046727466583251956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,64,0.04751360019048055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,32,0.06488746802012126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,32,0.04795733292897542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,32,0.04130133390426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,16384,2.6625706990559896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,16384,4.824644470214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,12288,3.550310516357422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,16384,6.801988220214843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,12288,5.041595967610677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,65536,11.934617106119791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,12288,1.9151531219482423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,10240,1.590886433919271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,10240,2.9124608357747395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,10240,4.194163258870443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,65536,18.924237060546876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,8192,3.3266006469726563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,8192,1.2024490356445312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,8192,2.255121103922526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,7168,1.0720255533854166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,7168,2.886587778727214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,7168,1.9039231618245442
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,6144,1.565932846069336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,65536,27.261167399088542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,6144,0.9210880279541016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,6144,2.5547414143880207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,5120,1.9496618906656902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,5120,0.7825066884358723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,5120,1.3042004903157554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,4096,1.2351487477620444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,4096,1.0066261291503906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,3584,1.0139989217122396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,4096,0.6267904281616211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,3584,0.8727893193562826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,3584,0.5687295913696289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,3072,0.7523328145345052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,3072,0.4864000002543132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,3072,0.7634944279988607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,2560,0.608904520670573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,2048,0.4896426518758138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,2560,0.3741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,2560,0.6272319793701172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,2048,0.2994858741760254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,2048,0.5041834513346355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,1536,0.37263358434041344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,1536,0.39516159693400066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,1024,0.2699946721394857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,1024,0.24712533950805665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,1536,0.2389333407084147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,1024,0.16332799593607586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,768,0.19927040735880536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,768,0.12741973400115966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,512,0.15035732587178546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,768,0.22654293378194174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,512,0.09929386774698892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,512,0.1591978708902995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,256,0.09680213133494059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,256,0.08861013253529867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,256,0.06683306694030762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,64,0.06208639939626058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,64,0.04123200178146362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,128,0.05038079818089804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,128,0.07276906967163085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,128,0.08034986654917399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,64,0.04130133390426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,32,0.05932373205820719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,32,0.0500053326288859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,32,0.042734932899475095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,16384,2.2856704711914064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,16384,4.059438832600912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,16384,6.499942525227864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,12288,2.9630465189615887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,12288,4.818500264485677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,65536,9.746227010091145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,12288,1.6571701049804688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,10240,4.0394752502441404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,10240,1.4080341339111329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,10240,2.276145172119141
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,8192,1.069329071044922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,8192,3.1775062561035154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,8192,1.898257064819336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,65536,19.973529052734374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,7168,0.9364479700724283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,7168,1.5815338134765624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,7168,2.761999003092448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,65536,26.01195475260417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,6144,2.2981630961100263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,6144,1.3415765126546224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,6144,0.7899477640787761
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,5120,1.9895295461018878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,5120,0.6399658838907878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,5120,1.0857471466064452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,4096,1.1902292887369792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,4096,0.514628251393636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,4096,0.9631743748982748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,3584,0.9405098597208659
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,3584,0.7513429641723632
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,3584,0.4645845413208008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,3072,0.654097048441569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,3072,0.6366549173990885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,3072,0.38942718505859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,2560,0.5506389617919922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,2560,0.2899968147277832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,2560,0.5287594795227051
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,2048,0.4232192039489746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,2048,0.4511743863423665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,2048,0.25466879208882653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,1536,0.34082132975260415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,1536,0.3356671969095866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,1024,0.22807893753051758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,1536,0.19172693888346354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,1024,0.2246250629425049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,1024,0.1396394729614258
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,768,0.19469653765360515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,768,0.1792682647705078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,768,0.11055786609649658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,512,0.13301760355631512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,512,0.1378986676534017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,512,0.08266560236612955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,256,0.0870741367340088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,256,0.05894826650619507
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,256,0.07782399654388428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,128,0.04457813501358032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,128,0.06430720090866089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,128,0.042632532119750974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,64,0.05546666781107584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,64,0.03549866676330567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,64,0.03570346832275391
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,32,0.05331626733144125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,32,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,32,0.0362496018409729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,16384,2.105446370442708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,16384,3.3361579895019533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,12288,2.392198435465495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,16384,6.269712829589844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,12288,4.690807596842448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,65536,9.647274780273438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,12288,1.569586181640625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,10240,1.274774424235026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,10240,1.956386184692383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,10240,3.873382314046224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,65536,16.608426920572917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,8192,3.0813524881998697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,8192,1.5071231842041015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,8192,0.9704106648763021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,7168,0.807253328959147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,7168,1.282867177327474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,7168,2.6366976420084636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,65536,25.21190388997396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,6144,0.7061845143636067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,6144,1.0822315216064453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,6144,2.3710378011067705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,5120,1.777288564046224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,5120,0.5576021194458007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,5120,0.8909088134765625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,4096,1.1196053822835288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,3584,0.8952149073282877
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,4096,0.48506879806518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,3584,0.6123519897460937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,4096,0.7122943878173829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,3584,0.3667626698811849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,3072,0.6220458984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,3072,0.5163626670837402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,3072,0.32713387807210287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,2560,0.4302165349324544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,2560,0.476364803314209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,2560,0.2812575976053874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,2048,0.3479893366495768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,2048,0.41622187296549484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,2048,0.2205344041188558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,1536,0.3137194633483887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,1536,0.2660693327585856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,1024,0.21234347025553385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,1536,0.17411413192749023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,1024,0.18315946261088054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,1024,0.12079786459604899
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,768,0.15564799308776855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,768,0.14619305928548176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,768,0.09963520367940268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,512,0.11496106783548991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,512,0.07461547056833903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,256,0.0754688024520874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,256,0.07700160344441732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,512,0.12960426807403563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,256,0.05249706506729126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,128,0.05748053391774496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,128,0.03696639935175578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,128,0.038126933574676516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,64,0.04997119903564453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,64,0.03580586512883504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,64,0.03191039959589641
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,32,0.04747946659723918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,32,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,32,0.032663466533025105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,16384,1.791453806559245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,16384,2.5143979390462237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,12288,1.8649770100911458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,16384,5.9635370890299475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,12288,4.4499969482421875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,65536,7.481310017903645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,12288,1.2799317677815756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,10240,1.5279103597005208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,65536,12.905131022135416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,10240,3.6904276529947913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,10240,1.0891605377197267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,8192,1.1794090270996094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,8192,0.7556437174479167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,8192,2.9191851298014324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,7168,0.682257080078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,7168,1.014749844868978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,7168,2.529143524169922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,6144,2.117358907063802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,65536,24.049186197916665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,6144,0.8472234725952148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,6144,0.5525781631469726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,5120,1.6950271606445313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,5120,0.6940330505371094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,5120,0.4925034523010254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,4096,0.5440511703491211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,4096,1.054310417175293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,4096,0.373417599995931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,3584,0.8435989379882812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,3584,0.4668757438659668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,3584,0.30197760264078777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,3072,0.5695829391479492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,3072,0.40417280197143557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,3072,0.2600618680318197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,2560,0.4264277458190918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,2560,0.339791997273763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,2560,0.21456212997436525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,2048,0.34024108250935875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,2048,0.17517226537068684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,2048,0.27494398752848304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,1536,0.28569599787394206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,1536,0.21650773684183755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,1536,0.13707946141560873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,1024,0.18035839398701986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,1024,0.15216426849365233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,1024,0.09768959681193033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,768,0.13069653511047363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,512,0.09837226867675782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,768,0.11796480019887287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,768,0.08202239672342936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,512,0.06058666706085205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,512,0.09413973490397134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,256,0.06645653247833253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,256,0.054169599215189615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,256,0.042905600865681966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,128,0.051575465997060144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,128,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,128,0.031573333342870075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,64,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,64,0.04386133352915446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,64,0.026589866479237872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,32,0.04184746742248535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,32,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,32,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,16384,1.4897492726643882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,16384,2.2449493408203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,12288,1.5756287892659506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,16384,5.817173258463542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,65536,6.926881917317708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,12288,4.3508051554361975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,12288,1.1866111755371094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,65536,10.614373779296875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,10240,1.2914688110351562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,10240,3.596356201171875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,10240,0.9206783930460611
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,8192,1.1340459187825522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,8192,2.9343401590983076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,8192,0.6729386647542317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,7168,2.556723276774089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,7168,0.5689343770345052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,7168,0.9938261032104492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,6144,2.0706986745198566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,6144,0.8408063888549805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,6144,0.517358938852946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,65536,23.43236287434896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,5120,1.6588459014892578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,5120,0.6984362920125325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,5120,0.42359466552734376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,4096,0.30197652180989587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,4096,0.5269845326741536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,4096,1.0414762496948242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,3584,0.821657625834147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,3584,0.483020814259847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,3584,0.261188268661499
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,3072,0.4068010648091634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,3072,0.5735424041748047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,3072,0.22138880093892416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,2560,0.4381653467814128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,2560,0.34348373413085936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,2560,0.18838186264038087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,2048,0.3181568145751953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,2048,0.28053760528564453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,2048,0.15694506963094074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,1536,0.2698560078938802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,1536,0.21858986218770346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,1536,0.11977386474609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,1024,0.15209813117980958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,1024,0.1731584072113037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,1024,0.08837119738260904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,768,0.12769280274709066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,768,0.1252010663350423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,768,0.07280639807383218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,512,0.09024853706359863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,512,0.08089599609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,512,0.05369173288345337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,256,0.06174720128377279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,256,0.04266666571299235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,256,0.03805866638819377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,128,0.048230401674906415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,128,0.02802346746126811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,128,0.028433066606521607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,64,0.04102826515833537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,64,0.022696532805760703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,64,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,32,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,32,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,32,0.03891199827194214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,16384,1.3665279388427733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,12288,1.307541275024414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,16384,1.8278400421142578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,65536,6.254523722330729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,12288,4.244002024332682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,16384,5.695008850097656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,65536,8.398267618815105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,10240,1.0821973164876302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,12288,1.0000725428263346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,10240,0.8765781402587891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,10240,3.514811706542969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,8192,0.6246058781941731
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,8192,0.8614570617675781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,8192,2.783300272623698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,7168,0.7342762629191081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,7168,2.3948959350585937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,7168,0.5057194709777832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,6144,0.6250495910644531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,6144,0.4568405469258626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,65536,22.814105224609374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,6144,2.028987757364909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,5120,1.5988394419352212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,5120,0.4966730753580729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,5120,0.312661329905192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,4096,0.39611733754475914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,4096,0.9699669520060221
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,4096,0.25726292928059896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,3584,0.8363349278767904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,3584,0.3470282554626465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,3584,0.2194431940714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,3072,0.5513557434082031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,3072,0.29982719421386717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,3072,0.19554773966471356
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,2560,0.4178911844889323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,2560,0.24999252955118814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,2560,0.16537599563598632
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,2048,0.3410602569580078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,2048,0.21271893183390297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,1536,0.21537599563598633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,1536,0.16319146156311035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,2048,0.13206186294555664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,1536,0.10366293589274089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,1024,0.1626794656117757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,1024,0.11386880079905193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,1024,0.07631573677062989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,768,0.11195733547210693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,768,0.08980480035146078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,768,0.06277120113372803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,512,0.046728531519571945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,512,0.07461547056833903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,512,0.08331946531931558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,256,0.056866133213043214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,256,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,256,0.03816106716791789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,128,0.04427093267440796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,128,0.025361067056655882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,128,0.024780799945195518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,64,0.037649067242940266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,64,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,64,0.02071466644605001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,32,0.03614720106124878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,32,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,32,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,16384,1.2188672383626302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,16384,1.4587904612223306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,12288,1.0599082946777343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,65536,5.260390218098959
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,12288,4.138495890299479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,65536,6.965248107910156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,16384,5.527347310384115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,12288,0.9149440129597982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,10240,3.41387939453125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,10240,0.8805717468261719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,10240,0.7675221125284831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,8192,0.5321013450622558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,8192,0.7219541549682618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,8192,2.716910807291667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,7168,2.328473663330078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,7168,0.7203498840332031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,7168,0.449399471282959
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,6144,0.5928618748982747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,6144,0.37154134114583337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,6144,1.949013392130534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,65536,22.14795939127604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,5120,1.5581525166829429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,5120,0.4845909436543782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,5120,0.26353813807169596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,4096,0.9507829030354819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,4096,0.3877504030863444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,4096,0.2112511952718099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,3584,0.7434239705403646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,3584,0.3497642517089844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,3584,0.18647039731343587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,3072,0.49097385406494143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,3072,0.2943317413330078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,3072,0.16059734026590983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,2560,0.4009984016418457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,2560,0.13783040046691894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,2560,0.24886612892150878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,2048,0.3059370676676432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,2048,0.2096127986907959
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,2048,0.11380053361256917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,1536,0.20558506647745767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,1536,0.16233812967936198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,1536,0.08830293019612631
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,1024,0.14496426582336425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,768,0.10482347011566162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,1024,0.06444373528162638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,1024,0.11284480094909669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,768,0.08325119813283285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,768,0.053282133738199865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,512,0.07325013478597006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,512,0.06198613246281942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,512,0.039287467797597245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,256,0.05304319858551025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,256,0.03290026585261027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,256,0.028262400627136232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,128,0.04085760116577149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,128,0.021572266022364298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,128,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,64,0.0347818652788798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,64,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,64,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,32,0.03283626635869344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,32,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,32,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,16384,1.1710464477539062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,16384,1.063321622212728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,65536,5.526561991373698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,12288,0.8933034896850586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,65536,4.666845703125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,16384,5.3685246785481775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,12288,4.028825632731119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,12288,0.8010751724243164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,10240,0.7384714762369792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,10240,0.6811989466349284
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,10240,3.3231531778971353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,8192,2.717081705729167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,8192,0.5646336237589519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,8192,0.46533972422281905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,7168,2.2544041951497396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,7168,0.4830890655517578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,7168,0.3623594601949056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,65536,21.4656005859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,6144,0.27019945780436194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,6144,0.39191360473632814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,6144,1.9226282755533854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,5120,1.5081802368164063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,5120,0.32085332870483396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,5120,0.21073919932047525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,4096,0.9026559829711914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,4096,0.17237332661946614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,4096,0.2642944018046061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,3584,0.7187456130981446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,3584,0.2306389331817627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,3584,0.15162026087443034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,3072,0.4918954531351726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,3072,0.19831466674804688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,3072,0.13076480229695636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,2560,0.36785494486490883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,2560,0.17015466690063477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,2560,0.11096746921539306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,2048,0.28067839940389
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,2048,0.1392298698425293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,2048,0.09226240317026774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,1536,0.2174293359120687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,1536,0.10973866780598958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,1536,0.0718506654103597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,1024,0.11468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,1024,0.0762880007425944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,1024,0.05218986670176188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,768,0.0713696002960205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,768,0.09297920068105062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,768,0.04379306634267171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,512,0.045909333229064944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,512,0.06669653256734212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,512,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,256,0.04771840174992879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,256,0.026997333765029906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,256,0.02334400018056234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,128,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,64,0.013753599921862283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,128,0.03682986497879028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,128,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,64,0.03126613299051921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,64,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2048,32,0.029417600234349566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2048,32,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2048,32,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,16384,0.9229930877685547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,16384,0.9373696009318033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,65536,3.7461334228515626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,65536,4.730333964029947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,12288,0.673792012532552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,16384,5.3537109375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,12288,0.6907562891642253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,12288,3.913109334309896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,10240,0.5797162373860677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,10240,3.2347819010416665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,10240,0.5925888061523438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,8192,0.5146623929341634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,8192,2.545493316650391
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,8192,0.3624277432759603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,7168,2.207436879475911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,7168,0.4296703974405925
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,7168,0.2783573468526205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,6144,1.8469547271728515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,6144,0.19630079269409179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,65536,20.91837361653646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,6144,0.3652949333190918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,5120,1.4652757008870443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,5120,0.1633610725402832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,5120,0.30866772333780923
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,4096,0.8658890406290689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,4096,0.1343488057454427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,4096,0.24832000732421874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,3584,0.2179072062174479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,3584,0.7311360041300456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,3584,0.12277759710947674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,3072,0.18899307250976563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,3072,0.4357461293538411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,3072,0.1023637294769287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,2560,0.3405824025472005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,2560,0.15336106618245443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,2048,0.2678773244222005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,2560,0.08663040002187093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,2048,0.1241429328918457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,2048,0.07273600101470948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,1536,0.2020352045694987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,1536,0.09741653601328532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,1536,0.057719465096791586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,1024,0.09932800134023032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,1024,0.0413696010907491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,1024,0.0710314671198527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,768,0.07823359966278076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,768,0.048913065592447916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,768,0.034645334879557295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,512,0.059869865576426186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,512,0.03577173153559367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,256,0.04307626485824585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,512,0.02576533357302348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,256,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,256,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,128,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,128,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,128,0.014642133315404256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,64,0.028467200199762982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,64,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,64,0.012525866429011026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1536,32,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1536,32,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1536,32,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,65536,2.64987309773763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,16384,0.593715222676595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,16384,0.7795029322306315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,12288,0.4355754534403483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,65536,3.3047892252604165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,16384,5.077845255533854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,12288,3.7914283752441404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,12288,0.5896533330281575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,10240,0.36505600611368816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,10240,0.4538709322611491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,10240,3.1460352579752606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,8192,0.26821972529093424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,8192,2.514227294921875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,8192,0.2957653363545736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,7168,0.22213973999023437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,7168,0.17189547220865886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,7168,2.1200213114420574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,65536,20.22437540690104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,6144,0.18964479764302572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,6144,0.14834346771240234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,6144,1.7801898956298827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,5120,1.407690684000651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,5120,0.16018773714701334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,5120,0.11980799833933513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,4096,0.830293337504069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,4096,0.09724586804707845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,4096,0.13267626762390136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,3584,0.6535167694091797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,3584,0.11755519707997639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,3584,0.08366080125172934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,3072,0.09881599744160971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,3072,0.39949652353922527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,3072,0.07485439777374267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,2560,0.3182250658671061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,2560,0.08427519798278808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,2560,0.06396586497624715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,2048,0.2605397383371989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,2048,0.07010986804962158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,1536,0.17339305877685546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,2048,0.05256533225377401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,1536,0.05512533187866211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,1536,0.04078933397928874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,1024,0.0864565372467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,1024,0.04495360056559245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,1024,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,768,0.06867626508076986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,768,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,768,0.034577067693074545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,512,0.05300906499226889
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,512,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,512,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,256,0.03775146802266439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,256,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,256,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,128,0.029422932863235475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,128,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,128,0.010342400272687275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,64,0.02501973311106364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,64,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,64,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,1024,32,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,1024,32,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,1024,32,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,65536,1.8989738464355468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,16384,0.47414614359537766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,65536,3.1576746622721354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,16384,0.7391189575195313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,12288,0.3492181460062663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,16384,5.064567565917969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,12288,3.7878443400065103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,12288,0.5672277450561524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,10240,0.4377600034077962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,10240,0.29146451950073243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,10240,3.1409835815429688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,8192,0.2171562671661377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,8192,2.460296630859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,8192,0.26077653566996256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,7168,2.1133312225341796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,7168,0.22339733441670737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,7168,0.1534293333689372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,6144,0.1927509307861328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,65536,20.16542765299479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,6144,0.13069653511047363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,6144,1.780736033121745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,5120,1.4057791392008463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,5120,0.1654784043629964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,5120,0.1129472017288208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,4096,0.09547093709309896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,4096,0.8226133346557617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,4096,0.13455360730489094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,3584,0.6461781183878581
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,3584,0.11963733037312825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,3584,0.08738133112589518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,3072,0.411135991414388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,3072,0.10656426747639973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,3072,0.07604906558990479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,2560,0.09506133397420248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,2560,0.33041067123413087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,2560,0.05918720165888468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,2048,0.24094506899515786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,2048,0.08068479696909586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,2048,0.048093867301940915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,1536,0.17015466690063477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,1536,0.06795946756998697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,1536,0.03741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,1024,0.08475306828816732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,1024,0.040277334054311116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,1024,0.027170133590698243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,768,0.06795946756998697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,768,0.03211946686108907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,768,0.023586134115854897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,512,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,512,0.05208746592203776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,256,0.03696639935175578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,512,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,256,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,256,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,128,0.0289792001247406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,128,0.010990933577219645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,128,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,64,0.02450773318608602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,64,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,768,32,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,64,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,768,32,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,768,32,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,16384,0.31621119181315105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,65536,1.3182975769042968
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,12288,0.23797760009765626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,16384,0.6267221450805665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,65536,2.569898732503255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,16384,4.943257649739584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,12288,3.699127451578776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,12288,0.4528810818990071
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,10240,0.1813162644704183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,10240,0.3551573435465495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,10240,3.11732915242513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,8192,0.1435647964477539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,8192,2.510199483235677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,8192,0.17889280319213868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,7168,0.1159168004989624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,7168,2.046361541748047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,7168,0.1002837340037028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,6144,0.0856725295384725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,65536,19.681791178385417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,6144,0.1047210693359375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,6144,1.7238357543945313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,5120,1.3579573313395181
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,5120,0.08717652956644693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,5120,0.07256746292114258
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,4096,0.05928959846496582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,4096,0.07280639807383218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,3584,0.6124160130818684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,3584,0.06471680005391439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,4096,0.7946240107218425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,3584,0.05208319822947184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,3072,0.3922602653503418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,2560,0.30498132705688474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,3072,0.04532906611760457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,3072,0.053521064917246494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,2560,0.045021867752075194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,2560,0.03843413194020589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,2048,0.041843199729919435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,2048,0.22312960624694825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,2048,0.03136853377024333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,1536,0.16315733591715495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,1536,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,1536,0.03334826628367106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,1024,0.07566933631896973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,1024,0.024234666426976522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,1024,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,768,0.019933867454528808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,768,0.059665067990620935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,768,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,512,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,512,0.04573866526285807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,512,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,256,0.032494932413101196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,256,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,256,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,128,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,128,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,128,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,64,0.021538132429122926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,64,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,64,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,512,32,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,512,32,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,512,32,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,16384,0.24917333920796714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,16384,0.5915989557902018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,65536,0.9209173202514649
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,65536,2.426197306315104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,12288,3.6587519327799476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,16384,4.895777893066406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,12288,0.19367252985636393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,12288,0.41581226984659836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,10240,0.10294186274210612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,10240,0.33535998662312827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,10240,3.0205930074055987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,8192,0.1290239969889323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,8192,0.07618559996287028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,8192,2.3851348876953127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,7168,0.06720853646596273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,7168,2.032093811035156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,7168,0.08188587029774984
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,65536,19.56809285481771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,6144,1.7077930450439454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,6144,0.05928959846496582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,6144,0.06959786415100097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,5120,0.050204801559448245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,5120,0.0589141329129537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,5120,1.343829345703125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,4096,0.7755776087443034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,4096,0.04130133390426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,4096,0.048019198576609294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,3584,0.6063103993733724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,3584,0.042456531524658205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,3072,0.36532907485961913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,3584,0.037546666463216145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,3072,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,3072,0.0373418649037679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,2560,0.29945173263549807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,2560,0.029798400402069092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,2560,0.031300266583760575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,2048,0.23364267349243165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,2048,0.02590720057487488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,1536,0.16274773279825847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,1536,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,2048,0.023995733261108397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,1536,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,1024,0.07335253556569418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,1024,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,1024,0.014233600099881491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,768,0.01181013286113739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,768,0.058641068140665685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,768,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,512,0.044811733563741046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,512,0.009179733196894328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,512,0.010102400183677673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,256,0.031846400101979574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,256,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,256,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,128,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,64,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,128,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,128,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,64,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,64,0.006757333377997081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,256,32,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,256,32,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,256,32,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,65536,0.896068255106608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,16384,0.23555413881937662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,65536,2.4199508666992187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,16384,0.5891754786173503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,12288,0.11826986471811932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,16384,4.898235575358073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,12288,3.6522326151529945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,12288,0.4336640040079753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,10240,0.04666026830673218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,10240,0.32225281397501626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,10240,3.021107228597005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,8192,2.375884755452474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,8192,0.03839999834696452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,8192,0.11502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,7168,0.03741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,7168,2.057420857747396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,65536,19.5637939453125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,7168,0.08195412953694661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,6144,0.031163734197616578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,6144,1.7090901692708331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,6144,0.06980266571044921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,5120,1.3349183400472007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,5120,0.029286400477091475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,5120,0.05850453376770019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,4096,0.04802560011545817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,4096,0.8573951721191406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,4096,0.027687466144561766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,3584,0.025565866629282636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,3584,0.5944661458333333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,3584,0.04212053219477336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,3072,0.35720532735188804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,3072,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,2560,0.2960042635599772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,3072,0.036962131659189865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,2048,0.23391572634379068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,2560,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,2560,0.0310261329015096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,2048,0.025361067056655882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,2048,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,1536,0.019933867454528808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,1536,0.16906240781148274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,1536,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,1024,0.07393279870351156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,1024,0.009723732868830364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,1024,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,768,0.05850453376770019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,768,0.008494933446248371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,768,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,512,0.006618666648864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,512,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,512,0.04450986782709758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,256,0.031573333342870075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,256,0.004911999901135763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,256,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,128,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,128,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,64,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,128,0.007098666826883952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,64,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,64,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,128,32,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,128,32,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,128,32,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,65536,0.8702293395996094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,16384,0.22155946095784507
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,65536,2.4101888020833333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,16384,0.5906773249308268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,12288,0.08608427047729492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,16384,4.886664326985677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,12288,0.43178666432698565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,10240,0.04710186719894409
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,12288,3.6509012858072913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,10240,0.3083584149678548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,10240,3.019775899251302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,8192,0.03857066631317139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,8192,2.3698089599609373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,8192,0.10181972980499268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,7168,0.08062293529510497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,7168,2.0573183695475263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,7168,0.04700160026550293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,65536,19.542289225260415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,6144,1.7041749318440753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,6144,0.040004265308380124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,6144,0.06908586819966635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,5120,1.3405867258707682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,5120,0.0340992013613383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,5120,0.058641068140665685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,4096,0.027101866404215497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,4096,0.7793664296468099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,4096,0.04771840174992879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,3584,0.025770666201909383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,3584,0.5921450932820638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,3584,0.04215466578801473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,3072,0.021640533208847047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,3072,0.3741696039835612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,3072,0.03682986497879028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,2560,0.018739199638366698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,2560,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,2048,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,2560,0.27845865885416665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,2048,0.21381120681762694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,2048,0.025388799111048382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,1536,0.15575040181477864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,1536,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,1024,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,1536,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,1024,0.0729087988535563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,1024,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,768,0.058641068140665685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,768,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,512,0.04468053181966146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,768,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,512,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,512,0.009796266754468281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,256,0.0315391997496287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,256,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,256,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,128,0.024745599428812663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,128,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,64,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,64,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,64,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,64,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,64,32,0.019352533419926963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,64,32,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,65536,0.870024553934733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,16384,0.21661012967427573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,65536,2.407492319742839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,12288,0.0659114678700765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,16384,0.5857280095418294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,12288,0.4309674580891927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,16384,4.8974146525065105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,12288,3.6487167358398436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,10240,0.04379306634267171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,10240,0.30839465459187826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,10240,3.0125396728515623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,8192,0.03597653309504191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,8192,2.3733929951985675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,8192,0.11113813718159993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,7168,0.0362496018409729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,7168,0.08147626717885335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,65536,19.528363037109376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,7168,2.037555185953776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,6144,0.028739200035731
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,6144,0.06925653616587321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,6144,1.702638880411784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,5120,1.3394602457682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,5120,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,5120,0.05840213298797607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,4096,0.04802560011545817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,4096,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,4096,0.8549034754435221
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,3584,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,3584,0.5912576039632161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,3584,0.04184746742248535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,3072,0.39840428034464515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,3072,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,3072,0.03682986497879028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,2560,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,2560,0.27856213251749673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,2560,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,2048,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,2048,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,2048,0.21319680213928222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,1536,0.16660480499267577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,1536,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,1536,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,1024,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,1024,0.0736255963643392
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,1024,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,768,0.05881173213322958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,768,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,768,0.013345066706339517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,512,0.04464639822642009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,512,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,512,0.010035199920336406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,256,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,256,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,256,0.03143253326416016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,128,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,128,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,128,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,64,0.020989867051442464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,64,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,64,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,32,32,0.019180800517400107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,32,32,0.002829866607983907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,32,32,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,12288,9.025501505533855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,10240,7.28627217610677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,16384,10.933930460611979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,10240,7.069730122884114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,8192,5.680810546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,10240,14.1591552734375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,12288,16.73512980143229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,16384,22.72010294596354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,8192,5.948108927408854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,7168,4.9126739501953125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,7168,5.460002136230469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,7168,9.191492716471355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,8192,10.93041483561198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,6144,4.05951156616211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,6144,7.815611775716145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,6144,4.348723347981771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,5120,3.419682057698568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,5120,3.630962117513021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,4096,2.6188117980957033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,5120,6.428057352701823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,4096,2.9927744547526043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,4096,5.058867390950521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,3584,2.2928726196289064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,3584,4.272298685709635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,3584,2.533307647705078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,3072,2.0386133829752606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,3072,2.0733268737792967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,2560,1.6482645670572917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,3072,3.8498303731282553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,2560,1.779097620646159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,2560,3.1833770751953123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,2048,1.3662208557128905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,2048,2.3073109944661456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,1536,1.0572799682617187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,2048,1.4937088012695312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,1536,1.1638100941975913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,1536,1.83910395304362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,1024,0.7482367833455403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,1024,0.8969216028849283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,768,0.6165162404378255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,1024,1.4555136362711587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,768,1.127526346842448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,512,0.5888341267903645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,512,0.5046613375345866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,768,0.7499093373616537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,512,0.7839061101277669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,256,0.46475947697957354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,256,0.5539840062459309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,256,0.473634147644043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,128,0.4520277341206868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,128,0.46250667572021487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,128,0.5502944310506185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,64,0.5233322779337566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,64,0.4898794809977214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,64,0.4803925196329753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,65536,32,0.48793598810831706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,65536,32,0.5672597249348958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,65536,32,0.5171541213989258
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,16384,2.8568575541178385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,12288,3.44801279703776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,16384,4.6747990926106775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,16384,5.589982096354166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,12288,4.089408111572266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,12288,1.9965952555338542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,10240,1.5773663838704428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,65536,13.922850545247396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,10240,2.813371785481771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,10240,3.342472585042318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,8192,1.9220138549804688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,65536,21.033711751302086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,65536,19.27198689778646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,8192,2.6110293070475263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,8192,1.3202091217041017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,7168,1.6621898651123046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,7168,1.2390740712483725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,7168,2.191564687093099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,6144,1.2856661478678384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,6144,1.0637994766235352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,5120,1.0135210673014323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,6144,1.8408789316813152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,5120,1.465002695719401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,4096,0.801860237121582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,5120,0.874666659037272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,4096,1.2111189524332682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,4096,0.720247459411621
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,3584,0.7004842758178711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,3584,0.6534133275349935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,3584,1.021610641479492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,3072,0.60808531443278
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,3072,0.5407402674357097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,2560,0.5001898765563965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,3072,0.9053866704305014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,2560,0.7648554484049479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,2560,0.4803925196329753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,2048,0.4103850682576497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,2048,0.6170965194702148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,2048,0.40417280197143557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,1536,0.314572811126709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,1024,0.2365407943725586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,1536,0.4759210586547852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,1536,0.2931370735168457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,1024,0.18629973729451496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,1024,0.3078144073486328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,768,0.1907360076904297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,768,0.25053866704305017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,768,0.14943572680155437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,512,0.14578347206115722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,512,0.11595093409220378
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,512,0.19309226671854657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,256,0.0992256005605062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,256,0.10752000013987224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,256,0.07884799639383952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,128,0.09499306678771972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,128,0.08075946966807047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,128,0.057275732358296715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,64,0.06553599834442139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,64,0.04778666496276855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,16384,32,0.06440960168838501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,64,0.08475306828816732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,16384,32,0.03287039995193482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,16384,32,0.045841066042582194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,16384,2.1241172790527343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,16384,4.031795247395833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,16384,4.161501820882162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,12288,2.977211761474609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,12288,2.9780309041341146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,65536,10.505045572916668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,12288,1.4501205444335938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,10240,1.2833109537760417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,10240,2.4280405680338544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,10240,2.4152064005533855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,8192,1.6250539143880207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,65536,15.987336222330729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,65536,16.850564575195314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,8192,1.8891765594482421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,8192,1.0462207794189453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,7168,1.297100830078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,7168,0.8885930379231771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,7168,1.5775744120279946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,6144,0.964744504292806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,6144,1.3409951527913413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,6144,0.7962602615356446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,5120,0.8167082468668619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,4096,0.648635737101237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,5120,1.096567408243815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,5120,0.6259370803833008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,4096,0.8760661443074544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,4096,0.5380095799763998
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,3584,0.7732906977335612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,3584,0.5943637212117513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,3584,0.47759361267089845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,3072,0.4856490770975749
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,3072,0.6473663965861003
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,3072,0.3942741394042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,2560,0.4043093363444011
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,2560,0.5574314753214519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,2560,0.3217407862345377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,2048,0.31802028020222983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,2048,0.4438357353210449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,2048,0.2537471930185954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,1536,0.24934399922688805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,1536,0.3389439900716146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,1536,0.19397974014282227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,1024,0.19073492685953777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,1024,0.2235391934712728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,768,0.1539413293202718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,1024,0.1349290688832601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,768,0.17285119692484538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,768,0.11543893019358317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,512,0.11124053001403808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,512,0.08523093064626058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,512,0.1423360029856364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,256,0.08052053451538085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,128,0.06079146862030029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,256,0.07703893184661866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,256,0.06048426628112793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,128,0.043144532044728595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,128,0.04474879900614421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,64,0.05386240084966024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,64,0.03631786505381267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,64,0.034918399651845296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,12288,32,0.05236053466796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,12288,32,0.03508586486180623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,12288,32,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,16384,1.732369105021159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,16384,3.3344853719075522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,16384,3.86338144938151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,12288,2.3214080810546873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,12288,2.831495412190755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,65536,8.062327575683593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,12288,1.2813312530517578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,10240,2.2511617024739583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,65536,13.461572265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,10240,1.9601407368977863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,10240,1.1241130828857422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,8192,1.5224831899007163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,8192,1.4944255828857422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,65536,16.025940958658854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,8192,0.8923818588256835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,7168,1.2883285522460937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,7168,1.3254314422607423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,7168,0.7960917154947917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,6144,0.6807882944742839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,6144,0.903987185160319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,6144,1.0730826059977212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,5120,0.729258664449056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,5120,0.8825514475504557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,5120,0.5466794967651367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,4096,0.6003018697102864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,4096,0.45844478607177735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,4096,0.7226687749226888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,3584,0.5241173426310222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,3584,0.6283605575561524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,3584,0.37246294021606446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,3072,0.4411391894022624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,3072,0.5268821398417155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,3072,0.33228801091512045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,2560,0.368230406443278
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,2560,0.4442432085673015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,2560,0.25862827301025393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,2048,0.29296639760335286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,2048,0.347816530863444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,2048,0.21210452715555825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,1536,0.22005759874979652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,1536,0.17131519317626953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,1536,0.2673664093017578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,1024,0.16384000778198243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,1024,0.18216959635416669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,768,0.14609066645304364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,1024,0.12741973400115966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,768,0.1262933333714803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,512,0.09936213493347168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,768,0.10045440196990967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,512,0.13028693199157715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,512,0.07673172950744629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,128,0.054271999994913736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,256,0.053452801704406736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,256,0.0748202641805013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,256,0.0688810666402181
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,128,0.037922132015228274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,64,0.04747946659723918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,128,0.03768320083618164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,64,0.031129600604375203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,64,0.030105600754419964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,10240,32,0.04648960034052531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,10240,32,0.02218666672706604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,10240,32,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,16384,1.4473898569742838
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,16384,2.577680969238281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,16384,3.544780731201172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,12288,1.7661951700846354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,12288,2.612326304117839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,65536,7.162197367350261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,65536,10.524910481770833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,12288,1.0847914377848307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,10240,0.8420010884602865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,10240,1.4770858764648438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,10240,2.1107711791992188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,8192,1.3926400502522787
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,65536,14.74532674153646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,8192,1.1912490844726562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,8192,0.7343445460001627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,7168,1.1171157836914063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,6144,0.8453770955403647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,7168,1.0095957438151042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,7168,0.6035445531209309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,6144,0.8511829376220703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,6144,0.5612192153930664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,5120,0.6163455963134765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,5120,0.693998908996582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,4096,0.48076801300048827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,4096,0.3150848070780436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,5120,0.4605578740437825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,4096,0.5594794591267903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,3584,0.4558848063151042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,3584,0.483020814259847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,3584,0.2875391960144043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,3072,0.3914069175720215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,3072,0.40208959579467773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,3072,0.24152746200561523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,2560,0.21189972559611
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,2560,0.31201279958089195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,2560,0.3367253303527832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,2048,0.2502997398376465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,2048,0.27436374028523763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,2048,0.17032532691955565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,1536,0.1942186673482259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,1536,0.21661012967427573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,1024,0.1516885280609131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,1536,0.13397332827250164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,1024,0.13335893948872884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,1024,0.09874773025512695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,768,0.108405335744222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,768,0.08113493124643961
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,768,0.11868159770965576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,512,0.06157653331756592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,512,0.08454826672871908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,512,0.08819519678751628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,256,0.05939199924468994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,256,0.0562175989151001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,256,0.043212799231211345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,128,0.04761600097020467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,128,0.029422932863235475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,128,0.032767999172210696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,64,0.04164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,64,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,64,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,8192,32,0.04041386842727661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,8192,32,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,8192,32,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,16384,1.1999231974283853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,16384,2.188049062093099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,16384,3.5070292154947915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,12288,2.4721748352050783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,12288,1.5771648406982421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,65536,6.150417073567708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,65536,9.315635172526042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,12288,0.9239210764567056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,10240,1.319219207763672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,10240,0.7890602747599285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,10240,2.002124786376953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,8192,1.2854272206624349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,8192,1.159918975830078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,65536,13.981183878580728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,8192,0.6284287770589192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,7168,0.5695829391479492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,7168,1.0031786600748698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,6144,0.7740074793497722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,7168,1.046186637878418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,6144,0.8348672231038412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,6144,0.4742485364278157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,5120,0.6304085413614909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,5120,0.6990847905476888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,5120,0.3919189453125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,4096,0.44489386876424153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,4096,0.5501269022623698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,3584,0.3761834780375163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,4096,0.28491093317667643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,3584,0.4750677426656087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,3584,0.24524374008178712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,3072,0.35778560638427737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,3072,0.4042069435119629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,3072,0.21070507367451988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,2560,0.2869248072306315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,2560,0.34344959259033203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,2048,0.23179945945739747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,2048,0.27849388122558594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,2560,0.18274985949198405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,2048,0.15144960085550946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,1536,0.17332906723022462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,1536,0.2263040065765381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,1024,0.12144640286763508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,1536,0.11786239941914875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,1024,0.15291733741760255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,1024,0.08642559846242269
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,768,0.0717141310373942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,768,0.12544000148773193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,768,0.10027946631113689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,512,0.0807253360748291
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,512,0.08574293454488119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,512,0.054135465621948244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,256,0.05529599984486898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,256,0.04423679908116658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,256,0.03870720068613688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,128,0.04416853189468384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,128,0.029525333642959596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,128,0.027716267108917236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,64,0.03874133427937825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,64,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,64,0.02461013396581014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,7168,32,0.03768320083618164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,7168,32,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,7168,32,0.024913066625595094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,16384,1.0957141876220704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,16384,1.7741483052571614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,16384,3.237444305419922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,12288,1.3159423828125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,65536,5.185158284505208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,12288,2.360797882080078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,65536,8.073250325520833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,12288,0.7998122533162435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,10240,1.078818130493164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,10240,1.8952533721923828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,10240,0.6949888229370117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,8192,0.8468821207682291
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,8192,1.2969642639160157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,65536,13.426585896809897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,8192,0.5393386840820312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,7168,0.9814997355143229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,7168,0.4656810760498047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,6144,0.715878423055013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,7168,0.7311018625895183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,6144,0.37191680272420247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,6144,0.6242645263671875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,5120,0.5909845352172851
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,3584,0.3469312032063802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,4096,0.393830394744873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,4096,0.2427903970082601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,5120,0.30853118896484377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,4096,0.4475562731424968
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,3584,0.3462826728820801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,5120,0.5139114697774251
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,3584,0.21087573369344076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,3072,0.29337600072224934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,3072,0.30088532765706383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,3072,0.18046293258666993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,2560,0.24903680483500162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,2560,0.15899306933085125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,2560,0.26869331995646156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,2048,0.20524373054504394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,2048,0.20821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,2048,0.13189120292663575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,1536,0.15062932968139647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,1536,0.1016149361928304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,1536,0.16223573684692383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,1024,0.07563947041829427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,1024,0.10912426312764485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,1024,0.11393706798553467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,768,0.09277439912160237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,768,0.08932693004608154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,768,0.06219093402226766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,512,0.06833493709564209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,512,0.06399999856948853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,512,0.04741119941075643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,256,0.050619733333587644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,256,0.038365864753723146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,256,0.03406293392181396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,128,0.040686933199564616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,128,0.024574933449427287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,128,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,64,0.03573760191599528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,64,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,64,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,6144,32,0.034747731685638425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,6144,32,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,6144,32,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,16384,3.1569236755371093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,65536,4.075724792480469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,16384,1.4900906880696616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,16384,0.8490666707356771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,12288,2.23812255859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,65536,6.486015828450521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,12288,1.0991605122884116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,12288,0.6761813481648763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,65536,12.6701904296875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,10240,0.5913941065470378
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,10240,0.8933034896850586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,10240,1.8688308715820312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,8192,1.128277333577474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,8192,0.7317162831624349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,8192,0.46451625823974607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,7168,0.9137834548950196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,7168,0.7182336171468099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,6144,0.6052864074707032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,6144,0.2962090810139974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,7168,0.35870720545450846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,6144,0.642355219523112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,5120,0.5215914726257325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,5120,0.4871509234110515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,5120,0.24384853045145669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,4096,0.39755093256632484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,4096,0.3870037396748861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,4096,0.19848532676696778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,3584,0.17841493288675944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,3584,0.35549653371175133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,3584,0.33952213923136393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,3072,0.2939157485961914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,3072,0.2691754659016927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,3072,0.1545215924580892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,2560,0.22432427406311034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,2560,0.2513919989267985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,2560,0.13247146606445312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,2048,0.18906453450520833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,2048,0.20759894053141276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,2048,0.10956799983978271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,1536,0.16629759470621747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,1536,0.08731306393941243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,1024,0.09669973055521647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,1536,0.13325653076171876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,1024,0.10639359951019287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,1024,0.0636245330174764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,768,0.08355840047200522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,768,0.08341973622639974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,768,0.05253119866053263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,512,0.06174720128377279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,512,0.06635520060857138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,256,0.046250665187835695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,512,0.04041386842727661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,256,0.029286400477091475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,256,0.032051199674606325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,128,0.037102933724721274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,128,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,128,0.022528000672658286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,64,0.03246080080668132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,64,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,64,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,5120,32,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,5120,32,0.031436800956726074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,5120,32,0.01337386667728424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,16384,2.911572265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,65536,3.6985824584960936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,16384,0.706935437520345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,65536,5.2235265096028645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,16384,1.1826122283935547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,12288,0.8691018422444662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,12288,2.1186901092529298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,12288,0.5629610697428385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,65536,11.967965698242187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,10240,0.6954271952311198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,10240,0.45769386291503905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,10240,1.7654783884684246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,8192,0.5483178456624349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,8192,1.0608298619588217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,8192,0.3073024113972982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,7168,0.8410111745198569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,7168,0.2711893399556478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,7168,0.46065705617268876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,6144,0.38789119720458987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,6144,0.6344021479288737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,6144,0.2360991954803467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,5120,0.3219797452290853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,5120,0.46090240478515626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,5120,0.19722240765889484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,4096,0.35413331985473634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,4096,0.2581823984781901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,4096,0.16202774047851562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,3584,0.31525545120239257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,3584,0.2308095932006836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,3584,0.14315519332885743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,3072,0.26030079523722327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,3072,0.1253717343012492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,3072,0.19821227391560872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,2560,0.19042347272237142
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,2560,0.16496639251708983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,2560,0.10728106498718262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,2048,0.14049280484517415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,2048,0.14626132647196452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,1536,0.11823786894480388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,1536,0.1093290646870931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,1536,0.07031466960906982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,2048,0.08878080050150552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,1024,0.08553813298543295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,1024,0.07710613409678141
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,768,0.06123520135879516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,1024,0.05188266833623251
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,768,0.06867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,768,0.042700799306233723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,512,0.05509119828542074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,512,0.04293973445892334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,512,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,256,0.04164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,256,0.02693120042483012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,128,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,128,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,256,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,64,0.014472533265749613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,64,0.029320534070332843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,128,0.033240532875061034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,64,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,4096,32,0.028706133365631104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,4096,32,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,4096,32,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,16384,2.9570388793945312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,16384,0.7285077412923177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,65536,3.2908971150716146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,16384,1.0273450851440429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,65536,4.7680511474609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,12288,0.7847935994466145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,12288,2.1079722086588544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,12288,0.5461333592732747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,8192,1.098581314086914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,10240,0.4708693186442058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,10240,1.6888832092285155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,10240,0.6544714609781901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,65536,11.9372802734375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,8192,0.5116223971048991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,8192,0.3006805419921875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,7168,0.8382805506388346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,7168,0.27415892283121746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,6144,0.5891754786173503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,5120,0.4501162528991699
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,7168,0.4413781483968099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,5120,0.3196586608886719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,6144,0.37908480962117513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,6144,0.2364415963490804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,5120,0.19647146860758463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,4096,0.3506175994873047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,4096,0.2580479939778646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,4096,0.16646827061971028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,3584,0.22814720471700034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,3584,0.317303466796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,3584,0.15083519617716473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,3072,0.2682880083719889
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,3072,0.19664212862650554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,3072,0.1295360008875529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,2560,0.19783573150634765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,2560,0.16554667154947916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,2560,0.10594560305277508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,2048,0.1454421361287435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,2048,0.13550933202107746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,1536,0.1182037353515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,2048,0.08792746861775716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,1536,0.10475520292917888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,1024,0.0834218660990397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,1536,0.0756053368250529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,1024,0.07379626433054606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,768,0.05710506836573283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,768,0.06727680365244547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,1024,0.05109759966532389
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,512,0.053725866476694736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,768,0.042496001720428465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,512,0.03211946686108907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,512,0.045670398076375324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,256,0.04010453224182129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,256,0.02645333409309387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,256,0.02375146746635437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,128,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,128,0.03293866713841756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,128,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,64,0.028910932938257854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,64,0.012935466567675271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,64,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3584,32,0.028123732407887774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3584,32,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3584,32,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,16384,0.8851093292236328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,16384,0.5927594502766926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,65536,2.8570281982421877
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,16384,2.890205891927083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,65536,4.787131754557292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,12288,0.6648149490356445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,12288,2.010760498046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,12288,0.4347221374511719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,10240,1.5979520161946614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,10240,0.569378153483073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,65536,11.394150797526041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,8192,0.2321407953898112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,10240,0.32539307276407875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,8192,0.49943892161051434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,8192,1.0596352259318034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,7168,0.7779616038004558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,7168,0.4240725199381511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,7168,0.20763306617736815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,6144,0.3642367998758952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,5120,0.3081557273864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,6144,0.5760682423909504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,5120,0.41277440388997394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,6144,0.17465707461039226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,5120,0.15025493303934734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,4096,0.318121592203776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,4096,0.24627200762430826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,4096,0.12335573037465412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,3584,0.27709439595540364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,3584,0.10799787044525147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,3584,0.21937386194864908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,3072,0.23773226737976075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,3072,0.1888256072998047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,3072,0.09564159711201986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,2560,0.17919999758402508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,2560,0.1563989321390788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,2560,0.08185173670450846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,2048,0.1263274669647217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,1536,0.09816746711730957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,2048,0.12704426447550457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,1536,0.0957098642985026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,2048,0.06891520023345947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,1536,0.05437440077463786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,1024,0.0713045358657837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,1024,0.06884693304697673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,768,0.059699201583862306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,1024,0.040686933199564616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,768,0.0338261326154073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,768,0.05130240122477213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,512,0.035327998797098796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,512,0.04833279848098755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,512,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,256,0.037034666538238524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,128,0.029730133215586346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,256,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,256,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,128,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,128,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,64,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,64,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,64,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,3072,32,0.02542933424313863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,3072,32,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,3072,32,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,16384,0.7409664154052734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,65536,2.6297342936197916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,16384,0.5638485590616862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,65536,3.140264638264974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,16384,2.749713134765625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,12288,1.985433578491211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,12288,0.5831680297851562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,12288,0.42052052815755203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,10240,0.46445226669311523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,65536,11.301648966471355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,10240,1.6479231516520183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,10240,0.30419626235961916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,8192,0.3558058738708496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,8192,0.22609920501708985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,8192,0.9738239924112955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,7168,0.7676533381144206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,7168,0.3385685284932455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,7168,0.2121386686960856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,6144,0.17947306632995605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,6144,0.5360640207926433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,6144,0.3063125292460124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,5120,0.4000085194905599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,5120,0.15462400118509928
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,5120,0.2690389315287272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,4096,0.3084287961324056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,4096,0.212990935643514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,4096,0.12100266615549723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,3584,0.27900587717692055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,3584,0.10711039702097576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,3584,0.19220159848531088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,3072,0.2268160025278727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,3072,0.09390079975128174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,2560,0.17411413192749023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,3072,0.16800427436828613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,2560,0.14257280031840008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,2560,0.08372906843821207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,2048,0.12441600163777669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,2048,0.12369919617970784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,2048,0.07618559996287028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,1536,0.09444693724314371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,1536,0.05256319840749105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,1536,0.09383253256479898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,1024,0.06983679930369059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,1024,0.07096319993336996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,1024,0.03915093342463176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,768,0.04795733292897542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,768,0.032255999247233075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,768,0.05870720148086548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,512,0.04747946659723918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,512,0.03389439980189006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,256,0.035598933696746826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,512,0.025326933463414508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,128,0.02924906611442566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,256,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,256,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,128,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,128,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,64,0.02573440074920654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,64,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,64,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2560,32,0.024985599517822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2560,32,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2560,32,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,65536,2.143163808186849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,65536,2.830811818440755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,16384,2.6121899922688803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,16384,0.40751787821451824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,16384,0.5739178975423177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,12288,0.4171776135762532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,12288,1.8912544250488281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,65536,10.768418375651041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,12288,0.29259093602498376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,10240,0.3247104008992513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,10240,0.2017279942830404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,8192,0.2611541271209717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,10240,1.6149162292480468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,8192,0.910472551981608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,8192,0.16424427032470704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,7168,0.13850986162821452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,7168,0.22026240030924477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,7168,0.7772501627604167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,6144,0.48906240463256834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,6144,0.19500373204549154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,5120,0.1650346597035726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,6144,0.12315306663513184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,5120,0.3546794573465983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,5120,0.10335573355356853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,4096,0.2905023892720541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,4096,0.13107199668884278
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,4096,0.0859818696975708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,3584,0.24128212928771972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,3584,0.11871573130289714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,3584,0.07365972995758056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,3072,0.20387840270996094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,3072,0.09943040211995444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,3072,0.06703786849975586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,2560,0.1569109280904134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,2560,0.08376320203145346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,2560,0.0589141329129537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,2048,0.10909012953440349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,2048,0.0703488032023112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,1536,0.080076797803243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,1536,0.05509119828542074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,1536,0.03887253204981486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,2048,0.0484661340713501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,1024,0.059869865576426186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,1024,0.044677333037058516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,1024,0.029013333717981975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,768,0.03570133447647095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,768,0.05065066814422607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,768,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,512,0.04119893312454224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,512,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,512,0.025361067056655882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,256,0.03136853377024333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,256,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,256,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,128,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,128,0.012148267030715943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,128,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,64,0.022835199038187662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,64,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,64,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,2048,32,0.02204479972521464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,2048,32,0.0066890666882197065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,2048,32,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,16384,0.39195305506388345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,65536,1.8903040568033855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,65536,1.8966506958007812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,16384,0.44717973073323564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,16384,2.601574452718099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,12288,0.3043327967325846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,12288,1.8752170562744142
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,12288,0.22548480033874513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,10240,0.26170026461283363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,10240,0.18486612637837727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,8192,0.21364053090413412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,10240,1.5600298563639323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,65536,10.65724589029948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,8192,0.893400510152181
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,8192,0.1510741392771403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,7168,0.7098709106445312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,6144,0.44946772257486983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,6144,0.11417600313822429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,7168,0.13096746603647869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,6144,0.17117865880330402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,7168,0.23282346725463868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,5120,0.3657045364379883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,5120,0.14305280049641927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,5120,0.09970346291859945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,4096,0.27627518971761067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,3584,0.07072426478068033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,4096,0.13271040121714275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,3584,0.24907093048095702
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,3584,0.11878399848937989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,4096,0.08850773175557455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,3072,0.1998165289560954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,3072,0.10584746996561686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,3072,0.06253226598103842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,2560,0.13697706858317057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,2560,0.05277013381322225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,2560,0.09249599774678549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,2048,0.10236266454060872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,2048,0.08205653031667073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,2048,0.0443391998608907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,1536,0.07918933232625326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,1536,0.06191786527633667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,1536,0.035292800267537436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,1024,0.05905066728591919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,1024,0.040106666088104245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,1024,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,768,0.05021013418833414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,768,0.0318122665087382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,768,0.022528000672658286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,512,0.04085760116577149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,512,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,512,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,256,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,256,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,256,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,128,0.011980799833933513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,128,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,128,0.025252266724904375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,64,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,64,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,64,0.010410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1536,32,0.021845332781473794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1536,32,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1536,32,0.010513066252072652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,65536,1.4665386199951171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,16384,0.28177067438761394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,16384,0.2631285349527995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,65536,1.3045077006022134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,16384,2.47019526163737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,12288,0.2053119977315267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,12288,1.780736033121745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,12288,0.13458773295084636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,10240,0.1737386703491211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,10240,1.4207317352294921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,10240,0.11147946516672771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,65536,10.196104939778646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,8192,0.835857073465983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,8192,0.14762667020161946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,8192,0.0889514684677124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,7168,0.12100266615549723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,7168,0.6594559987386067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,7168,0.08110079765319825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,6144,0.10656426747639973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,6144,0.07096319993336996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,5120,0.313480536142985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,6144,0.38874025344848634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,5120,0.06055253346761068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,5120,0.08690240383148193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,4096,0.2569215933481852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,4096,0.07153813044230142
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,4096,0.048264535268147786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,3584,0.04304213523864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,3584,0.22121814092000328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,3584,0.06433493296305338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,3072,0.1821013291676839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,3072,0.05205333232879639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,2560,0.11690346399943034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,3072,0.03781973520914714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,2560,0.04498773415883382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,2560,0.03338239987691243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,2048,0.08475306828816732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,2048,0.0366538683573405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,2048,0.02788693308830261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,1536,0.06543360153834024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,1536,0.035327998797098796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,1536,0.02225493391354879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,1024,0.04925440152486165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,1024,0.024779733022054037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,1024,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,768,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,768,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,768,0.041773867607116696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,512,0.0340992013613383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,256,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,512,0.01221440037091573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,512,0.014363732933998109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,256,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,256,0.010342400272687275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,128,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,128,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,128,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,64,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,64,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,1024,32,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,64,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,1024,32,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,1024,32,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,65536,1.3271711985270183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,16384,0.23190186818440756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,16384,0.22425600687662758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,65536,1.0095317204793295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,16384,2.4446975708007814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,12288,0.16718079249064127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,12288,1.7587540944417317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,12288,0.1165994644165039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,10240,0.09697279930114747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,10240,1.406395721435547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,10240,0.14281387329101564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,8192,0.8260255813598633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,65536,10.032469685872396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,8192,0.12133973439534504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,8192,0.07700479825337728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,7168,0.6434815724690754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,7168,0.0683690627415975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,7168,0.10317866802215576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,6144,0.05925546487172445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,6144,0.08949759801228842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,5120,0.30207999547322595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,6144,0.401476256052653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,5120,0.07512746651967367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,5120,0.051268267631530764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,4096,0.23036586443583168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,4096,0.040584532419840495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,4096,0.06082559823989868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,3584,0.20135253270467124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,3584,0.03689813216527303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,3072,0.17718613942464193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,3584,0.054681599140167236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,3072,0.04703573385874431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,3072,0.032631466786066696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,2560,0.12550826867421466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,2560,0.03973120053609212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,2048,0.08280746936798096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,2048,0.03829760154088338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,2560,0.02867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,2048,0.024166399240493776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,1536,0.06509226560592651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,1536,0.03051519989967346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,1536,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,1024,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,1024,0.048742401599884036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,1024,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,768,0.04133546749750773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,768,0.013925333817799887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,768,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,512,0.03392746845881144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,512,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,512,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,256,0.02590506672859192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,256,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,256,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,128,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,128,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,64,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,128,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,64,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,64,0.0077802668015162155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,768,32,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,768,32,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,768,32,0.004434133569399515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,16384,0.13438293139139812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,65536,0.6599338531494141
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,65536,1.2737876892089843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,16384,0.2020352045694987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,16384,2.5153877258300783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,12288,0.10410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,12288,1.7445205688476562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,12288,0.10577920277913411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,10240,0.09437867005666098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,10240,0.0881322701772054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,10240,1.4375253041585287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,65536,9.990826416015626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,8192,0.8186880111694336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,8192,0.07041599750518798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,8192,0.0771071990331014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,7168,0.6352213541666667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,7168,0.06700373490651448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,7168,0.062327468395233156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,6144,0.4336640040079753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,5120,0.3138901392618815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,6144,0.06103039979934692
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,5120,0.050619733333587644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,6144,0.054135465621948244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,4096,0.23456427256266274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,5120,0.04666026830673218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,4096,0.041365333398183185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,4096,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,3584,0.2004650592803955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,3584,0.036761601765950516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,3584,0.03379199902216594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,3072,0.16673812866210938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,2560,0.1230847994486491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,3072,0.03484586477279663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,3072,0.029795199632644653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,2560,0.029969066381454468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,2560,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,2048,0.08034986654917399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,2048,0.023859200874964397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,1536,0.06440960168838501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,1536,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,1536,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,2048,0.022459733486175536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,1024,0.04843519926071167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,1024,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,1024,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,768,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,768,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,768,0.04078720013300578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,512,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,512,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,256,0.025668267409006757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,512,0.010922666390736897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,256,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,256,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,128,0.021367466449737547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,128,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,128,0.004743466774622599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,64,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,64,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,512,32,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,512,32,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,512,32,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,16384,0.07509333292643229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,65536,1.2576768239339193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,65536,0.5584554672241211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,16384,0.16018773714701334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,16384,2.4357205708821614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,12288,0.05836799939473471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,12288,1.748855463663737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,12288,0.10557440121968586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,10240,0.08751786549886068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,10240,0.04856959978739421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,10240,1.4351701100667318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,65536,9.985774739583333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,8192,0.04113066593805949
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,8192,0.07024640242258708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,8192,0.8216234842936198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,7168,0.6305109024047851
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,7168,0.037614933649698895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,7168,0.0618837316830953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,6144,0.38942718505859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,6144,0.032085333267847696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,6144,0.054067198435465494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,5120,0.3042986551920573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,5120,0.02935466567675273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,5120,0.04638719956080119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,4096,0.2339146614074707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,4096,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,4096,0.037205334504445395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,3584,0.20729173024495445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,3584,0.02255786657333374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,3584,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,3072,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,3072,0.0295199990272522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,3072,0.16278187433878583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,2560,0.13001386324564618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,2560,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,2560,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,2048,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,2048,0.0816810687383016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,2048,0.021845332781473794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,1536,0.06423679987589517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,1536,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,1536,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,1024,0.009795199831326802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,1024,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,768,0.040550398826599124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,1024,0.048093867301940915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,768,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,768,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,512,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,512,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,512,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,256,0.02529279987017314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,256,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,256,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,128,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,128,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,64,0.019078399737675986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,64,0.007574399809042613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,256,32,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,64,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,256,32,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,256,32,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,65536,0.4651349385579427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,65536,1.2552874247233072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,16384,0.03799039920171102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,16384,0.15100053151448567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,12288,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,16384,2.426436360677083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,12288,1.7475242614746094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,12288,0.10639359951019287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,10240,0.031778132915496825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,10240,1.4053376515706382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,65536,9.972941080729168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,10240,0.08768426577250163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,8192,0.025975465774536133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,8192,0.8172202428181967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,8192,0.07014400164286295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,7168,0.6373034795125325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,7168,0.025156267484029132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,7168,0.06208853324254354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,6144,0.4314112027486165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,6144,0.054340267181396486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,5120,0.303547732035319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,6144,0.02225493391354879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,5120,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,5120,0.046284798781077066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,4096,0.22575573921203612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,4096,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,4096,0.037478399276733396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,3584,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,3584,0.20650666554768882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,3584,0.03334826628367106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,3072,0.17486507097880047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,3072,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,3072,0.029522132873535157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,2560,0.12342613538106281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,2560,0.025804799795150758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,2560,0.014509866635004679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,2048,0.08038400014241537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,2048,0.012769066294034324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,1536,0.06423893372217813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,2048,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,1536,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,1536,0.018051199118296304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,1024,0.048128000895182294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,1024,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,1024,0.014472533265749613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,768,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,768,0.040550398826599124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,512,0.03314346671104431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,768,0.012900267044703165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,512,0.005118933320045471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,512,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,256,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,256,0.004129066566626231
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,128,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,256,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,64,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,128,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,128,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,64,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,64,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,128,32,0.01808746655782064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,128,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,128,32,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,16384,0.03805866638819377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,65536,0.4464981396993001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,65536,1.2587690989176432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,16384,0.15155199368794758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,16384,2.4321024576822916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,12288,0.029969066381454468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,12288,1.746124776204427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,12288,0.105403733253479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,10240,0.08704000314076742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,10240,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,8192,0.8204287846883138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,65536,9.998267618815104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,10240,1.404381815592448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,8192,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,8192,0.07000746726989746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,7168,0.6348117192586262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,7168,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,7168,0.06157013177871704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,6144,0.4129791895548503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,6144,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,6144,0.05379413366317749
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,5120,0.3126954714457194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,5120,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,5120,0.04614826838175456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,4096,0.23889919916788735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,4096,0.037137067317962645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,4096,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,3584,0.20434986750284828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,3072,0.17131519317626953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,3584,0.03310933311780294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,3584,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,3072,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,3072,0.029320534070332843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,2560,0.12561066945393878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,2560,0.025565866629282636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,2560,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,2048,0.0816810687383016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,2048,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,2048,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,1536,0.06379520098368327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,1536,0.009687466422716777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,1536,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,1024,0.04775253136952718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,1024,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,1024,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,768,0.040516265233357746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,768,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,768,0.01242453356583913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,512,0.032767999172210696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,512,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,256,0.025326933463414508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,256,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,256,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,128,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,128,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,128,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,64,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,64,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,64,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,64,32,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,64,32,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,64,32,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,65536,0.39935998916625975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,65536,1.253819783528646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,16384,0.031129600604375203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,16384,0.14885546366373698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,16384,2.430225118001302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,12288,0.025634133815765382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,12288,1.7361578623453777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,12288,0.1055402676264445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,10240,0.022425599892934165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,10240,1.4040746053059896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,10240,0.0870741367340088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,65536,9.974306233723958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,8192,0.8275285085042319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,8192,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,8192,0.06980266571044921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,7168,0.6328320185343425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,7168,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,7168,0.062122666835784913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,6144,0.4204885482788086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,6144,0.05358933210372925
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,5120,0.2982912063598633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,6144,0.016109866897265117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,5120,0.012999467055002847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,5120,0.04597760041554769
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,4096,0.22763519287109374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,4096,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,4096,0.0370688001314799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,3584,0.20336640675862633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,3584,0.010069333513577779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,3072,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,3584,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,3072,0.029525333642959596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,3072,0.16950613657633465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,2560,0.12482559680938721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,2560,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,2560,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,2048,0.08045226732889811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,1536,0.06410239934921265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,2048,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,1536,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,1536,0.007469866673151653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,2048,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,1024,0.04788906574249267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,1024,0.005767466624577841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,1024,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,768,0.040516265233357746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,768,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,768,0.012458667159080505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,512,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,512,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,512,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,256,0.025081600745519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,256,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,256,0.008940800031026205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,128,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,128,0.020951465765635172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,128,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,64,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,64,0.007368533313274384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2048,32,32,0.01825493375460307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2048,32,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2048,32,32,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,16384,5.754914347330729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,12288,4.642235819498698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,16384,5.978350830078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,12288,4.186385091145834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,10240,3.41575673421224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,12288,8.514082336425782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,10240,6.876160176595052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,16384,11.451289876302083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,10240,3.8647466023763024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,8192,2.652637990315755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,8192,2.6689193725585936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,7168,2.350352986653646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,8192,5.566805521647135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,7168,2.553070831298828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,7168,4.860586547851563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,6144,2.0407615661621095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,6144,2.037282180786133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,5120,1.6121471405029297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,6144,3.975714111328125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,5120,1.8197845458984374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,5120,3.2635220845540367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,4096,1.323690668741862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,4096,1.4214144388834635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,4096,2.5361707051595053
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,3584,1.165994644165039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,3584,1.26112429300944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,3072,0.9916757583618164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,3584,2.1581141153971357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,3072,1.8542933146158855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,3072,1.0774528503417968
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,2560,0.8571903864542643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,2560,0.9348096211751302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,2560,1.568904495239258
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,2048,0.6917802810668945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,2048,0.7488170623779297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,2048,1.188522720336914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,1536,0.5590698877970378
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,1536,0.944093894958496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,1536,0.5837770462036133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,1024,0.41175041198730467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,1024,0.44817066192626953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,1024,0.6381226857503255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,768,0.32655359903971354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,768,0.5766485214233399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,768,0.3799722671508789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,512,0.28723198572794595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,512,0.41734825770060224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,256,0.23746560414632162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,512,0.30334294637044273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,256,0.28654934565226237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,128,0.22405120531717934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,128,0.2609834671020508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,256,0.2481152057647705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,128,0.2208768049875895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,64,0.2337109406789144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,64,0.22981972694396974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,64,0.22763519287109374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,65536,32,0.22654293378194174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,65536,32,0.26419199307759605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,65536,32,0.23975253105163574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,16384,2.010760498046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,16384,1.4474922180175782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,12288,1.250645319620768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,16384,2.654549407958984
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,12288,1.8701653798421223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,65536,6.773962910970051
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,12288,1.0724693298339845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,10240,1.0025301615397135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,10240,0.9221802393595377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,65536,10.115038045247395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,10240,1.5497557322184243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,8192,0.8033280054728189
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,65536,10.691959635416667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,8192,1.2124501546223958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,8192,0.6906506856282552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,7168,0.707754643758138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,7168,0.6387029647827148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,6144,0.6219776153564454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,6144,0.5804373423258464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,6144,0.900810686747233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,7168,1.0570069630940755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,5120,0.49339733123779295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,5120,0.7311701456705729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,5120,0.496230411529541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,4096,0.5903018951416016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,4096,0.4007253328959147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,4096,0.3992234547932943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,3584,0.36450986862182616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,3584,0.5207722663879395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,3584,0.3252565383911133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,3072,0.31825920740763347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,3072,0.4449962615966797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,3072,0.2506410598754883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,2560,0.2732330640157064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,2560,0.36829865773518883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,2560,0.21845332781473795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,2048,0.2899626731872559
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,2048,0.22951253255208334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,2048,0.17534292538960775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,1536,0.1737386703491211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,1536,0.2146986643473307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,1536,0.141482671101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,1024,0.11806720097859699
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,1024,0.10168320337931316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,1024,0.14813332557678222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,768,0.1002837340037028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,768,0.11779413223266602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,768,0.08226133187611898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,512,0.07618559996287028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,512,0.08447999954223633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,512,0.06215360164642334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,256,0.056797866026560465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,256,0.05573973258336386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,256,0.04300800164540609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,128,0.045499734083811444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,128,0.042530135313669844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,128,0.030958932638168336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,64,0.04072106679280599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,64,0.03218773404757182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,64,0.025941334168116253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,16384,32,0.040447998046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,16384,32,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,16384,32,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,16384,1.6682666778564452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,16384,1.8837844848632812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,16384,1.1146527608235677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,65536,4.910383097330729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,12288,1.0464256286621094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,12288,1.390762710571289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,65536,7.986381022135417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,65536,8.732023111979167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,12288,0.8488277435302735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,10240,0.8089258829752604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,10240,1.0997759501139321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,8192,0.6487040201822917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,10240,0.7105194727579753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,8192,0.9017344156901042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,8192,0.5825877507527669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,7168,0.5671253204345703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,7168,0.7663274765014648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,7168,0.4555775960286458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,6144,0.48079894383748367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,6144,0.441753609975179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,6144,0.6601386388142904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,5120,0.5487936019897461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,5120,0.40645545323689775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,4096,0.33153705596923827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,5120,0.3430741310119629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,4096,0.4396032015482585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,4096,0.25053332646687826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,3584,0.2996906598409017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,3584,0.3816416104634603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,3584,0.22036479314168295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,3072,0.3173376083374023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,3072,0.25931092898050945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,3072,0.19268159866333007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,2560,0.2123093287150065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,2560,0.25613652865091957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,2560,0.16424959500630695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,2048,0.17328853607177735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,2048,0.20650666554768882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,2048,0.1374890645345052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,1536,0.16332799593607586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,1536,0.12936533292134603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,1536,0.10564266840616862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,1024,0.09881599744160971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,1024,0.11451733112335205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,768,0.08376320203145346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,1024,0.07792320251464843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,768,0.0897706667582194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,768,0.06311253309249878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,512,0.06423893372217813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,512,0.06287360191345215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,512,0.04747626781463623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,256,0.04754773378372192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,256,0.03344426552454631
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,128,0.039082666238149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,128,0.024576000372568765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,256,0.03997013171513875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,128,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,64,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,64,0.03498239914576213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,64,0.020507733027140297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,12288,32,0.014711466431617738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,12288,32,0.0339626669883728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,12288,32,0.020002132654190062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,16384,1.5494144439697266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,16384,0.9287680308024088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,16384,1.5387306213378906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,12288,1.1067391713460286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,12288,0.9315669377644857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,65536,4.258440653483072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,65536,7.006202697753906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,65536,8.052565511067709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,12288,0.7087445576985677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,10240,0.7231829325358073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,10240,0.5608416239420573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,10240,0.9212245305379232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,8192,0.5703680038452148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,8192,0.7692970911661784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,8192,0.45093545913696287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,7168,0.5068117459615071
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,7168,0.4177237192789714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,7168,0.7485098520914714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,6144,0.42874879837036134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,6144,0.6425600051879883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,6144,0.3399680137634277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,5120,0.35611305236816404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,5120,0.5318655967712402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,5120,0.2673322677612305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,4096,0.2958677291870117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,4096,0.4273834546407064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,4096,0.20541440645853676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,3584,0.25784319241841636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,3584,0.3526656150817871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,3072,0.21821440060933434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,3584,0.18339840571085614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,3072,0.29532159169514977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,2560,0.2527573267618815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,3072,0.15981225967407225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,2560,0.1795072078704834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,2560,0.13854719797770182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,2048,0.13783040046691894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,2048,0.20654080708821615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,1536,0.11513173580169678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,2048,0.11018239657084147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,1536,0.09004373550415039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,1024,0.08874666690826416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,1024,0.06488639911015828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,1024,0.1101482629776001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,1536,0.16663893063863117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,768,0.06918826897939047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,768,0.08482133547465007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,512,0.056251732508341465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,768,0.05232640107472738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,512,0.03979946772257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,512,0.0612010677655538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,256,0.04341760079065959
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,256,0.031845333178838094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,128,0.03560106754302979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,256,0.02788693308830261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,128,0.020411733786265054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,64,0.03170453310012818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,64,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,64,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,128,0.020614399512608846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,10240,32,0.03112746675809224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,10240,32,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,10240,32,0.016924800475438435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,16384,0.7438336054484049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,16384,1.211797332763672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,16384,1.3568330128987631
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,65536,3.0447616577148438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,12288,0.8106666564941406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,12288,0.8747349421183268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,65536,5.32862294514974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,65536,7.3729705810546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,12288,0.5590325037638346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,10240,0.6249472300211589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,10240,0.455236275990804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,10240,0.7451306660970052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,8192,0.4944213231404622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,8192,0.5715573628743489
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,8192,0.34805758794148767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,7168,0.45315411885579426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,7168,0.5012480099995931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,6144,0.38604799906412757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,7168,0.26715946197509766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,6144,0.23053654034932455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,6144,0.428817081451416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,5120,0.30412801106770837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,5120,0.3452927907307943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,5120,0.1974613348642985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,4096,0.2402634620666504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,4096,0.1559893290201823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,3584,0.22244693438212076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,4096,0.2584917386372884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,3584,0.13919572830200194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,3072,0.18551467259724935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,3584,0.2290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,3072,0.19930453300476075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,3072,0.12178773085276287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,2560,0.14704640706380206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,2560,0.1055402676264445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,2560,0.16493226687113444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,2048,0.11950080394744873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,2048,0.08755199909210205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,2048,0.13745493888854982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,1536,0.10257066885630291
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,1536,0.10816853046417237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,1536,0.06901760101318359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,1024,0.07246507008870443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,1024,0.07758506933848063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,1024,0.05044906536738077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,768,0.06010239919026693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,768,0.04184746742248535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,768,0.060893865426381436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,512,0.04969813426335652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,512,0.04348586797714234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,512,0.031982932488123575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,256,0.03867306709289551
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,256,0.0269269327322642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,128,0.031846400101979574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,256,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,128,0.01750613252321879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,64,0.028706133365631104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,128,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,64,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,64,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,8192,32,0.027716267108917236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,8192,32,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,8192,32,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,16384,0.642355219523112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,16384,1.0263925552368165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,16384,1.363421885172526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,12288,0.7755093256632487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,65536,3.16409174601237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,65536,4.416473388671875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,12288,0.789299201965332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,65536,7.3441619873046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,12288,0.5213184038798014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,10240,0.6408575693766276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,10240,0.44762452443440753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,10240,0.6195199966430665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,8192,0.49315840403238936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,8192,0.541047477722168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,8192,0.33153705596923827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,7168,0.2666837374369303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,7168,0.45079574584960935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,7168,0.46260906855265305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,6144,0.3972096125284831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,6144,0.23364267349243165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,6144,0.3689130783081055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,5120,0.3035818735758463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,5120,0.32034133275349935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,5120,0.20032854080200196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,4096,0.1650005340576172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,4096,0.25825279553731284
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,3584,0.1932970682779948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,4096,0.2263381322224935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,3584,0.22743040720621743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,3584,0.14015146891276042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,3072,0.1773226737976074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,3072,0.12748800118764242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,3072,0.1974613348642985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,2560,0.14411093393961588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,2560,0.16609279314676922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,2048,0.12202560106913249
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,2048,0.13527040481567382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,2560,0.10728106498718262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,2048,0.08833599885304769
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,1536,0.10089813073476155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,1536,0.10523306528727214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,1536,0.06785706679026285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,1024,0.07041172981262207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,1024,0.04997119903564453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,1024,0.07383039792378744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,768,0.05857280095418295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,768,0.05754239956537882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,768,0.0413696010907491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,512,0.048162134488423665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,512,0.04119253158569336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,512,0.03109546701113383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,256,0.03689813216527303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,256,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,256,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,128,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,128,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,128,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,64,0.02805759906768799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,64,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,64,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,7168,32,0.027409066756566364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,7168,32,0.010269866387049357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,7168,32,0.014916266997655234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,16384,0.5576704025268555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,16384,1.1990357716878255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,16384,0.9295520146687826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,65536,2.2989481608072917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,65536,3.804740142822266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,12288,0.6906197230021159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,12288,0.6860458374023437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,65536,6.71829325358073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,12288,0.39816532135009763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,10240,0.3032064119974772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,10240,0.5842261632283529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,10240,0.5434709548950195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,8192,0.4459520022074382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,8192,0.23340373039245604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,7168,0.39628801345825193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,8192,0.5236735979715983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,7168,0.45813547770182295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,7168,0.19671039581298827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,6144,0.3221503893534342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,6144,0.3783679962158203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,6144,0.17264639536539714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,5120,0.14728533426920573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,5120,0.26699094772338866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,5120,0.30743894577026365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,4096,0.1955498695373535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,4096,0.11933013598124187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,4096,0.2499242623647054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,3584,0.1612405300140381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,3584,0.21913599967956543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,3584,0.10704212983449299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,3072,0.13933226267496746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,3072,0.18933760325113932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,3072,0.0933888037999471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,2560,0.12287999788920086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,2560,0.15964159965515137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,2048,0.10178559621175129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,2560,0.08024746576944987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,2048,0.13257386684417724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,2048,0.06737919648488364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,1536,0.0966655969619751
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,1536,0.05341866811116537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,1536,0.08724479675292969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,1024,0.0605183998743693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,1024,0.06693546772003174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,1024,0.03942399819691976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,768,0.05225813388824463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,768,0.047172268231709794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,768,0.03273386756579082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,256,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,512,0.03546239932378133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,512,0.043178665637969973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,256,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,512,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,256,0.018668800592422485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,128,0.014847999811172486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,128,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,128,0.02791680097579956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,64,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,64,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,64,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,6144,32,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,6144,32,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,6144,32,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,16384,1.1639466603597006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,16384,0.8014848073323568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,16384,0.49380693435668943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,12288,0.6510933558146159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,65536,2.333081563313802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,65536,3.0981460571289063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,12288,0.5620053609212239
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,65536,6.626781717936199
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,12288,0.38126932779947914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,10240,0.48581972122192385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,8192,0.43437973658243817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,10240,0.5525845209757487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,10240,0.2864469210306803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,8192,0.3999402681986491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,8192,0.21971626281738282
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,7168,0.3804842631022135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,7168,0.36085761388142906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,7168,0.19886080423990887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,6144,0.3060394605000814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,6144,0.17472853660583496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,6144,0.3001002629597982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,5120,0.25535146395365393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,5120,0.25296212832132975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,4096,0.1831253369649251
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,5120,0.1626453399658203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,4096,0.20753067334493003
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,4096,0.12059306303660075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,3584,0.10574506918589274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,3584,0.1574570655822754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,3584,0.18769920667012532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,3072,0.1386496067047119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,3072,0.09362773100535074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,3072,0.16820906003316244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,2560,0.11926186879475911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,2560,0.14257493019104003
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,2560,0.08383146921793619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,2048,0.06458026568094889
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,2048,0.10495573679606121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,2048,0.11597973505655926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,1536,0.08645973205566407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,1536,0.09461759726206462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,1536,0.05140479803085327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,1024,0.05922133525212606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,1024,0.07304320335388184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,1024,0.037853864828745525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,768,0.05048319896062216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,768,0.04764586687088013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,768,0.031300266583760575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,512,0.04181333382924397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,512,0.0338261326154073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,512,0.024576000372568765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,256,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,256,0.03218773404757182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,256,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,128,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,128,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,128,0.014298666516939798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,64,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,64,0.011092266440391541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,64,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,5120,32,0.024200532833735147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,5120,32,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,5120,32,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,16384,0.38321491877237956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,16384,0.5929983774820964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,65536,1.6992938995361329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,16384,1.0391551971435546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,65536,2.831121063232422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,12288,0.4320949236551921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,12288,0.5758634567260742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,65536,6.129728190104166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,12288,0.23251627286275228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,10240,0.45496320724487305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,10240,0.19810986518859863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,10240,0.36457811991373695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,8192,0.38140586217244465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,8192,0.2652501265207926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,8192,0.15837866465250652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,7168,0.22036479314168295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,7168,0.31662079493204753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,7168,0.1379327932993571
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,6144,0.2665130615234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,6144,0.18848427136739093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,6144,0.11707733472188313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,5120,0.20848639806111655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,5120,0.15994879404703777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,5120,0.09966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,4096,0.15148372650146485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,4096,0.13086720307668048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,4096,0.08093012968699137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,3584,0.12970133622487384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,3584,0.11762346426645916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,3584,0.07512746651967367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,3072,0.11216213703155517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,3072,0.10134186744689941
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,3072,0.0661845326423645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,2560,0.09966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,2560,0.08485013643900553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,2560,0.0562175989151001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,2048,0.06952959696451823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,2048,0.08696640332539876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,2048,0.0475818673769633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,1536,0.06331733465194703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,1536,0.05368853410085043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,1024,0.02802346746126811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,1024,0.0495957334836324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,1024,0.037239468097686766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,768,0.042733867963155106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,1536,0.037546666463216145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,768,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,768,0.03413333495457967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,512,0.01819093426068624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,512,0.03594239950180054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,512,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,256,0.028703999519348145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,256,0.013685333728790283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,256,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,128,0.024268800020217897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,128,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,128,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,64,0.02211839954058329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,64,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,64,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,4096,32,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,4096,32,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,4096,32,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,16384,0.5424810409545898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,16384,0.3505493481953939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,16384,1.0232149124145509
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,65536,1.5808170318603516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,65536,2.205865478515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,12288,0.5700607935587565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,12288,0.3980287869771322
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,65536,6.005521138509115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,12288,0.2246997356414795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,10240,0.18096960385640462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,10240,0.4528469403584798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,10240,0.3315711975097656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,8192,0.35952641169230143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,8192,0.14711467425028485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,8192,0.25832106272379557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,7168,0.3097599983215332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,7168,0.20787199338277182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,7168,0.1289898633956909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,6144,0.257365338007609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,6144,0.1787903944651286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,6144,0.11383466720581055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,5120,0.20787199338277182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,5120,0.15155199368794758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,4096,0.14609066645304364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,5120,0.09861120382944742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,4096,0.12161706288655598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,4096,0.07751680215199788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,3584,0.12547413508097333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,3584,0.10802986621856689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,3584,0.06949546337127685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,3072,0.11089920202891032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,3072,0.09393386840820313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,2560,0.07963199615478515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,2560,0.09697279930114747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,3072,0.06048426628112793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,2560,0.051950931549072266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,2048,0.06423893372217813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,2048,0.07813119888305664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,2048,0.043144532044728595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,1536,0.0630783994992574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,1536,0.03471253315607707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,1536,0.04966293176015218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,1024,0.04123306671778361
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,1024,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,1024,0.04908373355865479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,768,0.042393600940704344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,512,0.03604480028152466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,768,0.03269973397254944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,768,0.022046933571497597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,512,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,256,0.028433066606521607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,512,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,256,0.013516799608866373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,256,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,128,0.02430293361345927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,128,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,64,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,64,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,128,0.011025066177050274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,64,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3584,32,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3584,32,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3584,32,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,16384,0.29945173263549807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,16384,0.4456447919209798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,16384,1.0297685623168946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,65536,1.8556917826334636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,12288,0.5440511703491211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,65536,1.5747754414876303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,12288,0.32716798782348633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,65536,5.9247584025065105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,12288,0.21046613057454427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,10240,0.4527104059855144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,10240,0.18145279884338378
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,10240,0.25371306737264
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,8192,0.34983253479003906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,8192,0.21257813771565756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,8192,0.14441812833150228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,7168,0.303547732035319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,7168,0.12875093619028727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,7168,0.22302719751993813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,6144,0.24551572799682617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,6144,0.1688576062520345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,6144,0.11352746486663819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,5120,0.14202879269917806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,5120,0.19834346771240235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,5120,0.09970346291859945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,4096,0.14759252866109213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,4096,0.13400425910949706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,4096,0.08676693439483643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,3584,0.12707839806874593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,3584,0.07956480185190837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,3584,0.11908480326334636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,3072,0.11089920202891032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,3072,0.059596800804138185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,3072,0.10581333637237549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,2560,0.09478826522827148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,2560,0.09311573505401612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,2048,0.07707306543986002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,2560,0.05120000044504801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,2048,0.060893865426381436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,2048,0.04304213523864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,1536,0.06236159801483154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,1536,0.04689813454945882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,1536,0.03423573176066081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,1024,0.048674134413401286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,1024,0.02696533401807149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,1024,0.040482131640116374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,768,0.04201813141504924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,768,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,768,0.032085333267847696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,512,0.035259731610616046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,512,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,256,0.027852799495061236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,512,0.017334399620691936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,256,0.01395840048789978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,128,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,256,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,128,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,128,0.009588266412417095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,64,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,64,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,64,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,3072,32,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,3072,32,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,3072,32,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,65536,1.733393096923828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,16384,0.3717119852701823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,16384,1.028812789916992
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,65536,1.5386623382568358
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,16384,0.2929322560628255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,12288,0.5414229075113932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,12288,0.26204053560892737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,65536,5.931003824869792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,12288,0.21486934026082358
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,10240,0.429805850982666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,10240,0.22278827031453452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,8192,0.3408554712931315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,10240,0.18612586657206218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,8192,0.18708267211914062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,8192,0.1486848036448161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,7168,0.3015679995218913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,7168,0.18527572949727375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,6144,0.2565450668334961
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,7168,0.13035519917805988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,6144,0.16411306063334147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,6144,0.11933013598124187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,5120,0.1955839951833089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,5120,0.10629013379414876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,5120,0.1470058600107829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,4096,0.09103360176086425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,4096,0.144652795791626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,4096,0.11144213676452637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,3584,0.12670293649037678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,3584,0.06816426912943521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,3584,0.10127360026041668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,3072,0.10987520217895508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,3072,0.09048746426900228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,3072,0.06942719618479411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,2560,0.09591466585795085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,2560,0.08014187018076578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,2560,0.050551466147104894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,2048,0.0756053368250529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,2048,0.08642559846242269
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,2048,0.04235946734746297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,1536,0.052940801779429114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,1536,0.06208853324254354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,1536,0.0335530678431193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,1024,0.04836159944534302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,1024,0.03969706694285075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,1024,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,768,0.041602134704589844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,768,0.031573333342870075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,768,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,512,0.034747731685638425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,512,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,512,0.022488532463709514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,256,0.013892267147699991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,256,0.027340799570083618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,128,0.009796266754468281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,128,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,256,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,64,0.021367466449737547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,64,0.007229866584142049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,128,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,64,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2560,32,0.020989867051442464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2560,32,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2560,32,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,16384,0.2847743988037109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,16384,0.16281599998474122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,16384,0.9223168055216471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,65536,1.2577109018961588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,65536,1.0226687749226888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,12288,0.4820992151896159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,65536,5.419895426432292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,12288,0.20289279619852701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,12288,0.12840960025787354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,10240,0.3666261355082194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,10240,0.1724415938059489
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,10240,0.10103466510772705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,8192,0.2732373237609863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,8192,0.14714879989624025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,7168,0.2514261404673258
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,8192,0.08082773685455322
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,7168,0.11728213628133137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,7168,0.07355519930521646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,6144,0.19763199488321942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,6144,0.10284372965494792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,6144,0.06450986862182617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,5120,0.16387413342793783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,5120,0.08625493049621583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,4096,0.11263999938964844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,5120,0.05648746490478516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,4096,0.045465600490570066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,4096,0.07089493274688721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,3584,0.09987413088480632
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,3584,0.06550186475118001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,3584,0.04096000194549561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,3072,0.08475306828816732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,3072,0.052292267481486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,3072,0.03556693394978841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,2560,0.07017813523610433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,2560,0.031027199824651082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,2560,0.04437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,2048,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,2048,0.05922133525212606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,2048,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,1536,0.04911786715189616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,1536,0.020889600118001304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,1024,0.038740265369415286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,1536,0.034508800506591795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,1024,0.024576000372568765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,1024,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,768,0.033655468622843424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,768,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,768,0.013926399747530618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,512,0.028808534145355225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,512,0.014573867122332254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,256,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,256,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,512,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,256,0.009111467003822326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,128,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,128,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,128,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,64,0.018739199638366698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,64,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,64,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,2048,32,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,2048,32,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,2048,32,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,16384,0.21715946197509767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,16384,0.8628906885782877
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,65536,0.9208490371704101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,65536,1.0014378865559896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,16384,0.13892265955607097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,12288,0.16906240781148274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,12288,0.4700160026550293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,65536,5.278105672200521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,10240,0.3627690633138021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,12288,0.10383360385894776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,10240,0.1411754608154297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,10240,0.08949759801228842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,8192,0.26436266899108884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,8192,0.07195306619008382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,8192,0.12079786459604899
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,7168,0.24285866419474283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,7168,0.06379520098368327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,7168,0.10250133673350018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,6144,0.20169386863708497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,6144,0.0886783997217814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,5120,0.15045973459879558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,6144,0.05614826679229736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,5120,0.04798933267593384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,5120,0.07645866870880128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,4096,0.10407146612803142
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,4096,0.06109866698582968
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,4096,0.037853864828745525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,3584,0.09052159786224365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,3584,0.054135465621948244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,3584,0.034474666913350424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,3072,0.079257599512736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,3072,0.0475818673769633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,3072,0.03061760067939758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,2560,0.06823253631591797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,2560,0.040004265308380124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,2560,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,2048,0.05840213298797607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,2048,0.02225493391354879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,2048,0.032631466786066696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,1536,0.04860586722691854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,1536,0.030003199974695843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,1536,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,1024,0.03853653271993001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,1024,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,1024,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,768,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,768,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,768,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,512,0.028535467386245728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,512,0.012219732999801636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,256,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,512,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,256,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,256,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,128,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,128,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,128,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,64,0.018602667252222697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,64,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,64,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1536,32,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1536,32,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1536,32,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,65536,0.6493877410888672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,16384,0.13403946558634441
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,16384,0.12782933712005615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,16384,0.876680564880371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,65536,0.8350378672281901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,12288,0.4355754534403483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,65536,5.19021250406901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,12288,0.104584534962972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,12288,0.09755307038625081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,10240,0.33488213221232094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,10240,0.09499200185139975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,10240,0.0807253360748291
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,8192,0.2741930643717448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,8192,0.0762880007425944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,8192,0.06546773513158163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,7168,0.22240746815999352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,6144,0.19725653330485027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,7168,0.07041707038879394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,7168,0.05840213298797607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,6144,0.05925546487172445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,6144,0.05079040129979452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,5120,0.05058346589406332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,5120,0.14742186864217122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,5120,0.04358506600062052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,4096,0.10004479885101318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,4096,0.04116479953130086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,3584,0.08823466300964355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,4096,0.03519146839777629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,3584,0.03700053294499715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,3584,0.031675734122594196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,3072,0.07830186684926352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,3072,0.028091732660929365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,3072,0.03512320121129354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,2560,0.06751573085784912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,2560,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,2048,0.05792426665623983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,2048,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,2560,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,2048,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,1536,0.04778666496276855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,1536,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,1536,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,1024,0.037717334429423016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,1024,0.013516799608866373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,1024,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,768,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,768,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,768,0.011844266454378765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,512,0.027784534295399982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,256,0.022562134265899658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,512,0.00986346701780955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,512,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,256,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,256,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,128,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,128,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,128,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,64,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,1024,32,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,64,0.006617600222428639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,1024,32,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,1024,32,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,65536,0.49950720469156906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,65536,0.8216917037963867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,16384,0.12823893229166666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,16384,0.10844159921010335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,16384,0.8622079849243164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,12288,0.431547737121582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,65536,5.229670206705729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,12288,0.08741119702657064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,12288,0.09679893652598062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,10240,0.33495041529337566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,10240,0.07219200134277344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,10240,0.08079360326131185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,8192,0.06157653331756592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,8192,0.269482676188151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,8192,0.06519466638565063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,7168,0.0559445341428121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,7168,0.05754559834798177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,7168,0.23169066111246744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,6144,0.19483307202657063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,6144,0.0504149317741394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,6144,0.04904959996541341
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,5120,0.13858133951822918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,5120,0.04225706656773885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,5120,0.043281066417694095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,4096,0.09943040211995444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,4096,0.03508906761805217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,4096,0.03614720106124878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,3584,0.08815999825795492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,3584,0.032525867223739624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,3072,0.07813119888305664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,3584,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,3072,0.028910932938257854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,3072,0.027818665901819868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,2560,0.06727680365244547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,2560,0.02542933424313863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,2048,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,2048,0.057275732358296715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,2560,0.024166399240493776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,2048,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,1536,0.04768426815668742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,1536,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,1536,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,1024,0.03775146802266439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,1024,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,1024,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,768,0.010035199920336406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,768,0.03259733319282532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,768,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,512,0.027613866329193115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,512,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,256,0.022323199113210044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,512,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,256,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,256,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,128,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,128,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,128,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,64,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,64,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,64,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,768,32,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,768,32,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,768,32,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,65536,0.37908480962117513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,65536,0.8202922821044922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,16384,0.07348906993865967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,16384,0.8579754511515298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,16384,0.12810239791870118
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,12288,0.06014293432235718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,12288,0.4101461410522461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,65536,5.2137303670247395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,12288,0.09659413496653238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,10240,0.34696534474690754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,10240,0.048742401599884036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,10240,0.0808618704477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,8192,0.04167679945627849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,8192,0.2686293284098307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,8192,0.06481920083363851
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,7168,0.23497279485066733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,7168,0.03758080005645752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,7168,0.05768533150355021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,6144,0.18824532826741536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,6144,0.03351893424987793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,6144,0.05017600059509277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,5120,0.14779733022054037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,5120,0.029521065950393676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,5120,0.042871467272440594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,4096,0.10038613478342692
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,4096,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,4096,0.034983468055725095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,3584,0.023108265797297158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,3584,0.08861013253529867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,3584,0.03136853377024333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,3072,0.07774933179219565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,2560,0.06720853646596273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,3072,0.02768213351567586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,3072,0.020577067136764528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,2560,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,2560,0.024200532833735147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,2048,0.057275732358296715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,2048,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,2048,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,1536,0.0475818673769633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,1536,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,1024,0.03741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,1024,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,1536,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,1024,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,768,0.03246080080668132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,768,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,768,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,512,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,512,0.009796266754468281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,512,0.00654720018307368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,256,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,256,0.02215253313382467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,256,0.007985066870848339
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,128,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,128,0.007096533477306366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,64,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,64,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,64,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,512,32,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,512,32,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,512,32,0.006516266862551372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,65536,0.27347625096639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,65536,0.8210773468017578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,16384,0.8536064147949218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,16384,0.04089173475901286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,16384,0.12759040196736654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,12288,0.4501162528991699
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,65536,5.21338857014974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,12288,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,12288,0.0966655969619751
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,10240,0.3390122731526693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,10240,0.0291157325108846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,10240,0.08089599609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,8192,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,8192,0.06478506724039713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,8192,0.26391894022623696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,7168,0.21688106854756675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,7168,0.02222080032030741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,7168,0.0575488011042277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,6144,0.18629973729451496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,6144,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,6144,0.05024426778157552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,5120,0.14465707143147785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,5120,0.042905600865681966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,5120,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,4096,0.1000106652577718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,4096,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,4096,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,3584,0.08850773175557455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,3584,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,3584,0.031300266583760575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,3072,0.077619202931722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,2560,0.06700373490651448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,3072,0.02768213351567586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,3072,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,2560,0.012458667159080505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,2560,0.024132267634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,2048,0.057207465171813965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,2048,0.010752000411351522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,1536,0.0474453330039978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,2048,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,1536,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,1536,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,1024,0.037307735284169516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,1024,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,1024,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,768,0.032290132840474446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,768,0.011570133765538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,768,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,512,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,512,0.009658666451772054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,512,0.0051146666208903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,256,0.02208426594734192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,256,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,128,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,256,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,128,0.0070261334379514056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,128,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,64,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,64,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,256,32,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,64,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,256,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,256,32,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,65536,0.26808319091796873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,65536,0.8248597462972006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,16384,0.02624853253364563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,16384,0.8573269526163736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,16384,0.1279317299524943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,12288,0.4233215967814128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,65536,5.21519775390625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,12288,0.021538132429122926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,12288,0.09703466892242432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,10240,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,10240,0.3238911946614583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,10240,0.08096426328023275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,8192,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,8192,0.2637824058532715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,8192,0.06488746802012126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,7168,0.2285226662953695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,7168,0.016310399770736693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,7168,0.05768533150355021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,6144,0.18192532857259114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,6144,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,6144,0.050312534968058265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,5120,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,5120,0.13957014083862304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,5120,0.043006932735443114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,4096,0.09939093589782715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,4096,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,4096,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,3584,0.08830293019612631
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,3584,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,3584,0.03150506615638733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,3072,0.07795733610788981
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,3072,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,3072,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,2560,0.06707200209299723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,2560,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,2560,0.024132267634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,2048,0.057139201958974206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,2048,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,2048,0.02044266661008199
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,1536,0.04727466503779094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,1536,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,1536,0.016721065839131674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,1024,0.03741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,1024,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,1024,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,768,0.032358400026957196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,768,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,512,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,768,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,512,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,512,0.009656533598899841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,256,0.0035135999321937563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,256,0.022016000747680665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,256,0.00791786660750707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,128,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,128,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,128,0.0070271998643875126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,64,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,64,0.002661333233118057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,64,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,128,32,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,128,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,128,32,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,16384,0.030139732360839843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,65536,0.20804266929626464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,65536,0.8246613184611002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,16384,0.12728319962819418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,16384,0.8518997192382812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,12288,0.44735040664672854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,12288,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,65536,5.201578776041666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,12288,0.09629013538360595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,10240,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,10240,0.32235520680745444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,10240,0.08038400014241537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,8192,0.25398079554239905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,8192,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,8192,0.06464639902114869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,7168,0.22586026191711425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,7168,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,7168,0.057070934772491456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,6144,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,6144,0.04990080197652181
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,6144,0.1828181266784668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,5120,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,5120,0.13687465985616049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,4096,0.09936213493347168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,5120,0.042905600865681966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,4096,0.010682666301727295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,4096,0.034815998872121175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,3584,0.08785920143127442
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,3584,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,3584,0.031197865804036457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,3072,0.07799039681752523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,3072,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,3072,0.027374933163324993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,2560,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,2560,0.06690133412679036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,2560,0.024132267634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,2048,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,2048,0.057070934772491456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,2048,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,1536,0.04740906556447347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,1536,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,1536,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,1024,0.037273601690928145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,1024,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,1024,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,768,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,768,0.03246080080668132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,768,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,512,0.027135999997456868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,512,0.004025600105524063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,256,0.0033439998825391137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,512,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,256,0.021980800231297812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,256,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,128,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,128,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,128,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,64,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,64,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,64,32,0.01746986707051595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,64,32,0.0022528000175952912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,64,32,0.006450133522351582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,65536,0.20234239896138512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,16384,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,65536,0.8261290868123373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,16384,0.12748800118764242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,16384,0.8570880254109701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,12288,0.4459168116251628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,12288,0.014472533265749613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,65536,5.215983072916666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,12288,0.09659732977549235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,10240,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,10240,0.08014079729715982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,10240,0.32095572153727214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,8192,0.26231466929117836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,8192,0.06458026568094889
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,8192,0.010990933577219645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,7168,0.22644054094950355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,7168,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,7168,0.05734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,6144,0.1845589319864909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,6144,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,6144,0.04993706544240316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,5120,0.14243839581807455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,5120,0.008666666348775227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,5120,0.04256426493326823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,4096,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,4096,0.09905493259429932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,4096,0.034406399726867674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,3584,0.031129600604375203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,3584,0.0878922700881958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,3584,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,3072,0.07778986295064291
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,3072,0.0067562664548556015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,3072,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,2560,0.06690133412679036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,2560,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,2560,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,2048,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,2048,0.057036801179250085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,2048,0.020102399587631225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,1536,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,1536,0.047172268231709794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,1024,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,1024,0.0373418649037679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,1536,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,1024,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,768,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,768,0.03222186764081319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,768,0.011332266529401143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,512,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,512,0.02723840077718099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,512,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,256,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,256,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,256,0.0076789334416389465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,128,0.01925119956334432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,128,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,128,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,64,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,64,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,64,0.006377600133419037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1024,32,32,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1024,32,32,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1024,32,32,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,12288,3.5297620137532553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,16384,4.142796834309896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,16384,4.439517720540365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,10240,2.5466880798339844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,12288,3.0883839925130205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,12288,6.341051737467447
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,10240,5.291486104329428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,16384,8.583952840169271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,10240,2.648883310953776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,8192,1.930069351196289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,8192,2.077457046508789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,7168,1.709841028849284
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,8192,4.090197245279947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,7168,1.8852181752522785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,6144,1.488588841756185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,7168,3.3012736002604166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,6144,1.5920469919840494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,6144,2.929425048828125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,5120,1.2412245432535807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,5120,1.334067153930664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,5120,2.4128468831380205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,4096,1.0518869400024413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,4096,0.9840298970540365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,4096,1.8482154846191405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,3584,0.8770559946695963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,3584,1.6013994852701825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,3584,0.9730730692545573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,3072,0.7716864267985026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,3072,0.7726079940795898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,3072,1.3777578989664714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,2560,0.6527658462524414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,2560,1.1721728006998697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,2560,0.7275519688924154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,2048,0.5496490478515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,2048,0.5867861429850261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,2048,0.9096192042032877
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,1536,0.46404266357421875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,1536,0.42472108205159503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,1536,0.6954666773478191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,1024,0.31682027180989586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,1024,0.46994771957397463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,1024,0.35560105641682943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,768,0.26289493242899575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,768,0.36546452840169275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,768,0.3058687845865885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,512,0.22701652844746908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,512,0.33795413970947263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,512,0.23217493693033853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,256,0.18821120262145996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,256,0.2193749268849691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,256,0.19387733141581218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,128,0.18046293258666993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,128,0.20196693738301597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,64,0.1752405325571696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,128,0.17339733441670735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,64,0.1607338587443034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,64,0.175820795694987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,65536,32,0.17674239476521808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,65536,32,0.16032427151997883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,65536,32,0.17090026537577313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,16384,1.0290517171223958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,16384,1.875387700398763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,16384,1.2601685841878256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,12288,0.9123839696248373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,65536,5.068254089355468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,12288,1.3799381256103516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,65536,8.097723897298177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,12288,0.8581461588541666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,65536,7.510664367675782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,10240,0.7514794667561848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,10240,1.1256490071614584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,10240,0.6603776295979817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,8192,0.5909845352172851
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,8192,0.9044991811116537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,8192,0.532205867767334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,7168,0.5386922836303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,7168,0.4909397443135579
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,6144,0.45824000040690105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,7168,0.7964330673217773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,6144,0.6662485122680664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,5120,0.38847147623697914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,6144,0.4289877255757649
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,5120,0.5651456197102864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,5120,0.37352107365926107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,4096,0.3187029202779134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,4096,0.45496320724487305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,4096,0.2785279909769694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,3584,0.28282880783081055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,3584,0.2150400002797445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,3072,0.25221120516459145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,3072,0.3451221466064453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,3584,0.38492161432902017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,3072,0.18701653480529784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,2560,0.21271893183390297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,2560,0.289518928527832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,2560,0.16721919377644856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,2048,0.1657514731089274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,2048,0.13482666015625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,2048,0.20520960489908852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,1536,0.15977813402811686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,1536,0.10615466435750325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,1024,0.09567573070526122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,1024,0.1136629343032837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,1536,0.12731733322143554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,1024,0.07686826388041178
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,768,0.07536640167236328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,768,0.08929279645284018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,768,0.06191786527633667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,512,0.061713067690531406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,512,0.06338560183842977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,512,0.046728531519571945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,256,0.04648960034052531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,256,0.032631466786066696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,256,0.039833601315816244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,128,0.03839999834696452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,128,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,128,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,64,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,64,0.03495253324508667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,16384,32,0.0343722661336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,16384,32,0.014740266402562461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,32,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,16384,64,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,16384,1.105612818400065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,16384,0.8188245137532553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,65536,4.084018198649089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,12288,0.7881727854410807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,16384,1.6017749786376954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,65536,6.095598856608072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,12288,1.1782527923583985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,65536,6.839910380045573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,12288,0.6892202377319336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,10240,0.6455637613932292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,10240,0.5622101465861002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,10240,0.9704789479573568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,8192,0.5118293444315593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,8192,0.7935317357381184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,7168,0.4547925313313802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,8192,0.4458154678344727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,7168,0.7625727971394857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,7168,0.3891200065612793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,6144,0.3900416056315104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,6144,0.36048212051391604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,6144,0.661026128133138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,5120,0.3338239987691244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,5120,0.27883520126342776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,4096,0.4405589421590169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,4096,0.27876691818237304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,3584,0.23971840540568032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,4096,0.21319680213928222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,5120,0.5436757405598958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,3584,0.3845461209615072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,3584,0.1964373270670573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,3072,0.19176106452941893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,3072,0.15899306933085125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,3072,0.3249151865641276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,2560,0.15793493588765461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,2560,0.13608959515889485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,2048,0.13543465932210286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,2048,0.2168480078379313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,2560,0.26596693992614745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,2048,0.10939733187357585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,1536,0.11035306453704834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,1536,0.08714240392049154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,1536,0.1785173257191976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,1024,0.0861525297164917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,1024,0.107315198580424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,1024,0.06328320105870565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,768,0.06509226560592651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,768,0.08809599876403809
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,768,0.05208746592203776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,512,0.052940801779429114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,512,0.03898026545842488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,512,0.06526293357213339
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,256,0.04031146764755249
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,256,0.03099199930826823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,128,0.034030934174855546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,128,0.019933867454528808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,128,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,256,0.027340799570083618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,64,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,64,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,64,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,12288,32,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,12288,32,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,12288,32,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,16384,0.9591466903686523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,16384,1.1466410319010416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,65536,3.2466603597005212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,16384,0.6973781585693359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,12288,0.6626303990681965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,12288,0.8565760294596354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,65536,4.922743225097657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,65536,6.127991231282552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,12288,0.5503615697224935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,10240,0.5514922459920247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,8192,0.4317184130350749
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,10240,0.45786453882853195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,8192,0.5727914810180664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,8192,0.38615039189656575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,10240,0.6892490386962891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,7168,0.3916799863179525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,7168,0.3220821380615234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,7168,0.49541120529174804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,6144,0.24900266329447426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,6144,0.33713493347167967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,6144,0.42315092086791994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,5120,0.2897578557332357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,5120,0.2007381280263265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,5120,0.34843308130900064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,4096,0.2698549270629883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,4096,0.23825066884358725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,4096,0.16083626747131347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,3584,0.20486826896667482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,3584,0.22824959754943847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,3584,0.14755840301513673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,3072,0.1567402680714925
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,3072,0.1290239969889323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,2560,0.13810027440388997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,2560,0.16855039596557617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,3072,0.19640320142110188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,2560,0.10816853046417237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,2048,0.1176917314529419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,2048,0.1352021376291911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,2048,0.09069226582845052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,1536,0.09778772989908854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,1536,0.105403733253479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,1024,0.07396693229675293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,1024,0.06976853211720785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,1024,0.051677866776784265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,768,0.05785599946975708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,768,0.05758293469746908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,1536,0.07120213508605958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,512,0.04771840174992879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,768,0.04188160101572673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,512,0.031880533695220946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,512,0.041465600331624344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,256,0.02624853253364563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,256,0.037034666538238524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,256,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,128,0.03132800062497457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,128,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,128,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,64,0.028501333793004353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,64,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,64,0.014469333489735923
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,10240,32,0.027818665901819868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,10240,32,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,10240,32,0.013891200224558512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,16384,0.9271978378295899
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,16384,0.5111125310262044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,16384,0.8124074935913086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,12288,0.5573973337809245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,65536,2.445107269287109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,65536,3.8519808451334634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,65536,5.541171264648438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,12288,0.3819178581237793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,12288,0.8129877090454102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,10240,0.6402048110961914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,10240,0.348091729482015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,10240,0.46155093510945633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,8192,0.38174721399943035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,8192,0.5022037188212077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,8192,0.2434389273325602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,7168,0.32931626637776695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,7168,0.20111360549926757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,6144,0.2780159950256348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,6144,0.3959125200907389
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,6144,0.1774250666300456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,7168,0.4578986803690593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,5120,0.2411861260732015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,5120,0.3245738665262858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,5120,0.14482773145039876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,4096,0.17134933471679686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,4096,0.11693759759267171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,4096,0.2472618738810221
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,3584,0.15008427302042643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,3584,0.10560853481292724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,3072,0.13329066435496012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,3584,0.21780479749043785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,3072,0.18937172889709472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,3072,0.09267199834187825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,2560,0.16097280184427898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,2560,0.11741866270701092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,2560,0.07959893544514975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,2048,0.1008639971415202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,2048,0.13209599653879803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,1536,0.07488853136698405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,2048,0.06655999819437662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,1536,0.09560746351877848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,1536,0.05311146577199301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,1024,0.05778773228327433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,1024,0.06860586802164713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,1024,0.03898026545842488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,768,0.050312534968058265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,768,0.0475818673769633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,512,0.04198400179545085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,512,0.03604480028152466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,512,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,768,0.03211946686108907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,256,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,256,0.021367466449737547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,256,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,128,0.027955200274785357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,128,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,128,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,64,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,32,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,64,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,8192,32,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,8192,64,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,8192,32,0.012219732999801636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,16384,0.5258559862772624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,16384,0.7996074676513671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,16384,0.7960234959920247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,65536,2.0908031463623047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,12288,0.5907125473022461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,65536,3.6146517435709633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,12288,0.5315893491109211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,65536,5.440238952636719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,12288,0.39983787536621096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,10240,0.44390398661295577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,10240,0.3236522674560547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,10240,0.5023402531941732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,8192,0.3611647923787435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,8192,0.22016000747680664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,7168,0.3263829231262207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,8192,0.40625492731730145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,7168,0.41130666732788085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,7168,0.19131733576456705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,6144,0.2809856096903483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,6144,0.16940372784932453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,6144,0.3407178560892741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,5120,0.22927360534667968
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,5120,0.1542143980662028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,4096,0.16520533561706544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,4096,0.1160533348719279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,5120,0.2675711949666341
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,4096,0.21551787058512367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,3584,0.1472511927286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,3584,0.18728639284769694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,3584,0.10369706948598226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,3072,0.131169064839681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,3072,0.09147733052571615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,3072,0.1643839995066325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,2560,0.14079999923706055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,2560,0.07864320278167725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,2560,0.11557013193766277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,2048,0.10168320337931316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,2048,0.13841066360473633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,2048,0.06526293357213339
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,1536,0.08205653031667073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,1536,0.09318400224049886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,1536,0.05092693169911703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,1024,0.056695465246836343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,1024,0.037817601362864176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,768,0.04805546601613363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,1024,0.07106560071309408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,768,0.048981332778930665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,768,0.031402667363484696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,512,0.04119893312454224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,512,0.034474666913350424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,512,0.02430293361345927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,256,0.031914667288462324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,256,0.020814933379491172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,128,0.013993600010871887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,128,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,128,0.027374933163324993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,64,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,256,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,64,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,64,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,7168,32,0.02450773318608602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,7168,32,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,7168,32,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,16384,0.7955455780029297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,65536,2.1070507049560545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,16384,0.8144896189371744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,16384,0.5144234657287597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,12288,0.5368149439493816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,12288,0.6177109400431315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,65536,3.1248715718587237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,65536,5.422967529296875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,12288,0.3863893191019694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,10240,0.45875199635823566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,10240,0.5071872075398762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,10240,0.29040638605753577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,8192,0.36720641454060876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,8192,0.23691946665445962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,8192,0.42383359273274734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,7168,0.31873706181844075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,7168,0.3850581487019857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,7168,0.20210347175598145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,6144,0.2736469268798828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,6144,0.32013654708862305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,6144,0.18718719482421875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,5120,0.21927253405253092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,4096,0.1624725341796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,4096,0.21975040435791016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,5120,0.16438612937927247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,4096,0.12062719662984211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,5120,0.27306667963663733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,3584,0.14656853675842285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,3584,0.10669759909311931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,3584,0.1998165289560954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,3072,0.1311402638753255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,3072,0.17298773129781086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,2560,0.11537066300710042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,2560,0.0843775987625122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,3072,0.09724586804707845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,2048,0.10188799699147541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,2048,0.11318613688151043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,2560,0.1539413293202718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,2048,0.07406933307647705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,1536,0.07246507008870443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,1536,0.09308160146077474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,1536,0.05024426778157552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,1024,0.05570240020751953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,1024,0.07383039792378744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,1024,0.037102933724721274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,768,0.04792319933573405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,768,0.04795733292897542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,768,0.03037866751352946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,512,0.03973120053609212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,512,0.03392426570256551
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,256,0.03136853377024333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,256,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,512,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,128,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,128,0.013516799608866373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,128,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,256,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,64,0.024472532669703166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,64,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,64,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,6144,32,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,6144,32,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,6144,32,0.011707733074824016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,16384,0.636791483561198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,16384,0.5769898732503255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,65536,2.393292744954427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,16384,0.34949121475219724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,65536,1.511731211344401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,12288,0.44936532974243165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,12288,0.42284374237060546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,65536,4.814266459147135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,12288,0.2198186715443929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,10240,0.3798314730326334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,10240,0.3637248039245605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,8192,0.27149651845296224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,10240,0.18810879389444987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,8192,0.14854826927185058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,7168,0.26187094052632653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,8192,0.3090773264567057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,7168,0.21480107307434082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,7168,0.13311999638875324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,6144,0.21845226287841796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,6144,0.11407360235850017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,5120,0.17046186129252117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,6144,0.1845589319864909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,5120,0.15510187149047852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,4096,0.131549866994222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,4096,0.12492799758911133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,5120,0.09594559669494629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,4096,0.07898346583048502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,3584,0.11690666675567626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,3584,0.11161599953969319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,3072,0.10635840098063151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,3584,0.07061973412831625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,3072,0.09646080334981283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,3072,0.06167893409729004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,2560,0.08338773250579834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,2560,0.09291093349456787
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,2560,0.05403306484222412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,2048,0.07383039792378744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,2048,0.06734506289164224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,1536,0.059835731983184814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,2048,0.04423679908116658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,1536,0.05102933247884115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,1536,0.03604480028152466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,1024,0.03819520076115926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,1024,0.026282666126887004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,768,0.040994131565093996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,1024,0.04696746667226155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,768,0.02218666672706604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,512,0.034713598092397054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,768,0.03345066706339518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,512,0.024405332406361897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,512,0.017338667313257852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,256,0.02775040070215861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,256,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,128,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,256,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,128,0.00979306697845459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,128,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,64,0.022012799978256226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,64,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,64,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,5120,32,0.021572266022364298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,5120,32,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,5120,32,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,16384,0.6160725275675456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,65536,1.9511637369791668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,16384,0.4646570523579915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,16384,0.327782408396403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,65536,1.4839808146158853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,12288,0.4373141288757324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,65536,4.7336786905924475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,12288,0.20954453150431313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,10240,0.282862917582194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,12288,0.33402878443400064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,10240,0.17489919662475586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,10240,0.3893930753072103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,8192,0.21497173309326173
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,8192,0.2864469210306803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,8192,0.14609066645304364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,7168,0.24770560264587402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,7168,0.12622506618499757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,6144,0.2023082733154297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,6144,0.1698751926422119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,6144,0.11045546531677246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,7168,0.21992106437683107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,5120,0.1421994686126709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,5120,0.16294827461242675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,5120,0.09874773025512695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,4096,0.1295018672943115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,4096,0.11502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,4096,0.086899201075236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,3584,0.11741759777069091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,3584,0.07895039717356364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,3584,0.10141013463338215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,3072,0.10550506909688313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,3072,0.08830080032348633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,3072,0.05928959846496582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,2560,0.09171626567840577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,2560,0.07540053526560465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,2560,0.05085866848627726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,2048,0.0719871997833252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,2048,0.04225706656773885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,1536,0.05877759853998819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,2048,0.060279464721679686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,1536,0.046830932299296066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,1536,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,1024,0.04641493161519368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,1024,0.040243200461069745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,1024,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,768,0.032051199674606325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,768,0.04041386842727661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,512,0.0343722661336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,512,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,512,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,256,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,768,0.021297067403793335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,256,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,256,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,128,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,128,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,64,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,128,0.0234442671140035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,64,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,4096,32,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,64,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,4096,32,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,4096,32,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,16384,0.6176085154215495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,65536,1.4567423502604167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,16384,0.40912214914957684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,65536,1.8416277567545574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,16384,0.2871295928955078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,12288,0.44386987686157225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,65536,4.698111979166667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,12288,0.3007146517435709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,12288,0.21387947400410973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,10240,0.22760106722513834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,10240,0.3680938720703125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,10240,0.18756267229715984
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,8192,0.18609493573506672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,8192,0.15499946276346843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,7168,0.24296107292175292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,8192,0.27241385777791344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,7168,0.18688000043233235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,7168,0.13216426372528076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,6144,0.19906560579935711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,6144,0.11919360160827637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,6144,0.16411306063334147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,5120,0.15909546216328938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,5120,0.14837759335835773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,4096,0.12916053136189778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,5120,0.09891839822133383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,4096,0.11625920136769612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,4096,0.09052159786224365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,3584,0.11670186519622802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,3584,0.07854080200195312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,3072,0.10513066450754802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,3584,0.10492586294809977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,3072,0.09226240317026774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,2560,0.09209173520406087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,2560,0.08133973280588785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,3072,0.06208639939626058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,2560,0.0504149317741394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,2048,0.07191893259684244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,2048,0.08304639657338461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,1536,0.05874346494674683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,2048,0.04167679945627849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,1536,0.053452801704406736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,1536,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,1024,0.039765334129333495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,1024,0.04614826838175456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,1024,0.024779733022054037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,768,0.040209066867828366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,768,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,768,0.03163626591364543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,512,0.034195200602213545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,512,0.022528000672658286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,512,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,256,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,256,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,128,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,256,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,128,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,64,0.021640533208847047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,128,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,64,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,64,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3584,32,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3584,32,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3584,32,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,16384,0.6248106638590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,65536,1.4266026814778647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,16384,0.3693567911783854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,65536,1.655227788289388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,16384,0.27767467498779297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,12288,0.4459520022074382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,65536,4.723063659667969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,12288,0.25497600237528484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,12288,0.2158250649770101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,10240,0.36969814300537107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,10240,0.19196586608886718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,8192,0.2863786697387695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,10240,0.2226858615875244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,8192,0.17838080724080402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,8192,0.15602347056070964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,7168,0.18496853510538738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,7168,0.1377621332804362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,6144,0.19548160235087078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,7168,0.24586240450541177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,6144,0.16319146156311035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,5120,0.1573205312093099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,5120,0.14540799458821613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,6144,0.11748692989349366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,5120,0.10769066810607911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,4096,0.12847786744435627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,3584,0.11721386909484863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,4096,0.11011412938435872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,3584,0.09943040211995444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,4096,0.0886783997217814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,3584,0.0831829309463501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,3072,0.10567680199940999
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,3072,0.093013334274292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,3072,0.05918720165888468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,2560,0.09127253691355387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,2560,0.07956480185190837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,2560,0.048537600040435794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,2048,0.06877866586049398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,2048,0.0711680014928182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,2048,0.04119893312454224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,1536,0.05802666743596395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,1536,0.05253119866053263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,1536,0.03290453354517619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,1024,0.045636268456776936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,1024,0.03932160139083862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,1024,0.024950400988260905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,768,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,768,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,768,0.03945813179016113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,512,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,512,0.022386133670806885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,512,0.016381866733233132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,256,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,256,0.013585066795349121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,256,0.012663466731707254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,128,0.022835199038187662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,128,0.010170666376749675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,64,0.021640533208847047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,64,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,64,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,128,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,3072,32,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,3072,32,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,3072,32,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,16384,0.5079039891560873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,65536,1.2453887939453125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,65536,0.9578496297200522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,16384,0.1592309315999349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,16384,0.28596906661987304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,12288,0.33996693293253577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,65536,4.215735371907552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,12288,0.20018879572550455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,12288,0.12004693349202473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,10240,0.28955198923746744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,10240,0.16950292587280275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,10240,0.09983999729156494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,8192,0.14199466705322267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,8192,0.22298879623413087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,8192,0.07990612983703613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,7168,0.11171840031941731
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,6144,0.14981120427449543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,7168,0.07178133328755697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,7168,0.1839786688486735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,6144,0.09833813508351644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,6144,0.06359039942423503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,5120,0.08427519798278808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,5120,0.116974933942159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,5120,0.053521064917246494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,4096,0.09710933367411295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,4096,0.04437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,3584,0.08372906843821207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,4096,0.06983679930369059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,3584,0.05915306806564331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,3584,0.039250131448109946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,3072,0.051712000370025636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,3072,0.0723626693089803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,3072,0.03505493402481079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,2560,0.06304426590601603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,2560,0.043281066417694095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,2560,0.029798400402069092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,2048,0.054340267181396486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,2048,0.03597653309504191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,2048,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,1536,0.03160746693611145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,1536,0.01986560026804606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,1536,0.0451583981513977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,1024,0.03635199864705403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,1024,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,1024,0.02405973275502523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,768,0.031982932488123575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,768,0.019214934110641478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,768,0.013209600249926248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,512,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,512,0.013926399747530618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,256,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,512,0.010956799983978272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,256,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,256,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,128,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,128,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,128,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,64,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,64,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2560,32,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,64,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2560,32,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2560,32,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,65536,0.9737557093302408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,65536,0.8021674474080405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,16384,0.46994771957397463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,16384,0.21821440060933434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,65536,4.0953514099121096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,12288,0.1669119993845622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,12288,0.3241642634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,16384,0.13322239716847736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,12288,0.1033514658610026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,10240,0.1450655937194824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,10240,0.26801493962605794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,10240,0.08686933517456055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,8192,0.12124160130818684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,8192,0.21872426668802897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,8192,0.07072426478068033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,7168,0.18616320292154948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,7168,0.10280959606170655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,7168,0.06195199886957804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,6144,0.14441812833150228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,6144,0.08905386924743652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,6144,0.054749866326649986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,5120,0.10898773670196533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,5120,0.07615146636962891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,4096,0.08949759801228842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,4096,0.060893865426381436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,5120,0.046353065967559816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,4096,0.03741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,3584,0.07997439702351888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,3584,0.05410133202870687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,3584,0.03420159816741943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,3072,0.07048532962799073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,3072,0.04706986745198567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,3072,0.029832533995310467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,2560,0.03993599812189738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,2560,0.06202026605606079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,2560,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,2048,0.05386240084966024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,2048,0.032767999172210696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,1536,0.04478293259938558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,1536,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,2048,0.02208426594734192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,1536,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,1024,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,1024,0.014233600099881491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,768,0.0315391997496287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,1024,0.03635199864705403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,768,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,768,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,512,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,512,0.027135999997456868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,512,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,256,0.02249280015627543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,256,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,256,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,128,0.020138667027155558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,128,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,128,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,64,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,64,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,64,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,2048,32,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,2048,32,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,2048,32,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,65536,0.7192576090494792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,16384,0.16701439221700032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,16384,0.4900469462076823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,16384,0.12369919617970784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,65536,0.7563605626424154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,65536,4.003327941894531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,12288,0.333789857228597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,12288,0.1252351999282837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,12288,0.09533440272013347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,10240,0.2656938711802165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,10240,0.10796373685201008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,10240,0.07874560356140137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,8192,0.2102954705556234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,8192,0.08887893358866374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,8192,0.06423893372217813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,7168,0.09980586369832357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,7168,0.057207465171813965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,7168,0.1817599932352702
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,6144,0.1304917335510254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,6144,0.08816426595052083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,6144,0.04969813426335652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,5120,0.10673493544260662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,5120,0.07212373415629068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,5120,0.042461868127187094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,4096,0.08826879660288492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,4096,0.034542934099833174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,3584,0.07881386280059814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,4096,0.059494400024414064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,3584,0.03099306623140971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,3072,0.07075839837392171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,3584,0.05303893486658732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,3072,0.04703573385874431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,3072,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,2560,0.06164480050404867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,2560,0.043110398451487224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,2560,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,2048,0.053179732958475744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,2048,0.03556693394978841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,2048,0.020407466093699138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,1536,0.04457813501358032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,1536,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,1536,0.026692267258961993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,1024,0.03577173153559367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,768,0.031573333342870075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,1024,0.013514666755994161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,1024,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,768,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,768,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,512,0.011332266529401143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,512,0.027101866404215497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,512,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,256,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,256,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,128,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,256,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,128,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,128,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,64,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,64,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,64,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1536,32,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1536,32,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1536,32,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,65536,0.5009407997131348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,65536,0.7118506749471029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,16384,0.4917898813883464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,16384,0.11345919768015544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,16384,0.12318720022837322
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,65536,3.9764991760253907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,12288,0.32552960713704426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,12288,0.08447999954223633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,12288,0.09413973490397134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,10240,0.07386133670806885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,10240,0.26552319526672363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,10240,0.07871146996816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,8192,0.20517546335856118
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,8192,0.062395731608072914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,8192,0.06341973145802816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,7168,0.1768448034922282
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,7168,0.05618346532185873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,7168,0.05485226710637411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,6144,0.04983253479003906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,6144,0.048878931999206544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,6144,0.13035199642181397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,5120,0.10547200043996174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,5120,0.04188160101572673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,4096,0.08782506783803304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,5120,0.04072106679280599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,4096,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,4096,0.034474666913350424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,3584,0.07857493559519449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,3584,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,3584,0.031436800956726074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,3072,0.06962560017903646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,3072,0.027272532383600872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,2560,0.061166934172312414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,3072,0.027716267108917236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,2560,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,2560,0.023552000522613525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,2048,0.05287253459294637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,2048,0.019282132387161255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,2048,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,1536,0.04382719993591309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,1536,0.01665279964605967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,1024,0.03556693394978841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,1536,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,1024,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,1024,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,768,0.010035199920336406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,768,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,512,0.02648746569951375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,768,0.03092479904492696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,512,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,512,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,256,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,256,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,256,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,128,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,128,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,64,0.018739199638366698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,128,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,64,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,64,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,1024,32,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,1024,32,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,1024,32,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,65536,0.3785045305887858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,16384,0.4569087982177734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,16384,0.0825002670288086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,16384,0.12339200178782146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,65536,0.7136255900065105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,65536,3.961719512939453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,12288,0.32754348119099935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,12288,0.06638933420181274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,12288,0.09355946381886801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,10240,0.2640896002451579
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,10240,0.05843626658121744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,10240,0.07860906918843588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,8192,0.04604586760203044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,8192,0.19848532676696778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,8192,0.0632149338722229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,7168,0.1752405325571696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,7168,0.04341760079065959
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,7168,0.056081068515777585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,6144,0.12707733313242595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,6144,0.03829760154088338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,6144,0.048913065592447916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,5120,0.0338261326154073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,5120,0.04177920023600261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,5120,0.10598399639129638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,4096,0.0877568006515503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,4096,0.028910932938257854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,4096,0.03420159816741943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,3584,0.025873066981633504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,3584,0.07829866409301758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,3584,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,3072,0.06939307053883871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,3072,0.027170133590698243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,3072,0.022835199038187662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,2560,0.06106453339258829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,2560,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,2560,0.02347946763038635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,2048,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,2048,0.019933867454528808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,2048,0.052292267481486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,1536,0.044100264708201095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,1536,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,1024,0.03560106754302979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,1536,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,1024,0.010683733224868774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,1024,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,768,0.03089066743850708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,768,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,768,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,512,0.026692267258961993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,512,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,512,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,256,0.021367466449737547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,256,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,256,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,128,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,128,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,128,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,64,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,64,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,64,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,768,32,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,768,32,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,768,32,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,65536,0.28262399037679037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,65536,0.7261184056599934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,16384,0.4803584098815918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,16384,0.12311893304189045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,12288,0.32467625935872396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,16384,0.10236586729685467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,65536,3.973290761311849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,12288,0.07987199624379476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,12288,0.09332053661346436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,10240,0.2622773329416911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,10240,0.04143786827723185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,10240,0.07833600044250488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,8192,0.20421973864237467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,7168,0.17397759755452474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,8192,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,8192,0.06311253309249878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,7168,0.030719999472300214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,7168,0.05618026653925577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,6144,0.12796586354573566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,6144,0.027340799570083618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,6144,0.048878931999206544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,5120,0.10536959966023762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,5120,0.02405866583188375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,4096,0.0877568006515503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,5120,0.04160853226979573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,4096,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,4096,0.0339626669883728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,3584,0.07847253481547037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,3584,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,3584,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,3072,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,3072,0.06942719618479411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,2560,0.0606549342473348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,3072,0.027272532383600872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,2560,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,2560,0.023411200443903605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,2048,0.052292267481486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,2048,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,2048,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,1536,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,1536,0.04369066556294759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,1024,0.034986666838328045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,1536,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,1024,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,1024,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,768,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,768,0.007234133283297221
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,768,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,512,0.026520532369613648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,512,0.005866666634877523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,256,0.021572266022364298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,512,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,256,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,128,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,128,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,64,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,128,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,64,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,64,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,512,32,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,512,32,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,512,32,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,65536,0.7403178532918294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,65536,0.22002347310384116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,16384,0.4760181427001953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,16384,0.03351893424987793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,16384,0.12342613538106281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,65536,3.9830474853515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,12288,0.026862933238347368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,12288,0.32221546173095705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,12288,0.09331733385721842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,10240,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,10240,0.07860906918843588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,8192,0.20329813957214354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,10240,0.24582826296488441
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,8192,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,8192,0.06338560183842977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,7168,0.1650346597035726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,7168,0.05642240047454834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,7168,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,6144,0.12929706573486327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,6144,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,6144,0.048913065592447916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,5120,0.10509653091430664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,5120,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,5120,0.04167679945627849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,4096,0.08779093424479166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,4096,0.013582932949066161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,4096,0.03423573176066081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,3584,0.078438401222229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,3584,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,3584,0.030481066306432086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,3072,0.06970026493072509
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,3072,0.027170133590698243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,2560,0.06075733502705892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,2560,0.010342400272687275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,3072,0.011945600310961407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,2560,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,2048,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,2048,0.01993173360824585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,2048,0.05225813388824463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,1536,0.04358826478322347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,1536,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,1536,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,1024,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,1024,0.03498133420944214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,1024,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,768,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,768,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,768,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,512,0.026316799720128375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,512,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,512,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,256,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,256,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,256,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,128,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,128,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,128,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,64,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,64,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,64,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,256,32,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,256,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,256,32,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,65536,0.20493440628051757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,65536,0.7502506891886394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,16384,0.48435198465983076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,16384,0.02208426594734192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,16384,0.1230847994486491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,65536,3.976840464274088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,12288,0.3221845308939616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,12288,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,12288,0.09332053661346436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,10240,0.25995945930480957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,10240,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,10240,0.07881386280059814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,8192,0.19776852925618488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,8192,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,7168,0.013755733768145243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,8192,0.06328320105870565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,7168,0.16134826342264813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,7168,0.056149331728617344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,6144,0.12875093619028727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,6144,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,6144,0.048913065592447916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,5120,0.1052671988805135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,4096,0.08755199909210205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,5120,0.04181333382924397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,4096,0.009796266754468281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,5120,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,4096,0.03399680058161418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,3584,0.07874560356140137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,3584,0.009177600344022114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,3584,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,3072,0.06925653616587321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,3072,0.026961066325505573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,2560,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,2560,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,3072,0.008699733018875121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,2560,0.06082559823989868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,2048,0.05215573310852051
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,2048,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,2048,0.019656533002853395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,1536,0.043246932824452716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,1536,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,1536,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,1024,0.034986666838328045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,1024,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,768,0.030719999472300214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,1024,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,768,0.005015466610590617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,768,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,512,0.026726400852203368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,512,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,256,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,512,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,256,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,256,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,128,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,128,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,128,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,64,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,64,0.0024874667326609296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,128,32,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,64,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,128,32,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,128,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,65536,0.15762772560119628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,65536,0.752125867207845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,16384,0.44721492131551105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,16384,0.024064000447591147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,16384,0.12253866990407307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,65536,3.9825365702311197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,12288,0.32160425186157227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,12288,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,12288,0.0928767999013265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,10240,0.24773866335550943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,10240,0.01665173371632894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,10240,0.07898453076680502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,8192,0.19268266359965008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,8192,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,8192,0.06283946832021078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,7168,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,7168,0.168448003133138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,6144,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,7168,0.055910400549570714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,6144,0.12878293196360271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,6144,0.04860586722691854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,5120,0.10492586294809977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,5120,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,5120,0.04160853226979573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,4096,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,4096,0.03430399894714355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,3584,0.07809706528981528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,4096,0.08754879633585612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,3584,0.03027626673380534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,3584,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,3072,0.06935359636942545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,3072,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,3072,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,2560,0.06082559823989868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,2560,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,2560,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,2048,0.0521888017654419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,2048,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,2048,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,1536,0.04382719993591309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,1536,0.007371733089288075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,1536,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,1024,0.03511999845504761
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,1024,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,1024,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,768,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,768,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,768,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,512,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,512,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,512,0.009112532933553059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,256,0.021367466449737547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,256,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,128,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,128,0.0026933332284291584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,128,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,64,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,64,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,64,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,64,32,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,64,32,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,64,32,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,65536,0.09854293664296468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,16384,0.48233706156412764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,65536,0.7529130935668945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,16384,0.014165332913398743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,16384,0.12277759710947674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,65536,3.9811070760091147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,12288,0.31580158869425456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,12288,0.0928767999013265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,12288,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,10240,0.010575999816258747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,10240,0.2593450705210368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,10240,0.07823359966278076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,8192,0.20183040301005045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,8192,0.06311253309249878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,8192,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,7168,0.1658880074818929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,7168,0.05573973258336386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,7168,0.009484799702962239
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,6144,0.12980906963348388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,6144,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,6144,0.048537600040435794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,5120,0.10550506909688313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,5120,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,5120,0.04157439867655437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,4096,0.08707199891408285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,4096,0.03379199902216594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,4096,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,3584,0.07802879810333252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,3584,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,3584,0.0315391997496287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,3072,0.006415999929110209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,3072,0.06946132977803549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,3072,0.026652799050013228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,2560,0.06055253346761068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,2560,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,2560,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,2048,0.05208746592203776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,2048,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,2048,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,1536,0.043246932824452716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,1536,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,1536,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,1024,0.03495253324508667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,1024,0.0124917338291804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,1024,0.004841599861780802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,768,0.030958932638168336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,768,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,768,0.010922666390736897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,512,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,512,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,512,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,256,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,256,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,128,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,128,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,128,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,64,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,64,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,768,32,32,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,768,32,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,768,32,32,0.006243200103441874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,16384,3.0481407165527346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,12288,2.240614318847656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,16384,2.8068522135416667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,12288,2.024516296386719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,16384,5.732113138834635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,12288,4.238233693440756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,10240,1.6903850555419921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,10240,3.4721450805664062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,10240,1.789301300048828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,8192,1.4080000559488932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,8192,1.3318826039632161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,8192,2.7822420756022135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,7168,1.1885908762613933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,7168,1.2422442118326822
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,7168,2.405853780110677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,6144,1.0248853047688802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,6144,0.9665877024332682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,6144,2.0137301127115887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,5120,0.8858922958374024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,5120,0.8713898976643881
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,4096,0.7177898406982421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,5120,1.6919211069742839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,4096,1.303210703531901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,4096,0.7203498840332031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,3584,0.6192810694376628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,3584,0.6307839711507162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,3584,1.1360596974690755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,3072,0.5496831893920898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,3072,0.551526387532552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,3072,0.9744384129842123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,2560,0.46731945673624675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,2560,0.469162654876709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,2560,0.8197802861531576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,2048,0.3915093421936035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,2048,0.3634175936381022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,2048,0.668398920694987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,1536,0.3122175852457682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,1536,0.5121024131774903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,1536,0.2916351954142252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,1024,0.23708693186442056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,1024,0.23145813941955568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,768,0.19919999440511066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,768,0.1851029396057129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,1024,0.34812587102254233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,768,0.3160405476888021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,512,0.156330664952596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,512,0.23763626416524253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,512,0.12905813058217366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,256,0.09782613118489583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,256,0.12455253601074219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,256,0.08014506498972575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,128,0.105403733253479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,128,0.05792426665623983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,128,0.07447893619537353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,64,0.08212479750315348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,64,0.06539520025253295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,64,0.046284798781077066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,65536,32,0.06341973145802816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,65536,32,0.06164480050404867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,65536,32,0.04689919948577881
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,16384,0.7687168121337891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,16384,1.2477781931559244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,12288,0.6317056020100911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,16384,0.8535722732543946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,12288,0.9605461120605469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,65536,3.4661376953125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,65536,4.972612508138021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,65536,5.925785827636719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,12288,0.5940191904703777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,10240,0.5249365488688151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,10240,0.49674240748087567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,8192,0.4200757344563802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,10240,0.781277847290039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,8192,0.6401365280151368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,8192,0.3915775934855143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,7168,0.38826665878295896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,7168,0.5518335978190104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,7168,0.3333802541097005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,6144,0.3359402656555176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,6144,0.2987349192301432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,6144,0.48421548207600906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,5120,0.2897237459818522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,5120,0.38819840749104817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,5120,0.2393770694732666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,4096,0.21654186248779297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,4096,0.30931625366210935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,4096,0.1645568052927653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,3584,0.1795072078704834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,3584,0.27326720555623374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,3584,0.14387200673421224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,3072,0.1575935999552409
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,3072,0.12877973715464275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,2560,0.1315840005874634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,2560,0.10840746561686199
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,2560,0.18947307268778485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,3072,0.2333354632059733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,2048,0.11151359875996907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,2048,0.1378986676534017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,2048,0.09096533457438151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,1536,0.09588053226470947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,1536,0.0714698632558187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,1536,0.10560853481292724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,1024,0.06748159726460776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,1024,0.0769706646601359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,768,0.057139201958974206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,768,0.05905066728591919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,1024,0.051507198810577394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,768,0.04201813141504924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,512,0.04710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,512,0.042871467272440594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,512,0.032153600454330446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,256,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,256,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,256,0.03747626543045044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,128,0.03133440017700195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,128,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,128,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,64,0.028398933013280232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,64,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,64,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,16384,32,0.027852799495061236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,16384,32,0.010683733224868774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,16384,32,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,16384,0.6910293579101563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,65536,2.32260258992513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,16384,0.9482581456502279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,16384,0.5716650644938152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,12288,0.724070421854655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,65536,4.375995890299479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,12288,0.5070848147074382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,65536,4.302506510416666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,12288,0.42110293706258134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,10240,0.42055679957071945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,10240,0.3630762736002604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,10240,0.5961727778116862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,8192,0.3322538693745931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,8192,0.4831232070922852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,7168,0.3073024113972982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,8192,0.28436158498128256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,7168,0.24466773668924965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,7168,0.4801194508870442
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,6144,0.2653183937072754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,6144,0.413865598042806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,5120,0.2352128028869629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,6144,0.19899733861287433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,5120,0.3438591957092285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,4096,0.1552725315093994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,4096,0.27978668212890623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,3584,0.14138026237487794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,4096,0.12287999788920086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,5120,0.15404373804728191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,3584,0.11250346501668293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,3584,0.23197013537089028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,3072,0.12564480304718018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,3072,0.18991786638895672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,3072,0.09788906574249268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,2560,0.11301546891530354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,2560,0.16138240496317546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,2048,0.09639253616333007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,2048,0.1316810687383016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,2048,0.06860480308532715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,2560,0.08366080125172934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,1536,0.07106560071309408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,1536,0.05553493499755859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,1536,0.10577920277913411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,1024,0.055569068590799964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,1024,0.040106666088104245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,768,0.04802560011545817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,1024,0.07120213508605958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,768,0.032529066006342575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,512,0.040447998046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,512,0.035327998797098796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,512,0.02491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,768,0.051438931624094644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,256,0.032529066006342575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,256,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,256,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,128,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,128,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,128,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,64,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,64,0.011571199695269267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,64,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,12288,32,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,12288,32,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,12288,32,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,16384,0.5627893447875977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,16384,0.8512170791625977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,65536,3.329570007324219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,65536,2.302259318033854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,12288,0.4985173225402832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,12288,0.6332757314046223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,16384,0.6619818369547527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,65536,4.222463989257813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,12288,0.4204202651977539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,10240,0.39656108220418296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,10240,0.5155839920043945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,10240,0.35047785441080725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,8192,0.4222293217976888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,8192,0.3078826586405436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,7168,0.29085013071695964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,8192,0.2746709187825521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,7168,0.4058112144470215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,7168,0.20667732556660973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,6144,0.2537471930185954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,6144,0.35519145329793295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,6144,0.1773226737976074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,5120,0.2964138666788737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,5120,0.20391252835591636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,4096,0.15612586339314777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,4096,0.12062719662984211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,4096,0.21725865999857583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,3584,0.14178986549377443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,5120,0.15503360430399576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,3584,0.1822389284769694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,3584,0.10980693499247234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,3072,0.12489386399586995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,3072,0.09499306678771972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,3072,0.16274773279825847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,2560,0.11185493469238281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,2560,0.08512852986653646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,2048,0.09349119663238525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,2048,0.13984427452087403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,2560,0.1702229340871175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,2048,0.0764245351155599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,1536,0.07970133622487387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,1536,0.05273600021998087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,1536,0.09383253256479898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,1024,0.05393066803614298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,1024,0.06079146862030029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,1024,0.038126933574676516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,768,0.048537600040435794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,768,0.04706986745198567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,768,0.03133440017700195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,512,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,512,0.03901439905166626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,512,0.034406399726867674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,256,0.031129600604375203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,256,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,256,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,128,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,128,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,64,0.024541866779327393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,64,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,64,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,128,0.026658133665720625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,10240,32,0.024405332406361897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,10240,32,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,10240,32,0.011025066177050274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,16384,0.6345727920532227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,16384,0.5371221542358399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,65536,1.6114687601725262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,65536,2.8171605428059894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,16384,0.3863552093505859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,65536,3.682338205973307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,12288,0.3930794715881348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,12288,0.4815189361572266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,12288,0.3010218620300293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,10240,0.30365012486775717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,10240,0.39618558883666993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,10240,0.23756799697875977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,8192,0.15264426867167155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,7168,0.22862505912780762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,8192,0.3156309445699056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,8192,0.24326400756835936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,7168,0.27402238845825194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,7168,0.14766079584757488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,6144,0.20162453651428222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,6144,0.23906985918680826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,6144,0.12455253601074219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,5120,0.1495039939880371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,5120,0.16237227121988934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,4096,0.12373332977294922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,4096,0.13393492698669435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,4096,0.0877568006515503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,5120,0.10577600002288819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,3584,0.11321919759114582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,3584,0.07731200059254964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,3584,0.11772586504618328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,3072,0.1008639971415202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,3072,0.09816746711730957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,3072,0.06877866586049398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,2560,0.08352426687876383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,2560,0.05922133525212606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,2560,0.08379733562469482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,2048,0.07604906558990479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,2048,0.0688810666402181
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,1536,0.05577386617660522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,2048,0.04751253525416056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,1536,0.05358933210372925
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,1536,0.03860479990641276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,1024,0.02792106668154399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,768,0.03911679983139038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,1024,0.0443391998608907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,768,0.023586134115854897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,1024,0.03686399857203166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,512,0.03314346671104431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,768,0.029183999697367353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,512,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,512,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,256,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,256,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,256,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,128,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,128,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,64,0.021980800231297812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,128,0.023859200874964397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,64,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,64,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,8192,32,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,8192,32,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,8192,32,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,16384,0.5111808141072591
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,16384,0.5683541615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,65536,1.5236095428466796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,16384,0.35799039204915367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,65536,3.6024660746256507
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,12288,0.3764906565348307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,65536,2.3421951293945313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,12288,0.4313130696614583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,12288,0.27784531911214194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,10240,0.3040255864461263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,10240,0.19278507232666015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,10240,0.3549525260925293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,8192,0.28665386835734047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,8192,0.23807786305745443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,8192,0.1516885280609131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,7168,0.21913599967956543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,7168,0.25572373072306315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,6144,0.18285226821899414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,7168,0.13598720232645672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,6144,0.19213652610778809
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,6144,0.11898880004882813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,5120,0.14592000643412273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,4096,0.12274346351623536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,5120,0.10212693214416504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,4096,0.12069546381632487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,4096,0.0828757365544637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,3584,0.10837333202362061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,5120,0.1508010705312093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,3584,0.10697387059529621
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,3584,0.07522986729939779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,3072,0.0997717301050822
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,3072,0.0934229294459025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,2560,0.08673280080159505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,2560,0.07901866436004638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,3072,0.06352213223775229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,2048,0.06662826538085938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,2560,0.05403306484222412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,2048,0.06492160161336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,2048,0.0444757342338562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,1536,0.05509119828542074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,1536,0.0495957334836324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,1536,0.037273601690928145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,1024,0.04126720031102498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,1024,0.044100264708201095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,768,0.03866879940032959
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,1024,0.025838933388392132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,768,0.027784534295399982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,768,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,512,0.03297280073165894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,512,0.023995733261108397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,512,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,256,0.014711466431617738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,256,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,128,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,128,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,128,0.010069333513577779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,64,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,256,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,64,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,64,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,7168,32,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,7168,32,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,7168,32,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,16384,0.5027125358581543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,65536,1.9333461761474608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,65536,1.4734335581461588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,65536,3.5496960957845056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,16384,0.35478188196818033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,12288,0.35730772018432616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,16384,0.48179200490315754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,12288,0.3517439842224121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,12288,0.23135573069254556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,10240,0.3049471855163574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,10240,0.18350079854329426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,8192,0.22896639506022134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,10240,0.3502079963684082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,8192,0.14649386405944825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,7168,0.20391252835591636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,7168,0.21602880160013832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,8192,0.2797909418741862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,7168,0.13441707293192545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,6144,0.17431893348693847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,6144,0.16889173189798992
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,6144,0.11799893379211426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,5120,0.14458880424499512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,5120,0.10059093634287516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,5120,0.14254080454508464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,4096,0.11502719720204671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,4096,0.12219733397165936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,4096,0.08488960266113281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,3584,0.11219627062479656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,3584,0.10134186744689941
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,3584,0.07720959981282552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,3072,0.09949653148651123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,3072,0.062395731608072914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,3072,0.0884394645690918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,2560,0.07499093214670817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,2048,0.06621760129928589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,2560,0.05324800014495849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,2560,0.08304639657338461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,2048,0.0440661350886027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,2048,0.06055253346761068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,1536,0.054749866326649986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,1536,0.03495253324508667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,1024,0.04396373430887858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,1024,0.04037973483403524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,1536,0.04700160026550293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,768,0.03819520076115926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,768,0.0321834663550059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,1024,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,768,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,512,0.03242666721343994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,512,0.02286826570828756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,512,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,256,0.02638186613718669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,256,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,256,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,128,0.010069333513577779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,64,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,128,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,128,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,64,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,6144,32,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,64,0.008598400155703227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,6144,32,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,6144,32,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,16384,0.5023061434427898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,65536,1.4873940785725912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,16384,0.39990612665812175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,65536,1.7292971293131512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,65536,3.5111254374186194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,12288,0.29296639760335286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,12288,0.3565567970275879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,16384,0.34973014195760094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,12288,0.21978453000386558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,10240,0.30617599487304686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,10240,0.1934677282969157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,8192,0.1547946612040202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,8192,0.2317311922709147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,8192,0.18746026357014972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,10240,0.250982395807902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,7168,0.20804266929626464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,7168,0.1850709279378255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,7168,0.13916160265604655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,6144,0.16373759905497234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,6144,0.12410879929860433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,6144,0.16756052970886232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,5120,0.14172159830729167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,4096,0.12137813568115234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,5120,0.14039039611816406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,4096,0.10990933577219646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,5120,0.106494935353597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,4096,0.09236053625742593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,3584,0.10905600388844808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,3584,0.09885013103485107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,3584,0.0843775987625122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,3072,0.1000106652577718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,3072,0.062054399649302164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,3072,0.08973653316497802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,2560,0.0802186648050944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,2560,0.051712000370025636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,2560,0.08570880095163981
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,2048,0.065774933497111
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,2048,0.08546986579895019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,2048,0.04256426493326823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,1536,0.05451093514760336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,1536,0.03409706751505534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,1024,0.04348586797714234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,1536,0.05348693529764811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,1024,0.039867734909057616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,768,0.03781973520914714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,1024,0.025120000044504803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,768,0.031402667363484696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,768,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,512,0.032153600454330446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,512,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,256,0.02590720057487488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,512,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,256,0.013820800185203552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,128,0.022801067431767783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,128,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,256,0.01242453356583913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,128,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,64,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,64,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,64,0.008594133456548055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,5120,32,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,5120,32,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,5120,32,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,65536,0.8675327936808268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,16384,0.37645654678344725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,65536,1.3058048248291017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,16384,0.31812267303466796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,12288,0.2636799971262614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,12288,0.24552106857299805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,16384,0.156330664952596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,65536,3.013427225748698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,12288,0.12946773370107015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,10240,0.21207040150960288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,10240,0.17855146725972493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,10240,0.10598293145497639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,8192,0.08659626642862955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,8192,0.16069973309834798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,8192,0.1384042739868164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,7168,0.11506346861521403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,7168,0.14899199803670246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,7168,0.07778986295064291
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,6144,0.12148053646087646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,6144,0.10045440196990967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,6144,0.06611626545588176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,5120,0.08536746501922607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,5120,0.10673387050628662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,5120,0.05816320180892944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,4096,0.07058239777882894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,4096,0.04686506589253743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,4096,0.09127253691355387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,3584,0.08041813373565673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,3584,0.06393173138300577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,3584,0.04208319981892904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,3072,0.05273600021998087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,3072,0.03717120091120402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,2560,0.05666133165359497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,3072,0.06536533435185751
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,2560,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,2560,0.04498773415883382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,2048,0.04877653519312541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,2048,0.03700053294499715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,2048,0.02624853253364563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,1536,0.04106239875157674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,1536,0.03239253362019857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,1536,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,1024,0.03389439980189006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,768,0.03017279903093974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,1024,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,768,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,768,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,512,0.025941334168116253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,1024,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,512,0.010613333185513813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,512,0.014331733187039694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,256,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,256,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,256,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,128,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,128,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,128,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,64,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,64,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,64,0.00631039987007777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,4096,32,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,4096,32,0.01832746664683024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,4096,32,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,65536,2.936115264892578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,65536,0.8018250783284506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,65536,1.226751963297526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,16384,0.2812949180603027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,16384,0.15291733741760255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,12288,0.19885973930358886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,12288,0.25309866269429526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,16384,0.3478869438171387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,12288,0.11472213268280029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,10240,0.21695146560668946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,10240,0.15356586774190267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,10240,0.09966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,8192,0.15472639401753743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,8192,0.12823893229166666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,8192,0.07898453076680502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,7168,0.13363199234008788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,7168,0.10693973700205486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,7168,0.06973439852396647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,6144,0.11451733112335205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,6144,0.09362346331278483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,6144,0.06341973145802816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,5120,0.10120533307393391
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,5120,0.07997439702351888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,5120,0.05379413366317749
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,4096,0.08557226657867431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,4096,0.06468266646067301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,4096,0.04304213523864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,3584,0.07232853571573893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,3584,0.05683199961980184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,3584,0.03959466616312663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,3072,0.0636245330174764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,3072,0.03420159816741943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,3072,0.04990293184916179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,2560,0.05604693492253622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,2560,0.04235946734746297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,2048,0.04860586722691854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,2560,0.029013333717981975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,2048,0.039043200016021726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,1536,0.04113066593805949
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,1536,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,2048,0.023756800095240276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,1024,0.03362133502960205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,1024,0.014472533265749613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,1024,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,1536,0.030446932713190718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,768,0.02986666758855184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,768,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,768,0.012458667159080505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,512,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,512,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,256,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,256,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,512,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,256,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,128,0.01966080069541931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,128,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,128,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,64,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,64,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,64,0.006074666480223338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3584,32,0.018357332547505698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3584,32,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3584,32,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,16384,0.24911252657572427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,65536,0.9760768254597982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,16384,0.33576958974202475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,65536,0.7762261072794596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,65536,2.8540585835774737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,16384,0.13605546951293945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,12288,0.16711680094401044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,12288,0.24647679328918456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,12288,0.10649279753367107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,10240,0.20421973864237467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,10240,0.14213120142618815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,10240,0.08970026969909668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,8192,0.1516544024149577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,8192,0.12168533007303875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,8192,0.07072426478068033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,7168,0.1266976038614909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,7168,0.10185279846191406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,7168,0.06345173517862955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,6144,0.10963626702626546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,6144,0.08779093424479166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,6144,0.05451093514760336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,5120,0.09407146771748862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,5120,0.07594666481018067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,4096,0.07768106460571289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,5120,0.047991466522216794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,4096,0.06143999894460043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,4096,0.03904853264490764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,3584,0.07007573445638021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,3584,0.05369173288345337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,3584,0.03423573176066081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,3072,0.030958932638168336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,2560,0.055637331803639736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,2560,0.040072532494862874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,3072,0.06277120113372803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,2560,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,3072,0.047577599684397384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,2048,0.048128000895182294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,2048,0.037375998497009275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,2048,0.02222080032030741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,1536,0.04082346757253011
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,1536,0.030003199974695843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,1024,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,1536,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,1024,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,768,0.02959360082944234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,1024,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,768,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,768,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,512,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,512,0.01225279966990153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,256,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,256,0.008088533580303193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,512,0.009556266665458679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,256,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,128,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,128,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,128,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,64,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,64,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,64,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,3072,32,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,3072,32,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,3072,32,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,65536,0.8374634424845377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,65536,0.721510378519694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,16384,0.33228801091512045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,16384,0.1887231985727946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,65536,2.808421325683594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,16384,0.1288533369700114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,12288,0.24019625981648765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,12288,0.13189120292663575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,12288,0.1022976001103719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,10240,0.20582399368286133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,10240,0.11478400230407715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,10240,0.08461653391520182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,8192,0.14025386174519855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,8192,0.06638933420181274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,8192,0.09359359741210938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,7168,0.09936213493347168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,7168,0.06041599909464518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,6144,0.10816853046417237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,7168,0.12445013523101807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,6144,0.0527018666267395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,6144,0.0870741367340088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,5120,0.09144319693247477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,5120,0.07962666352589926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,5120,0.045465600490570066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,4096,0.06260053316752115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,4096,0.07662826379140218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,4096,0.037034666538238524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,3584,0.06949546337127685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,3584,0.03283626635869344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,3584,0.05584213336308798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,3072,0.049322664737701416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,3072,0.062327468395233156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,3072,0.02921813329060872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,2560,0.05522773265838623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,2560,0.025461333990097045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,2048,0.04774826765060425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,2048,0.03444053332010905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,2560,0.04410453240076701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,2048,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,1536,0.026999467611312868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,1536,0.017167999347050985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,1536,0.040516265233357746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,1024,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,1024,0.01925119956334432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,1024,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,768,0.02945706645647685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,768,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,768,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,512,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,512,0.011362133423487346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,512,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,256,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,256,0.007980800171693166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,128,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,256,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,128,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,128,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,64,0.006141866743564606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,64,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,32,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2560,64,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2560,32,0.0038901334007581077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2560,32,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,65536,0.6609578450520833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,16384,0.3209898630777995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,16384,0.13513387044270833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,65536,0.6961493174235026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,65536,2.787293752034505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,16384,0.12789759635925294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,12288,0.23145813941955568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,12288,0.10099519888559978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,10240,0.092740265528361
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,10240,0.18387626012166342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,10240,0.0843775987625122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,12288,0.10397013028462727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,8192,0.1365674654642741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,8192,0.06625279982884726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,8192,0.0771071990331014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,7168,0.0669696013132731
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,7168,0.060279464721679686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,7168,0.12137493292490642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,6144,0.10615466435750325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,6144,0.05259946584701538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,6144,0.05980159838994344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,5120,0.09062399864196777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,5120,0.05017600059509277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,5120,0.04478293259938558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,4096,0.07581013043721517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,4096,0.04191253185272217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,4096,0.03638613224029541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,3584,0.06908586819966635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,3584,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,3584,0.036892799536387126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,3072,0.06164373159408569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,3072,0.0289792001247406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,3072,0.03433813254038493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,2560,0.05443946520487467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,2560,0.029047467311223346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,2560,0.024951465924580894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,2048,0.04737706581751506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,2048,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,1536,0.039765334129333495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,2048,0.02101973295211792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,1536,0.019588265816370645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,1536,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,1024,0.03259733319282532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,1024,0.01395840048789978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,1024,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,768,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,512,0.02481493353843689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,512,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,768,0.028808534145355225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,768,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,512,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,256,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,256,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,256,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,128,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,128,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,64,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,32,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,128,0.0066549330949783325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,2048,64,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,2048,32,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,2048,32,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,16384,0.3153919855753581
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,65536,0.511795202891032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,65536,0.6802090962727865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,16384,0.10813439687093099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,16384,0.12785706520080567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,65536,2.7513514200846356
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,12288,0.22725973129272461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,12288,0.08495786984761557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,12288,0.09966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,10240,0.1911466598510742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,10240,0.07393279870351156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,8192,0.06099626620610556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,10240,0.08379733562469482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,8192,0.12479146321614583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,8192,0.06597973505655924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,7168,0.12072959740956624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,7168,0.055910400549570714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,6144,0.10581333637237549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,7168,0.05973333517710368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,6144,0.049352534612019855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,6144,0.052019198735555015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,5120,0.043178665637969973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,5120,0.04464319944381714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,5120,0.09048746426900228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,4096,0.07635626792907715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,4096,0.03570346832275391
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,4096,0.03618133465449015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,3584,0.06874026457468668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,3584,0.032529066006342575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,3072,0.06143999894460043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,3584,0.03170986572901408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,3072,0.029183999697367353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,3072,0.028703999519348145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,2560,0.054408534367879235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,2560,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,2560,0.024678399165471397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,2048,0.04703573385874431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,2048,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,1536,0.03997013171513875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,2048,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,1536,0.016583466529846193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,1536,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,1024,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,1024,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,1024,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,768,0.02867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,768,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,768,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,512,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,512,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,256,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,512,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,256,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,256,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,128,0.019180800517400107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,128,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,64,0.017948800325393678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,128,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1536,32,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,64,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1536,32,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1536,32,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,16384,0.32133121490478517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,65536,0.6767616271972656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,16384,0.07676586310068766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,65536,0.3601408004760742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,12288,0.2288917382558187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,65536,2.721996815999349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,16384,0.12759040196736654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,12288,0.05857280095418295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,12288,0.09921920299530029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,10240,0.18568533261617023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,10240,0.04983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,8192,0.12127573490142822
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,10240,0.08366080125172934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,8192,0.06615039904912313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,7168,0.12100266615549723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,8192,0.039628799756368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,7168,0.037717334429423016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,6144,0.10598399639129638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,7168,0.059562667210896814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,6144,0.0335530678431193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,5120,0.09069226582845052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,5120,0.02935466567675273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,6144,0.05181440114974976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,5120,0.04450986782709758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,4096,0.07587306499481201
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,4096,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,4096,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,3584,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,3584,0.03239253362019857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,3584,0.06877866586049398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,3072,0.06133120059967041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,3072,0.020409599939982096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,3072,0.028706133365631104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,2560,0.054135465621948244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,2560,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,2560,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,2048,0.047033600012461346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,2048,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,2048,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,1536,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,1536,0.03952639897664388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,1536,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,1024,0.032153600454330446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,1024,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,1024,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,768,0.02874026695887248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,768,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,768,0.011025066177050274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,512,0.00658133327960968
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,512,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,256,0.020889600118001304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,512,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,256,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,256,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,128,0.003884800026814143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,128,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,64,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,128,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,64,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,1024,32,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,1024,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,64,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,1024,32,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,16384,0.30935039520263674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,65536,0.2797525405883789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,16384,0.10233173370361329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,65536,0.6833152135213216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,16384,0.1276586691538493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,12288,0.22022825876871743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,65536,2.726673126220703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,12288,0.04788906574249267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,12288,0.09915733337402344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,10240,0.04130133390426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,10240,0.18301973342895508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,8192,0.12264106273651124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,10240,0.08314666748046876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,8192,0.035225598017374675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,7168,0.12076373100280761
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,8192,0.06580906709035238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,7168,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,7168,0.059699201583862306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,6144,0.10591359933217366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,6144,0.027170133590698243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,6144,0.051780267556508386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,5120,0.04413439830144246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,5120,0.09035092989603678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,5120,0.023893332481384276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,4096,0.07598079840342203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,4096,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,4096,0.03594239950180054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,3584,0.06894933382670085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,3584,0.01925119956334432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,3584,0.03218773404757182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,3072,0.028467200199762982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,2560,0.05410133202870687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,3072,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,3072,0.061269334952036535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,2560,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,2560,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,2048,0.04655786752700806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,2048,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,2048,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,1536,0.039389868577321366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,1024,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,1024,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,1024,0.032153600454330446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,1536,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,1536,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,768,0.028568534056345622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,768,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,768,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,512,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,512,0.0059349333246548975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,256,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,512,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,256,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,128,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,128,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,128,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,64,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,64,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,64,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,768,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,768,32,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,768,32,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,65536,0.22070612907409667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,16384,0.31590401331583656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,65536,0.662220827738444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,65536,2.731178792317708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,16384,0.1273845354715983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,16384,0.03887786865234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,12288,0.033928533395131424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,12288,0.21868906021118165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,12288,0.0992256005605062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,10240,0.0291157325108846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,10240,0.1784490744272868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,10240,0.08301226298014322
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,8192,0.12349440256754558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,8192,0.06584320068359376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,8192,0.024473599592844644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,7168,0.12059306303660075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,7168,0.02218666672706604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,7168,0.059765334924062094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,6144,0.10588160355885823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,6144,0.020002132654190062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,6144,0.05139946540196737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,5120,0.09052159786224365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,5120,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,5120,0.04413439830144246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,4096,0.07591040134429931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,4096,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,3584,0.06857386430104574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,4096,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,3584,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,3584,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,3072,0.061269334952036535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,3072,0.02836479942003886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,2560,0.053896534442901614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,2560,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,2560,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,3072,0.01378666659196218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,2048,0.046830932299296066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,2048,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,1536,0.03956053256988525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,1536,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,2048,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,1536,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,1024,0.032290132840474446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,1024,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,1024,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,768,0.02867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,768,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,768,0.010990933577219645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,512,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,512,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,512,0.024473599592844644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,256,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,256,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,256,0.004124800115823746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,128,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,128,0.0033088001112143196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,128,0.006380799909432728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,64,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,64,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,64,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,512,32,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,512,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,512,32,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,65536,0.12847786744435627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,65536,0.6105770746866862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,16384,0.3179861386617025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,16384,0.12776106993357342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,16384,0.026037333408991496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,65536,2.7226793924967447
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,12288,0.21763413747151694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,12288,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,12288,0.0992256005605062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,10240,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,10240,0.17865386009216308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,10240,0.08304639657338461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,8192,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,8192,0.12349440256754558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,7168,0.12096853256225586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,7168,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,8192,0.065774933497111
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,7168,0.05946026643117269
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,6144,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,6144,0.1060863971710205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,6144,0.051712000370025636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,5120,0.09014613628387451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,5120,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,5120,0.04382719993591309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,4096,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,4096,0.07587839762369791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,4096,0.0358730673789978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,3584,0.06843732992808024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,3584,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,3584,0.032051199674606325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,3072,0.06143999894460043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,3072,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,3072,0.028467200199762982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,2560,0.053998935222625735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,2560,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,2560,0.024331732590993246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,2048,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,2048,0.04665813446044922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,2048,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,1536,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,1536,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,1024,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,1024,0.006037333110968272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,1536,0.03942399819691976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,1024,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,768,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,768,0.028398933013280232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,768,0.010956799983978272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,512,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,512,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,512,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,256,0.02068480054537455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,256,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,256,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,128,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,128,0.0029333333174387617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,128,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,64,0.01774719953536987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,256,32,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,64,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,256,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,256,32,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,65536,0.056109865506490074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,65536,0.5774677276611329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,16384,0.3162453333536784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,65536,2.7297793070475262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,16384,0.12759040196736654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,12288,0.22913707097371422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,16384,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,12288,0.09874773025512695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,12288,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,10240,0.1747968037923177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,10240,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,10240,0.08260266780853272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,8192,0.12311893304189045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,8192,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,8192,0.06574079990386963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,7168,0.12090026537577311
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,7168,0.05922133525212606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,6144,0.10588160355885823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,7168,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,6144,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,6144,0.051541332403818765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,5120,0.0904533306757609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,5120,0.04392960071563721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,5120,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,4096,0.07587839762369791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,4096,0.03594239950180054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,4096,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,3584,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,3584,0.06830080350240073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,3584,0.03218773404757182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,3072,0.06133759816487631
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,3072,0.028262400627136232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,3072,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,2560,0.05406293471654257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,2560,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,2048,0.04655786752700806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,2560,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,2048,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,2048,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,1536,0.039389868577321366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,1536,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,1536,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,1024,0.03287039995193482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,1024,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,768,0.028603732585906982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,1024,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,768,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,768,0.01095360020796458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,512,0.024370133876800537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,512,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,256,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,256,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,512,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,256,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,128,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,128,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,128,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,64,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,64,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,64,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,128,32,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,128,32,0.0024874667326609296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,128,32,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,65536,0.05488640069961548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,65536,0.5531989415486653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,16384,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,16384,0.3066197395324707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,65536,2.7311444600423176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,16384,0.12673707008361818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,12288,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,12288,0.2291711966196696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,12288,0.09615360101064047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,10240,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,10240,0.1712469259897868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,10240,0.08079360326131185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,8192,0.12472319602966309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,8192,0.06526293357213339
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,8192,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,7168,0.12079786459604899
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,7168,0.05809493462244669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,7168,0.007643733421961467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,6144,0.10601812998453777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,6144,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,6144,0.05034666856129964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,5120,0.09048746426900228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,5120,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,5120,0.042837333679199216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,4096,0.07594666481018067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,4096,0.03536213239034017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,4096,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,3584,0.06823253631591797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,3584,0.007403733332951863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,3584,0.031300266583760575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,3072,0.06143999894460043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,3072,0.02775040070215861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,2560,0.053896534442901614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,2560,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,2560,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,3072,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,2048,0.046728531519571945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,2048,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,1536,0.03956053256988525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,1536,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,1024,0.031982932488123575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,1536,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,2048,0.020167466004689535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,1024,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,1024,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,768,0.028398933013280232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,768,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,768,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,512,0.024405332406361897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,512,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,512,0.008871466914812724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,256,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,256,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,256,0.00709440012772878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,128,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,128,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,128,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,64,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,64,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,64,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,64,32,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,64,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,64,32,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,65536,0.029934932788213093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,65536,0.5471541086832683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,16384,0.31494827270507814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,65536,2.735616048177083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,12288,0.22824959754943847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,16384,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,16384,0.12782933712005615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,12288,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,12288,0.09639253616333007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,10240,0.17523520787556965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,10240,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,10240,0.08140799999237061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,8192,0.12509866555531818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,8192,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,8192,0.06601386864980062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,7168,0.12090026537577311
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,7168,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,7168,0.05809493462244669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,6144,0.1060863971710205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,6144,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,6144,0.050722134113311765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,5120,0.04304213523864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,5120,0.006138666470845541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,5120,0.09028266270955404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,4096,0.0756053368250529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,4096,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,4096,0.03519146839777629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,3584,0.06843732992808024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,3584,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,3584,0.031880533695220946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,3072,0.061269334952036535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,3072,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,3072,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,2560,0.05389333168665568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,2560,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,2560,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,2048,0.0466261347134908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,2048,0.005117866893609365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,1536,0.03932160139083862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,2048,0.020067199071248373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,1536,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,1024,0.03211946686108907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,1024,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,1536,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,1024,0.012590932846069335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,768,0.028569600979487103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,768,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,768,0.010682666301727295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,512,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,512,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,512,0.024541866779327393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,256,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,256,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,128,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,128,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,64,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,128,0.0026229334374268847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,512,32,32,0.01730453372001648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,64,0.005627733469009399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,512,32,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,512,32,32,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,16384,2.1799593607584637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,12288,1.536307144165039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,12288,1.646933364868164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,16384,2.2188372294108074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,12288,3.4632703145345056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,16384,4.3325439453125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,10240,1.3668693542480468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,10240,2.7327786763509114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,10240,1.3504511515299478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,8192,1.1075242360432944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,8192,1.1294719696044921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,8192,2.259899648030599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,7168,0.9659392038981119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,7168,1.0110250473022462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,7168,1.9397973378499347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,6144,0.8866133371988931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,6144,0.8353450775146485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,6144,1.6882347106933593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,5120,0.7618218739827474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,5120,0.6681941350301106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,5120,1.3332436879475913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,4096,1.0627413431803385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,4096,0.6267562866210937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,4096,0.5720064163208007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,3584,0.5592746734619141
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,3584,0.9021781285603842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,3584,0.4959232012430827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,3072,0.4640085220336914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,3072,0.4330453236897786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,3072,0.8095690409342448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,2560,0.4037973403930664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,2560,0.6613333384195964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,2560,0.3743061383565267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,2048,0.3388415972391764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,2048,0.5496831893920898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,2048,0.3141962687174479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,1536,0.2766506512959798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,1536,0.4225642522176106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,1536,0.23896212577819825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,1024,0.20910080273946127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,1024,0.33344853719075523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,1024,0.1928533395131429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,768,0.14537386894226073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,768,0.2664778709411621
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,768,0.12277759710947674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,512,0.10301439762115479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,512,0.20271786053975424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,512,0.08290987014770508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,256,0.07427413463592529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,256,0.08823466300964355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,256,0.05717333157857259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,128,0.05850453376770019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,128,0.0724992036819458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,128,0.04211733341217041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,64,0.05225813388824463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,64,0.03461120128631592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,65536,32,0.05116586685180664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,64,0.06754986445109049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,65536,32,0.025565866629282636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,65536,32,0.035327998797098796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,16384,0.722432009379069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,16384,0.62969175974528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,12288,0.5352106730143229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,12288,0.8226816177368164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,16384,1.0497023900349935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,65536,2.350455474853516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,65536,3.8450859069824217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,65536,4.292060852050781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,12288,0.46315520604451493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,10240,0.44397226969401044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,10240,0.39335254033406575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,10240,0.674235725402832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,8192,0.37324587504069007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,8192,0.5205290794372559
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,8192,0.31989758809407554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,7168,0.3463850657145182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,7168,0.26726398468017576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,7168,0.5134336153666179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,6144,0.29030399322509765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,6144,0.43796478907267256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,5120,0.23418773015340172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,6144,0.22923946380615234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,5120,0.18046293258666993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,5120,0.3623594601949056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,4096,0.1583082675933838
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,4096,0.2962090810139974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,4096,0.12619093259175618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,3584,0.14049280484517415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,3584,0.2580138683319092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,3072,0.1217194636662801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,3584,0.11165013313293456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,3072,0.09762132962544759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,2560,0.10922666390736897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,3072,0.2233344078063965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,2560,0.1650346597035726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,2560,0.08308053016662598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,2048,0.09543680349985759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,2048,0.067686398824056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,2048,0.13178880214691163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,1536,0.07983787059783935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,1536,0.10690560340881347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,1024,0.05437440077463786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,1024,0.06884160041809081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,1024,0.03969706694285075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,1536,0.054579198360443115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,768,0.04700160026550293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,768,0.0552618662516276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,768,0.031982932488123575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,512,0.03935573498408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,512,0.03560106754302979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,512,0.02491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,256,0.032015999158223465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,256,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,256,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,128,0.02719893256823222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,128,0.014165332913398743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,128,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,64,0.02505386670430501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,64,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,64,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,16384,32,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,16384,32,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,16384,32,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,16384,0.6771029154459636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,16384,0.9277066548665365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,65536,2.298845926920573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,16384,0.5688320159912109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,65536,3.298611195882162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,65536,3.6831573486328124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,12288,0.508620802561442
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,12288,0.41850881576538085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,10240,0.41840639114379885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,12288,0.7315114974975586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,10240,0.5840170542399089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,10240,0.35081812540690105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,8192,0.33966080347696936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,8192,0.4621312141418457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,8192,0.2958677291870117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,7168,0.28883625666300455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,7168,0.23985066413879394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,6144,0.24272212982177735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,6144,0.3732480049133301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,6144,0.18742613792419432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,7168,0.42175038655598956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,5120,0.19633493423461915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,5120,0.1544874668121338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,5120,0.31665493647257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,4096,0.15158613522847492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,4096,0.1215488036473592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,4096,0.25439573923746744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,3584,0.13649919827779133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,3584,0.10857813358306885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,3584,0.21070507367451988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,3072,0.12267519632975261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,3072,0.09704106648763021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,2560,0.1101471980412801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,3072,0.17834666570027669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,2560,0.15605759620666504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,2560,0.08608427047729492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,2048,0.09646080334981283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,2048,0.07662933667500814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,1536,0.07666347026824952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,1536,0.0506880005200704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,1024,0.05181440114974976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,2048,0.11263999938964844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,1536,0.09328640302022298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,1024,0.06905173460642497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,1024,0.03768320083618164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,768,0.04505600134531657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,768,0.04724053144454956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,768,0.03054400086402893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,512,0.037205334504445395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,512,0.03379199902216594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,512,0.024029866854349772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,256,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,256,0.029832533995310467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,256,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,128,0.025941334168116253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,128,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,128,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,64,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,64,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,12288,32,0.02379093368848165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,32,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,12288,64,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,12288,32,0.010921600461006164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,65536,1.6525311787923176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,16384,0.6654293060302734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,16384,0.408951473236084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,16384,0.5245269457499186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,65536,3.0548937479654947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,12288,0.523093318939209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,65536,2.7258880615234373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,12288,0.3996671994527181
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,12288,0.29614079793294273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,10240,0.3268266677856445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,10240,0.4128767967224121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,10240,0.2543274720509847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,8192,0.25654613176981605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,8192,0.3311615943908691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,8192,0.1802570660909017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,7168,0.22633493741353355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,7168,0.14021973609924315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,7168,0.290338134765625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,6144,0.18688000043233235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,6144,0.1202133337656657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,6144,0.2454528013865153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,5120,0.14243839581807455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,5120,0.22173013687133789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,4096,0.11694080034891766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,4096,0.13042346636454266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,4096,0.08260266780853272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,3584,0.10745173295338947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,3584,0.11042133172353108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,3584,0.07348906993865967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,5120,0.10291199684143067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,3072,0.09697066942850749
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,3072,0.09629013538360595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,3072,0.06430720090866089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,2560,0.08178346951802572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,2560,0.08468480110168457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,2560,0.05437440077463786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,2048,0.06458026568094889
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,2048,0.06734506289164224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,2048,0.04485119978586833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,1536,0.05348693529764811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,1536,0.05102933247884115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,1536,0.03628373146057129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,1024,0.042871467272440594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,1024,0.026555732885996504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,1024,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,768,0.028603732585906982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,768,0.03774826526641846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,768,0.02211839954058329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,512,0.032255999247233075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,512,0.021640533208847047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,512,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,256,0.026521599292755126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,256,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,256,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,128,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,128,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,64,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,64,0.007743999858697255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,64,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,32,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,10240,128,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,10240,32,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,10240,32,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,16384,0.4866389274597168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,16384,0.5359957377115886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,65536,1.541802724202474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,65536,2.061821873982747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,65536,2.9352617899576825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,16384,0.3711999893188477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,12288,0.3599018732706706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,12288,0.39297707875569665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,12288,0.2802346547444662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,10240,0.2942634582519531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,10240,0.32703145345052087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,10240,0.21596159934997558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,8192,0.22517760594685873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,8192,0.14551040331522624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,8192,0.2590378602345785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,7168,0.20285439491271973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,7168,0.13144746621449788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,7168,0.2528597354888916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,6144,0.1597098668416341
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,6144,0.20237654050191245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,6144,0.11953492959340413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,5120,0.13991146087646483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,5120,0.1008298635482788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,4096,0.11441493034362793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,4096,0.11758933067321778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,4096,0.0861525297164917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,5120,0.1458175977071126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,3584,0.1066325346628825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,3584,0.10134186744689941
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,3584,0.07379626433054606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,3072,0.08765439987182617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,3072,0.06208853324254354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,2560,0.080076797803243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,3072,0.09669973055521647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,2560,0.053452801704406736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,2048,0.0632149338722229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,2560,0.09458346366882324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,2048,0.08127146561940511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,2048,0.04348586797714234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,1536,0.05304319858551025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,1536,0.034542934099833174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,1536,0.05550080140431722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,1024,0.04225706656773885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,1024,0.03307519952456157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,768,0.037649067242940266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,1024,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,768,0.032358400026957196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,768,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,512,0.03160746693611145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,256,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,512,0.022835199038187662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,512,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,256,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,128,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,128,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,128,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,256,0.012285866340001424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,64,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,64,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,8192,32,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,64,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,8192,32,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,8192,32,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,16384,0.3600384076436361
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,65536,1.4484480539957683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,65536,1.8818037668863934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,16384,0.44741973876953123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,16384,0.48247467676798506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,65536,2.924505615234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,12288,0.3431082725524902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,12288,0.3395925203959147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,12288,0.2580479939778646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,10240,0.19186347325642902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,10240,0.28108692169189453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,10240,0.2774698575337728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,8192,0.23524692853291831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,8192,0.15486292839050292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,7168,0.1887231985727946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,8192,0.21705387433369955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,7168,0.2288640022277832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,7168,0.13701119422912597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,6144,0.158788267771403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,6144,0.12055892944335937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,6144,0.1759552001953125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,5120,0.13837653795878094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,5120,0.1413109302520752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,5120,0.10523306528727214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,4096,0.11847679615020752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,4096,0.11277653376261394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,4096,0.08970239957173666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,3584,0.10850986639658611
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,3584,0.09966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,3584,0.0793941338857015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,3072,0.09598293304443359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,3072,0.06338560183842977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,2560,0.07359146277109782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,2560,0.09588053226470947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,3072,0.11001173655192058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,2560,0.05218986670176188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,2048,0.08372373580932617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,2048,0.04222293297449748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,2048,0.06369173526763916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,1536,0.053179732958475744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,1536,0.053623465696970615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,1024,0.042049066225687666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,1024,0.0318122665087382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,1536,0.034167468547821045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,1024,0.02518826723098755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,768,0.03696639935175578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,768,0.024985599517822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,768,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,512,0.031197865804036457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,512,0.022528000672658286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,512,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,256,0.012251733740170797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,256,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,128,0.022562134265899658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,128,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,256,0.025770666201909383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,128,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,64,0.02102400064468384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,64,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,64,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,7168,32,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,7168,32,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,7168,32,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,65536,1.5790794372558594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,16384,0.486741320292155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,65536,1.505894343058268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,16384,0.35238612492879234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,16384,0.4007935841878255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,12288,0.3368618647257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,65536,2.931438954671224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,12288,0.30419839223225914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,12288,0.23524692853291831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,10240,0.26255359649658205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,10240,0.2568192005157471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,10240,0.19667627016703287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,8192,0.19872426986694336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,8192,0.2017279942830404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,7168,0.17902933756510417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,8192,0.15790079434712728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,7168,0.20565226872762046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,7168,0.1405951976776123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,6144,0.1555456002553304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,6144,0.16285012563069662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,6144,0.12458666960398357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,5120,0.13714772860209148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,5120,0.14151360193888346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,4096,0.11844053268432617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,5120,0.1052671988805135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,3584,0.10809386571248372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,4096,0.1094655990600586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,4096,0.08840533097585043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,3584,0.10024960041046142
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,3584,0.08277333577473958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,3072,0.09700693289438883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,3072,0.07058773040771485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,2560,0.08243199984232584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,2560,0.04983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,2560,0.07874560356140137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,2048,0.06273706754048666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,2048,0.08587946891784667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,3072,0.08993919690450033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,1536,0.051609599590301515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,1536,0.050756267706553136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,1024,0.04143786827723185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,1536,0.03362133502960205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,1024,0.03577173153559367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,1024,0.02491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,2048,0.04133546749750773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,768,0.03621546824773152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,768,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,768,0.031194667021433514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,512,0.030446932713190718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,256,0.025702399015426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,512,0.022459733486175536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,512,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,256,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,256,0.012009599804878235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,128,0.02228906750679016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,128,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,128,0.009520000219345093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,64,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,64,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,64,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,6144,32,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,6144,32,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,6144,32,0.005593599875768026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,16384,0.3355989456176758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,16384,0.3389098803202311
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,65536,0.8335701624552408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,65536,1.360315704345703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,65536,2.4053385416666666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,12288,0.24644266764322914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,16384,0.1810431957244873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,12288,0.12233386834462483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,10240,0.21534719467163085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,10240,0.09768959681193033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,10240,0.18674346605936687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,12288,0.25077546437581383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,8192,0.13605546951293945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,8192,0.14421332677205403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,8192,0.08294400374094645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,7168,0.12677120367685954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,7168,0.11618986924489338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,7168,0.07109973430633545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,6144,0.11168426672617596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,6144,0.06198613246281942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,6144,0.09738240242004395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,5120,0.09495893319447836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,5120,0.05536426703135172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,4096,0.0841386636098226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,4096,0.06638933420181274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,4096,0.04358826478322347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,5120,0.08270506858825684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,3584,0.06877866586049398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,3584,0.039867734909057616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,3584,0.05884586572647095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,3072,0.051336534818013514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,3072,0.06007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,3072,0.034884266058603924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,2560,0.05314559936523437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,2560,0.030446932713190718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,2560,0.0443391998608907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,2048,0.04604586760203044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,2048,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,1536,0.03921600182851155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,1536,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,2048,0.0356661319732666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,1536,0.019761067628860474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,1024,0.03218773404757182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,1024,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,1024,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,768,0.029252266883850096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,768,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,768,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,512,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,512,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,256,0.021845332781473794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,512,0.02505386670430501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,256,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,128,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,256,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,128,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,128,0.00689279983441035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,64,0.018257067600886027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,64,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,64,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,5120,32,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,5120,32,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,5120,32,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,16384,0.30477654139200844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,65536,0.7624021530151367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,65536,1.0781397501627603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,65536,2.2920191446940104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,16384,0.2625194708506266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,12288,0.20360533396402994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,16384,0.1351648012797038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,12288,0.1042090654373169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,10240,0.17353386878967286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,12288,0.2075647989908854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,10240,0.15086933771769206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,8192,0.11209386984507244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,10240,0.08731093406677246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,8192,0.1268053372701009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,8192,0.0713045358657837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,7168,0.11455146471659343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,7168,0.1035264015197754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,6144,0.10253653526306153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,6144,0.05628586610158285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,6144,0.08816640377044678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,7168,0.06345386505126953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,5120,0.09164799849192301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,5120,0.04645546674728394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,4096,0.0726698637008667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,5120,0.07519573370615641
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,4096,0.06082559823989868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,4096,0.03829760154088338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,3584,0.065774933497111
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,3584,0.03436799844106038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,3584,0.054169599215189615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,3072,0.05901653369267782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,3072,0.046826668580373126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,2560,0.05239466826121012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,2560,0.039765334129333495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,2560,0.025702399015426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,3072,0.03037866751352946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,2048,0.04532906611760457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,2048,0.03293866713841756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,2048,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,1536,0.038809601465861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,1536,0.025873066981633504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,1536,0.0177130659421285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,1024,0.031880533695220946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,1024,0.020889600118001304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,1024,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,768,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,768,0.02877333362897237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,768,0.011571199695269267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,512,0.02481493353843689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,512,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,512,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,256,0.02095253268877665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,256,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,256,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,128,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,128,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,128,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,64,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,64,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,64,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,4096,32,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,4096,32,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,4096,32,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,16384,0.23569067319234213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,65536,0.7483733495076497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,65536,0.9299285252888998
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,65536,2.2467241923014325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,16384,0.2916693369547526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,16384,0.12683946291605633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,12288,0.19776852925618488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,12288,0.10011306603749592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,10240,0.12957013448079427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,12288,0.16233812967936198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,10240,0.16482987403869628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,8192,0.11185493469238281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,8192,0.1107968012491862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,10240,0.08502613703409831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,8192,0.0664576013882955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,7168,0.11400533517201741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,7168,0.06041599909464518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,6144,0.05256533225377401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,6144,0.09980586369832357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,6144,0.07598079840342203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,7168,0.08499200344085693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,5120,0.08731306393941243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,5120,0.06628693342208862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,5120,0.045021867752075194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,4096,0.07195306619008382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,4096,0.058641068140665685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,4096,0.03645439942677815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,3584,0.06553599834442139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,3584,0.03290453354517619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,3072,0.05833386580149332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,3584,0.05205333232879639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,3072,0.03020799954732259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,3072,0.04556693236033122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,2560,0.052257064978281656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,2560,0.03857066631317139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,2560,0.02501973311106364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,2048,0.035293865203857425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,2048,0.045465600490570066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,2048,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,1536,0.03904853264490764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,1536,0.02788693308830261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,1536,0.01726933320363363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,1024,0.031948800881703696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,1024,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,1024,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,768,0.028569600979487103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,768,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,512,0.024405332406361897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,512,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,768,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,512,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,256,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,256,0.007674666742483775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,256,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,128,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,128,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,128,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,64,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,64,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,64,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3584,32,0.01798400084177653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3584,32,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3584,32,0.004060799876848856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,65536,0.8315562566121419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,16384,0.2687317212422689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,65536,2.166783905029297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,65536,0.7118848164876301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,16384,0.12622506618499757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,16384,0.21940906842549643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,12288,0.19322880109151203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,12288,0.14107306798299152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,12288,0.09915733337402344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,10240,0.1505621274312337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,10240,0.0839680035909017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,10240,0.11953492959340413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,8192,0.10810026327768962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,8192,0.06587733427683512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,8192,0.09949866930643717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,7168,0.11281066735585529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,7168,0.0994645357131958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,6144,0.09786026477813721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,7168,0.059869865576426186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,6144,0.08727893034617105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,6144,0.052121599515279136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,5120,0.08526506423950195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,5120,0.08219306468963623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,4096,0.07120213508605958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,5120,0.04427093267440796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,4096,0.059665067990620935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,4096,0.036111998558044436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,3584,0.06485333442687988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,3584,0.053350400924682614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,3584,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,3072,0.04686506589253743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,2560,0.05215573310852051
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,3072,0.029013333717981975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,3072,0.05819733142852783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,2560,0.042530135313669844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,2048,0.04481706619262695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,2560,0.024678399165471397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,2048,0.03345066706339518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,1536,0.03863893349965413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,2048,0.0210207998752594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,768,0.02867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,1024,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,1024,0.0318122665087382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,1536,0.026726400852203368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,1024,0.018567466735839845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,1536,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,768,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,768,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,512,0.024678399165471397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,512,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,256,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,512,0.009213866790135701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,256,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,256,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,128,0.018874667088190713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,128,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,128,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,64,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,32,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,3072,64,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,64,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,3072,32,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,3072,32,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,65536,2.156168619791667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,16384,0.16093866030375165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,16384,0.2675370534261068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,65536,0.6852266947428386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,65536,0.7063552220662435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,16384,0.12625919977823893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,12288,0.18800640106201172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,12288,0.11956906318664551
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,12288,0.09854293664296468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,8192,0.10905600388844808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,10240,0.11489280064900716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,10240,0.14400854110717773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,10240,0.0834218660990397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,8192,0.06550186475118001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,8192,0.09297920068105062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,7168,0.11072853406270344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,7168,0.06492160161336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,7168,0.06045013268788656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,6144,0.09717760086059571
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,6144,0.05696533521016439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,5120,0.08471893469492595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,6144,0.05208746592203776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,5120,0.048878931999206544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,4096,0.07075839837392171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,4096,0.040072532494862874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,5120,0.04471466541290283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,4096,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,3584,0.032494932413101196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,3584,0.06444373528162638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,3584,0.03580586512883504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,3072,0.0582314650217692
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,3072,0.032631466786066696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,3072,0.02867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,2560,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,2560,0.02908160090446472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,2048,0.04468053181966146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,2048,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,2560,0.05113173325856527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,2048,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,1536,0.038159998257954915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,1536,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,1536,0.016962132851282754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,1024,0.03109546701113383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,1024,0.013685333728790283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,768,0.02754453420639038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,1024,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,768,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,768,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,512,0.023995733261108397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,512,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,512,0.009078400333722432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,256,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,256,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,256,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,128,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,128,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,128,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,64,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,64,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,64,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2560,32,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2560,32,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2560,32,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,65536,0.5769557317097982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,65536,0.6637909571329753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,16384,0.2668544133504232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,65536,2.1045247395833333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,16384,0.1257472038269043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,16384,0.10868053436279297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,12288,0.185753599802653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,12288,0.0982698678970337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,12288,0.08475306828816732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,10240,0.13649919827779133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,10240,0.07396693229675293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,8192,0.10789546966552735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,10240,0.0834559996922811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,8192,0.0614741325378418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,8192,0.065774933497111
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,7168,0.11093333562215169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,7168,0.05550080140431722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,6144,0.09704106648763021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,7168,0.05925546487172445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,6144,0.048230401674906415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,5120,0.08441173235575358
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,5120,0.04229120016098022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,6144,0.05147306521733602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,5120,0.044100264708201095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,4096,0.07092906634012858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,4096,0.03604480028152466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,3584,0.06410239934921265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,3584,0.032290132840474446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,3584,0.032255999247233075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,4096,0.03566933472951253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,3072,0.05765120188395182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,3072,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,3072,0.028330665826797486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,2560,0.02450559933980306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,2560,0.05109759966532389
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,2560,0.023822933435440063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,2048,0.0443391998608907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,1536,0.037853864828745525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,2048,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,2048,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,1536,0.017061332861582436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,1536,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,1024,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,1024,0.0310591995716095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,1024,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,768,0.027955200274785357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,768,0.010062932968139648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,768,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,512,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,512,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,512,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,256,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,256,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,256,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,128,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,128,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,64,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,128,0.004363733530044556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,64,0.0038901334007581077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,2048,32,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,64,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,2048,32,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,2048,32,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,65536,0.44226560592651365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,16384,0.2584917386372884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,16384,0.08755093415578206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,65536,2.0944895426432293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,16384,0.1257472038269043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,65536,0.6474751790364583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,12288,0.1784490744272868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,12288,0.06662826538085938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,12288,0.09717760086059571
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,10240,0.05775146484375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,10240,0.13076480229695636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,8192,0.10782079696655274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,10240,0.0822271982828776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,8192,0.065228799978892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,8192,0.048640000820159915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,7168,0.11048853397369385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,7168,0.04430506626764934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,7168,0.05925546487172445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,6144,0.09689813454945882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,6144,0.03843413194020589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,6144,0.05120000044504801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,5120,0.0839680035909017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,5120,0.034030934174855546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,4096,0.0713045358657837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,4096,0.028945066531499225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,5120,0.043656531969706217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,4096,0.03556693394978841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,3584,0.06413653294245401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,3584,0.025873066981633504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,3584,0.032085333267847696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,3072,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,3072,0.028330665826797486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,3072,0.05734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,2560,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,2560,0.05102933247884115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,2560,0.024234666426976522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,2048,0.04426666498184204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,2048,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,2048,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,1536,0.037853864828745525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,1536,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,1024,0.010683733224868774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,1536,0.014264532923698425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,1024,0.031027199824651082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,1024,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,768,0.02757866581281026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,768,0.009077333410580953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,768,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,512,0.023819732666015624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,512,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,256,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,256,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,512,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,256,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,128,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,128,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,128,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,64,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,64,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,64,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1536,32,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1536,32,0.01747093399365743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1536,32,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,65536,2.0795733133951826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,65536,0.3048789342244466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,65536,0.6381568272908529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,16384,0.2600618680318197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,16384,0.10233173370361329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,16384,0.12567466894785564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,12288,0.04778666496276855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,12288,0.16834559440612792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,12288,0.09775786399841309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,10240,0.13175466855367024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,10240,0.04123306671778361
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,10240,0.08246613343556722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,8192,0.10840746561686199
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,8192,0.06519466638565063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,8192,0.034508800506591795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,7168,0.11072853406270344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,7168,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,7168,0.058641068140665685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,6144,0.09724586804707845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,6144,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,6144,0.05102613369623819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,5120,0.08447999954223633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,5120,0.023586134115854897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,4096,0.07058773040771485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,5120,0.04379306634267171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,4096,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,4096,0.03549866676330567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,3584,0.032051199674606325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,3584,0.06410239934921265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,3584,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,3072,0.057275732358296715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,3072,0.018771199385325114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,2560,0.050653866926829015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,3072,0.027955200274785357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,2560,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,2560,0.024166399240493776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,2048,0.04413439830144246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,2048,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,2048,0.020581332842508952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,1536,0.037478399276733396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,1536,0.010752000411351522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,1024,0.031089067459106445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,1536,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,1024,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,1024,0.012560000022252401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,768,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,768,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,768,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,512,0.023654399315516154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,512,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,512,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,256,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,256,0.004535466432571411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,256,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,128,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,128,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,128,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,64,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,64,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,1024,32,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,64,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,1024,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,1024,32,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,65536,0.24793920516967774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,65536,0.6216021219889323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,65536,2.0667391459147137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,16384,0.12567893664042157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,16384,0.2567509333292643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,12288,0.17093973159790038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,16384,0.04471146663029989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,12288,0.03775146802266439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,12288,0.09744959672292074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,10240,0.13271040121714275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,10240,0.08120319843292237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,8192,0.06516053279240927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,7168,0.11100160280863444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,8192,0.10898773670196533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,10240,0.031027199824651082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,8192,0.027784534295399982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,7168,0.024541866779327393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,7168,0.05877759853998819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,6144,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,6144,0.09697279930114747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,5120,0.019797333081563315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,6144,0.05116586685180664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,5120,0.08383146921793619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,5120,0.04354986747105916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,4096,0.0703488032023112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,4096,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,4096,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,3584,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,3584,0.06396586497624715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,3584,0.032255999247233075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,3072,0.057241598765055336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,3072,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,3072,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,2560,0.050756267706553136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,2560,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,2560,0.024268800020217897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,2048,0.04392533302307129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,2048,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,2048,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,1536,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,1536,0.037375998497009275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,1536,0.016445866227149962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,1024,0.031300266583760575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,1024,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,1024,0.012626133362452188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,768,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,768,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,768,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,512,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,512,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,512,0.023552000522613525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,256,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,256,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,256,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,128,0.01853440006573995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,128,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,64,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,128,0.006137600044409434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,64,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,64,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,768,32,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,768,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,768,32,0.005596800148487091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,16384,0.25439573923746744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,65536,0.5695146560668946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,65536,0.19613332748413087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,16384,0.03314346671104431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,65536,2.0643157958984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,16384,0.12561066945393878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,12288,0.16749226252237956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,12288,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,12288,0.09734826882680257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,10240,0.13165226777394612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,10240,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,10240,0.0818176031112671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,8192,0.10816853046417237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,8192,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,8192,0.06512639919916788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,7168,0.018739199638366698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,7168,0.05928959846496582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,7168,0.1106602668762207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,6144,0.09717760086059571
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,6144,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,6144,0.0506880005200704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,5120,0.08386560281117758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,5120,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,4096,0.07072426478068033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,4096,0.03543039957682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,4096,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,3584,0.06386346817016601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,3584,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,3584,0.0318122665087382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,3072,0.05720426638921102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,5120,0.04379200140635173
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,3072,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,3072,0.028296534220377607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,2560,0.05147306521733602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,2560,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,2560,0.023991467555363973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,2048,0.04427093267440796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,2048,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,1536,0.0375434676806132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,1536,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,1536,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,1024,0.030753066142400105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,1024,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,1024,0.012663466731707254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,768,0.027409066756566364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,768,0.006552533308664958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,768,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,2048,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,512,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,512,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,512,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,256,0.020309333006540933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,256,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,256,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,128,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,128,0.006141866743564606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,64,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,64,0.0026954665780067446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,64,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,512,32,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,32,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,512,32,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,512,128,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,65536,0.07229440212249756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,65536,2.082473627726237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,65536,0.4930560111999512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,16384,0.2532010714213053
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,16384,0.022016000747680665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,16384,0.12581546306610109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,12288,0.16233493487040201
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,12288,0.09772373040517171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,10240,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,10240,0.13445119857788085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,10240,0.08209066390991211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,8192,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,8192,0.10953386624654134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,12288,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,8192,0.06539946794509888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,7168,0.11127466360727947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,6144,0.05102933247884115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,7168,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,6144,0.09741653601328532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,7168,0.05915306806564331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,6144,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,5120,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,5120,0.08389973640441895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,4096,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,4096,0.07157759666442871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,5120,0.04423679908116658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,3584,0.03211946686108907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,4096,0.03594239950180054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,3584,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,3584,0.06430720090866089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,3072,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,2560,0.023859200874964397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,3072,0.02815999984741211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,2560,0.007946666578451793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,2560,0.050551466147104894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,2048,0.04396373430887858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,2048,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,3072,0.05785599946975708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,2048,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,1536,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,1536,0.037205334504445395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,1024,0.03078826665878296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,1536,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,1024,0.005767466624577841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,1024,0.012589866916338602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,768,0.027716267108917236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,768,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,768,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,512,0.023586134115854897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,512,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,512,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,256,0.020172800620396933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,256,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,128,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,128,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,128,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,64,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,64,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,64,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,256,32,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,256,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,256,32,0.002661333233118057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,256,32,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,65536,0.04474879900614421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,16384,0.257365338007609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,65536,0.49343147277832033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,16384,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,65536,2.071517817179362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,12288,0.1640447934468587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,16384,0.12588373025258381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,12288,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,12288,0.09714346726735433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,10240,0.13363199234008788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,10240,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,8192,0.10850986639658611
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,10240,0.08130559921264649
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,8192,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,7168,0.1102847973505656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,7168,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,7168,0.05836799939473471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,6144,0.09693866570790609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,6144,0.050995198885599766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,6144,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,5120,0.08420693079630534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,8192,0.06516053279240927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,5120,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,5120,0.043315200010935466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,4096,0.07055253187815348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,4096,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,4096,0.03543039957682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,3584,0.06389760176340739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,3584,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,3072,0.057241598765055336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,3072,0.0070271998643875126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,3072,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,2560,0.05079040129979452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,3584,0.031948800881703696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,2560,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,2048,0.04440746704737346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,2560,0.024098134040832518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,2048,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,2048,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,1536,0.03741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,1536,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,1024,0.03065173427263896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,1024,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,1024,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,1536,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,768,0.027033599217732747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,768,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,768,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,512,0.023586134115854897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,512,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,512,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,256,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,256,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,128,0.018602667252222697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,128,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,128,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,256,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,64,0.0024906667570273083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,64,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,64,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,128,32,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,128,32,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,128,32,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,16384,0.24941226641337075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,65536,0.48882347742716475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,65536,2.0755797068277997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,16384,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,65536,0.04208639860153198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,16384,0.12489386399586995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,12288,0.16523946126302083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,12288,0.011707733074824016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,10240,0.13397119839986166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,12288,0.09468586444854736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,10240,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,10240,0.07976960341135661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,8192,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,8192,0.06468266646067301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,7168,0.11025066375732422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,7168,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,8192,0.10847573280334473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,7168,0.05717333157857259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,6144,0.09700693289438883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,6144,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,6144,0.0497322678565979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,5120,0.08376320203145346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,5120,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,5120,0.042837333679199216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,4096,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,4096,0.035225598017374675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,4096,0.07085866928100586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,3584,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,3584,0.031027199824651082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,3584,0.06925653616587321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,3072,0.05713493426640829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,3072,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,3072,0.027272532383600872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,2560,0.05079040129979452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,2560,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,2048,0.04392960071563721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,1536,0.037649067242940266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,2048,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,2048,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,1536,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,1536,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,2560,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,1024,0.03051519989967346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,1024,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,1024,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,768,0.027135999997456868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,768,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,768,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,512,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,512,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,512,0.008799999952316284
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,256,0.020171733697255452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,256,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,256,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,128,0.0025589334468046824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,128,0.018500266472498576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,128,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,64,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,32,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,64,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,64,32,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,65536,0.023552000522613525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,64,64,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,65536,2.072098159790039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,65536,0.4915199915568034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,16384,0.25453227361043296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,16384,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,16384,0.12564480304718018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,12288,0.16199679374694825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,12288,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,10240,0.13305173714955648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,12288,0.09509546756744384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,10240,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,10240,0.08004266421000163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,8192,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,8192,0.10885120232899984
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,8192,0.06471680005391439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,7168,0.11100160280863444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,7168,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,6144,0.05003946622212728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,6144,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,6144,0.09683626492818197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,5120,0.08406933148701987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,5120,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,7168,0.05748053391774496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,5120,0.042632532119750974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,4096,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,4096,0.034747731685638425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,3584,0.06416853268941244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,4096,0.07082666556040446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,3584,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,3584,0.03133440017700195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,3072,0.05877759853998819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,3072,0.027988266944885255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,3072,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,2560,0.04966400067011516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,2560,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,2048,0.04464639822642009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,2048,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,2560,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,1536,0.037614933649698895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,2048,0.02068480054537455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,1536,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,1024,0.030481066306432086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,1024,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,1536,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,768,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,1024,0.011980799833933513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,768,0.02645333409309387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,512,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,512,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,256,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,768,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,512,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,256,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,128,0.026726400852203368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,256,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,128,0.0022528000175952912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,128,0.006035199761390686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,64,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,64,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,384,32,32,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,32,0.0026602665583292644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,384,32,64,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,384,32,32,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,16384,4.463035583496094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,12288,3.3464319864908854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,12288,3.8010538736979163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,16384,8.646792602539062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,10240,2.8052139282226562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,10240,5.411020914713541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,8192,2.329565938313802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,8192,4.491229756673177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,7168,2.139443206787109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,8192,2.6746538798014323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,7168,3.858978017171224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,7168,2.2589439392089843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,6144,1.7062911987304688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,16384,56.98822835286459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,6144,3.345680999755859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,12288,45.51246337890625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,10240,42.55457356770833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,4096,1.2306432088216146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,5120,2.8050773620605467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,4096,0.5951114654541015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,5120,1.5815338134765624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,4096,2.2756009419759113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,3584,1.2376405080159505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,5120,0.6154581069946289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,12288,55.72567138671875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,3584,0.8805717468261719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,3584,1.0811050415039063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,3072,0.43318611780802413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,3072,0.7095637639363607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,3072,0.9761109034220377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,2560,0.9233407974243164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,2560,1.4172159830729165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,2560,0.7559167861938476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,10240,64.17121175130208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,2048,0.7704234441121419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,2048,1.1202560424804688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,2048,0.6176085154215495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,1536,0.6260735829671223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,1536,0.38202025095621744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,1536,0.5048320134480794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,1024,0.37195094426472985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,1024,0.35136852264404295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,1024,0.697924296061198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,16384,82.74530436197917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,768,0.31460692087809244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,768,0.4943872133890788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,768,0.3247786521911621
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,512,0.17749333381652832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,12288,79.7444091796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,512,0.22050132751464843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,512,0.07789226373036703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,256,0.16957440376281738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,256,0.06335146824518839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,256,0.0559445341428121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,128,0.156330664952596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,128,0.04222293297449748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,128,0.051370668411254886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,8192,25.716188557942708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,64,0.067413330078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,64,0.024574933449427287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,64,0.06242986520131429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,65536,32,0.04724053144454956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,65536,32,0.037478399276733396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,65536,32,0.062463998794555664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,10240,33.13035888671875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,16384,0.6459391911824544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,8192,31.451578776041668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,16384,1.0948949178059897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,65536,4.393164571126302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,16384,2.1664085388183594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,12288,1.1532970428466798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,12288,1.8580820719401043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,7168,27.200712076822917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,12288,0.3720874786376953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,65536,9.14374491373698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,65536,3.1680511474609374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,10240,1.3960875193277995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,16384,110.4095947265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,10240,0.7617535909016927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,10240,1.0174805323282876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,8192,0.3149792035420736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,8192,47.96514892578125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,8192,0.8299861272176108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,7168,24.660957845052085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,8192,0.5214549382527669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,7168,0.3179861386617025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,6144,21.460308837890626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,7168,0.49411414464314773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,6144,0.26596693992614745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,7168,0.9686016082763672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,6144,0.837939198811849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,6144,0.1817599932352702
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,5120,0.18421759605407714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,5120,0.6548479715983073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,5120,0.33682772318522136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,4096,0.3319125175476074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,4096,0.5186218579610189
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,4096,0.21203625996907555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,3584,0.20817920366923012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,6144,20.65012003580729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,3584,0.21159253120422364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,3584,0.198963197072347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,3072,0.21084159215291343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,5120,16.181965128580728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,3072,0.3668992042541504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,3072,0.1868117332458496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,2560,0.17718613942464193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,7168,43.8818115234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,2560,0.06550186475118001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,2560,0.09526613553365072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,4096,12.672887166341145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,2048,0.19353599548339845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,2048,0.15616000493367513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,2048,0.18404693603515626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,5120,15.929241943359376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,1536,0.056012801329294835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,1536,0.042393600940704344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,1536,0.15947093963623046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,1024,0.04949333270390828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,1024,0.04990293184916179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,768,0.05522773265838623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,1024,0.03304106593132019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,6144,35.055240885416666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,768,0.0446122686068217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,4096,11.339707438151041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,512,0.04167679945627849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,768,0.03993599812189738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,5120,28.4404052734375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,512,0.031982932488123575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,3584,10.556654866536459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,256,0.027370667457580565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,256,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,512,0.022323199113210044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,256,0.01884160041809082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,128,0.03246080080668132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,4096,22.64238077799479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,128,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,128,0.01986560026804606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,3584,11.236897786458332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,64,0.021363200743993123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,64,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,3072,9.009356689453124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,64,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,16384,32,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,16384,32,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,16384,32,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,2560,8.469674682617187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,3072,10.595840454101562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,3584,21.990155029296876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,16384,0.851251220703125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,16384,0.8102229436238607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,16384,1.6973140716552735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,12288,0.9373696009318033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,65536,3.5246081034342445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,12288,1.4258516947428386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,2560,8.212991841634114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,65536,3.3745236714680993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,2560,15.665288289388021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,3072,19.98564656575521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,65536,5.612544250488281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,2048,10.061209106445313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,10240,0.329690678914388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,12288,0.6162090937296549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,10240,0.23500800132751465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,8192,0.6351872126261393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,10240,0.4562944094340007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,2048,12.034970092773438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,8192,0.1826815923055013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,8192,0.37959680557250974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,2048,6.740309143066407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,1536,5.1933141072591145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,7168,0.24466773668924965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,1536,4.957696024576823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,7168,0.13919572830200194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,7168,0.8112127939860025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,1536,9.049532063802083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,6144,0.36290559768676756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,6144,0.11103146870930988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,1024,3.587993621826172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,6144,0.30030508041381837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,5120,0.37287254333496095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,1024,3.5058347066243494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,768,2.5202687581380205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,5120,0.2949119885762533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,1024,6.947089131673176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,5120,0.09618773460388183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,768,2.9480959574381513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,4096,0.09881599744160971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,3584,0.13161813418070475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,4096,0.07540053526560465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,3584,0.09482239882151286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,4096,0.1724405288696289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,3584,0.0695637305577596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,512,2.116266632080078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,512,2.1569536844889323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,3072,0.19520853360493978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,256,1.986730702718099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,768,5.267387898763021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,3072,0.16407893498738607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,512,3.5334144592285157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,2560,0.06693546772003174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,3072,0.06331733465194703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,256,2.2751230875651043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,128,1.7792981465657551
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,2560,0.05085866848627726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,2560,0.09652907053629557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,2048,0.07372799714406332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,256,2.1662378946940106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,2048,0.0823637326558431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,2048,0.04198400179545085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,128,2.2885035196940104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,128,1.9187029520670573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,1536,0.048674134413401286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,64,2.054417037963867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,1536,0.03379199902216594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,1536,0.054408534367879235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,64,2.321509297688802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,1024,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,1024,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,1024,0.025870933135350542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,65536,32,2.50248540242513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,768,0.03256319959958394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,65536,32,2.000451151529948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,768,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,64,2.0774912516276043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,65536,32,2.241297149658203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,512,0.02815999984741211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,768,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,512,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,512,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,256,0.024268800020217897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,256,0.014643200238545737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,256,0.014131200313568116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,128,0.03031040032704671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,128,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,64,0.020002132654190062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,64,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,128,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,64,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,12288,32,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,12288,32,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,12288,32,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,12288,0.7974570592244465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,16384,0.95098876953125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,16384,1.3241002400716146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,12288,1.0565290451049805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,65536,3.134020233154297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,65536,4.8806910196940105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,16384,1.3343060811360679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,16384,17.77773234049479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,65536,5.327769470214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,12288,16.101990763346354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,12288,0.6177450815836589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,10240,0.20138667424519857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,10240,0.7407616297403972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,10240,0.8732704162597656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,16384,24.162406412760415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,8192,0.4845568021138509
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,16384,24.92238566080729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,8192,0.3642026583353678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,12288,19.330389404296874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,8192,0.7222954432169597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,7168,0.18589013417561848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,7168,0.3084287961324056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,7168,0.647441037495931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,6144,0.22814720471700034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,6144,0.21309439341227213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,6144,0.3194197336832682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,5120,0.11465386549631755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,5120,0.1817599932352702
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,5120,0.19039573669433593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,12288,11.281852213541667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,4096,0.21927253405253092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,4096,0.4744874636332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,4096,0.16175786654154461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,3584,0.23524692853291831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,3584,0.2125482718149821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,3584,0.16042666435241698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,10240,7.639722696940105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,3072,0.21073919932047525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,3072,0.12769280274709066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,3072,0.05509119828542074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,2560,0.1959935983022054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,2560,0.04655786752700806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,2560,0.20927146275838218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,2048,0.1668437321980794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,8192,9.785037231445312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,2048,0.18834773699442547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,10240,13.5437255859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,10240,12.625271606445313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,1536,0.04843519926071167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,2048,0.04666026830673218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,1536,0.04532906611760457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,1536,0.17619627316792805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,1024,0.03826346794764201
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,1024,0.037273601690928145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,1024,0.03092479904492696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,8192,5.739451599121094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,768,0.04440746704737346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,8192,10.820676676432292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,768,0.020680532852808634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,512,0.04157439867655437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,768,0.02481493353843689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,512,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,512,0.015871999661127727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,256,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,256,0.027784534295399982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,256,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,128,0.02068480054537455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,128,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,7168,5.231343078613281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,64,0.027409066756566364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,128,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,7168,8.340685017903645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,64,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,65536,64.79571533203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,64,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,7168,10.735718790690104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,10240,32,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,6144,7.077068583170574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,10240,32,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,10240,32,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,6144,9.522994995117188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,65536,0.9686698913574219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,16384,0.5002922693888346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,16384,0.5009440104166667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,16384,0.9455615997314453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,12288,0.7976618448893229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,12288,0.9655295689900717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,6144,5.167991638183594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,65536,4.36490224202474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,5120,7.490491739908855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,5120,6.439833577473958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,12288,0.3840000152587891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,4096,5.578990681966146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,65536,4.565435791015625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,10240,0.13687465985616049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,5120,6.810555521647136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,10240,0.6522538503011067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,10240,0.23722666104634604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,4096,5.870660400390625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,8192,0.15383893648783367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,65536,84.69070638020834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,8192,0.26289493242899575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,8192,0.10130773385365803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,7168,0.12878506978352863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,7168,0.48602453867594403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,6144,0.10956799983978271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,7168,0.19189759890238445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,6144,0.32754348119099935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,3584,2.504430898030599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,3584,4.46013437906901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,4096,3.3026046752929688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,6144,0.18148694038391114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,3584,4.530585734049479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,5120,0.18135040601094563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,5120,0.06427306731541951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,3072,3.8526634216308593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,3072,3.791155242919922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,5120,0.09468586444854736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,3072,2.2897664388020833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,65536,97.3490478515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,4096,0.12485973040262859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,4096,0.07229440212249756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,4096,0.051370668411254886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,3584,0.07239680290222168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,3584,0.06608213186264038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,2560,1.684991963704427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,3072,0.057241598765055336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,3584,0.047650134563446044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,2560,2.8357289632161455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,3072,0.042803200085957845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,2048,2.0136960347493487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,3072,0.06259733438491821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,2048,2.414489491780599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,2560,0.05147306521733602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,2560,0.064955735206604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,2560,3.162760416666667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,1536,1.3708287556966146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,2048,1.4364330291748046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,2048,0.04416853189468384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,1536,1.784217580159505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,2560,0.037137067317962645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,2048,0.037956265608469646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,1536,1.1004586537679037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,2048,0.029661866029103597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,1536,0.04355413516362508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,1024,0.937506103515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,1536,0.030242133140563964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,1024,1.3942442576090495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,1536,0.029013333717981975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,1024,0.045533867677052815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,1024,0.8743935902913412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,1024,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,768,0.027784534295399982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,768,1.080968475341797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,1024,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,512,0.5408767700195313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,768,0.774073600769043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,512,0.7606271743774414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,768,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,768,0.735914675394694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,768,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,256,0.4576256116231282
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,512,0.024951465924580894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,512,0.5460959752400716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,512,0.014643200238545737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,512,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,256,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,256,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,256,0.5133653322855631
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,128,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,128,0.47418025334676106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,256,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,128,0.4442111968994141
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,256,0.4957866668701172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,128,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,128,0.41813119252522785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,64,0.3874815940856934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,64,0.5056511878967285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,128,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,64,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,64,0.516812801361084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,64,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,64,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,8192,32,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,8192,32,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,16384,32,0.531660811106364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,8192,32,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,16384,32,0.5104640007019043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,16384,32,0.5497514724731445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,12288,0.25163092613220217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,16384,0.37348693211873374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,65536,0.9581567764282226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,16384,0.4084735870361328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,16384,0.23012693723042807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,12288,0.31928319931030275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,65536,1.727249018351237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,65536,1.6901120503743488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,12288,0.3637930552164713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,10240,0.4658176104227702
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,10240,0.5192021369934082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,8192,0.20739413897196451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,10240,0.2269183953603109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,8192,0.21384533246358237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,8192,0.4991658528645833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,16384,12.299161783854167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,7168,0.3611647923787435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,7168,0.1784490744272868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,7168,0.2371583938598633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,6144,0.19800747235616048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,6144,0.07789013385772706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,12288,12.965990193684897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,12288,12.955886840820312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,6144,0.2016597270965576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,5120,0.20538026491800943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,5120,0.2130261262257894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,16384,17.402162679036458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,16384,17.040418497721355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,5120,0.06311253309249878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,4096,0.18438827196757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,4096,0.048365867137908934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,4096,0.1509717305501302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,3584,0.17254400253295898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,3584,0.0575488011042277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,3584,0.056524801254272464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,3072,0.05642240047454834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,3072,0.16476160685221355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,3072,0.05051733255386352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,12288,8.508791605631512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,2560,0.06707200209299723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,2560,0.05509119828542074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,2560,0.036761601765950516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,2048,0.04471466541290283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,2048,0.05280426740646362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,10240,5.747945658365885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,2048,0.02635093331336975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,1536,0.05536426703135172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,10240,10.406468709309895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,1536,0.04638719956080119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,10240,10.5814697265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,1536,0.02222080032030741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,1024,0.03068586587905884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,1024,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,8192,8.497220357259113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,1024,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,768,0.040994131565093996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,768,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,768,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,512,0.03321173389752706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,512,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,512,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,8192,7.8657877604166675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,8192,4.250965372721354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,256,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,256,0.013346133629480996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,256,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,128,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,128,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,128,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,64,0.005321600039800008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,64,0.018500266472498576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,7168,6.918724060058594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,7168,3.8029652913411462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,64,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,7168,32,0.025770666201909383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,7168,32,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,7168,11.508019002278646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,7168,32,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,65536,52.069173177083336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,6144,6.203938293457031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,6144,5.788906860351562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,16384,0.33143466313680015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,65536,0.8680447896321615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,65536,2.19139404296875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,16384,0.49701547622680664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,6144,3.2585728963216147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,65536,1.5538516998291017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,16384,0.7892309188842773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,12288,0.448034127553304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,12288,0.6826037089029948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,12288,0.3247786521911621
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,5120,5.99534912109375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,5120,4.989439900716146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,10240,0.3465898513793945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,10240,0.5041152000427246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,5120,3.112925974527995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,8192,0.11281066735585529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,10240,0.32201385498046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,65536,67.08640543619791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,4096,4.7982935587565105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,8192,0.1737727959950765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,8192,0.3832490603129069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,7168,0.11284373601277668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,4096,4.238370259602865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,4096,2.3937705993652343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,7168,0.08632319768269857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,7168,0.10622293154398602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,6144,0.10031466484069824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,6144,0.1646933396657308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,6144,0.09113600254058837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,65536,70.62476399739583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,3584,3.497608439127604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,5120,0.0884394645690918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,5120,0.07137280305226644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,3584,1.8211157480875653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,5120,0.1602560043334961
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,3072,2.8438186645507812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,3584,3.85976308186849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,3072,1.6117759704589845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,3072,3.136204783121745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,4096,0.07058773040771485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,4096,0.07007573445638021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,2560,2.463500722249349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,3584,0.06236159801483154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,4096,0.04655786752700806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,3584,0.0549888014793396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,3584,0.04174506664276123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,3072,0.05553493499755859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,2560,1.3307210286458333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,3072,0.04775253136952718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,3072,0.037614933649698895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,2560,0.0679253339767456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,2048,1.8868565877278645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,1536,1.3096277872721354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,1536,1.0935296376546224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,2560,0.04123306671778361
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,2560,2.3193941752115887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,2048,1.6639317830403644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,2048,1.1346602121988931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,2560,0.032255999247233075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,2048,0.04659200112024943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,1536,0.8449024200439453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,1536,0.030958932638168336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,2048,0.03925333420435588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,1536,0.037102933724721274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,2048,0.02645333409309387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,1024,0.7486122767130534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,1024,0.030719999472300214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,1536,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,1024,0.6949546813964844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,1024,1.0358783721923828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,768,0.5769557317097982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,1024,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,1024,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,768,0.8142847696940103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,512,0.43697493871053056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,768,0.5618005116780599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,512,0.5708799997965495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,768,0.027306665976842243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,512,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,768,0.014233600099881491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,512,0.024132267634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,768,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,256,0.3850581487019857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,256,0.3504128138224284
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,512,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,512,0.429533863067627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,256,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,256,0.44663467407226565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,256,0.008327466746171314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,128,0.35351893107096355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,256,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,64,0.35085652669270834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,128,0.29241921106974283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,128,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,128,0.37451092402140296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,64,0.28641281127929685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,64,0.39669758478800454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,128,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,128,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,64,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,64,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,12288,32,0.3718144098917643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,64,0.006960000097751618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,6144,32,0.01941653291384379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,12288,32,0.28822186787923176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,6144,32,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,12288,32,0.40755198796590164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,6144,32,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,16384,0.27275946935017903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,65536,1.45349973042806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,16384,0.35911680857340494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,65536,0.8147274653116862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,12288,0.1990997314453125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,65536,1.1897504170735678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,16384,0.3368618647257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,12288,0.5053098678588868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,12288,0.307370662689209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,10240,0.3158357302347819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,10240,0.47889067331949875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,10240,0.36078933080037434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,8192,0.19828054110209148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,8192,0.1798144022623698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,8192,0.21616640090942382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,7168,0.3273728052775065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,16384,10.241158040364583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,12288,10.606490071614584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,16384,14.489087931315103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,16384,16.209988403320313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,7168,0.08779093424479166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,7168,0.30501546859741213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,12288,12.234377034505208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,6144,0.200874662399292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,6144,0.12431360085805257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,6144,0.19677866299947103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,5120,0.0575488011042277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,5120,0.08267093499501546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,5120,0.19851946830749512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,4096,0.16534186999003092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,4096,0.16390825907389323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,4096,0.05259946584701538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,3584,0.040277334054311116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,3584,0.06062080065409342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,3072,0.05437440077463786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,3584,0.05655893484751383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,3072,0.05236053466796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,12288,7.168648783365886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,10240,4.4224853515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,3072,0.04778666496276855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,2560,0.051336534818013514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,2560,0.048571733633677165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,2560,0.03628373146057129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,2048,0.06297599871953329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,10240,8.612961832682291
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,2048,0.035225598017374675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,2048,0.025361067056655882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,1536,0.039287467797597245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,10240,10.143675740559896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,8192,7.854625956217449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,1536,0.045841066042582194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,1024,0.0370688001314799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,1536,0.021640533208847047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,1024,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,1024,0.021367466449737547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,768,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,8192,3.655884806315104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,768,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,768,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,65536,39.59422200520833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,8192,6.661086018880208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,512,0.03560106754302979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,512,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,512,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,7168,3.0489547729492186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,256,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,256,0.007815466821193695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,7168,6.726348876953125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,256,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,7168,5.8939051310221355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,128,0.026897066831588747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,128,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,64,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,128,0.010513066252072652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,64,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,64,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,5120,32,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,6144,5.282747904459635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,6144,4.058180236816407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,6144,6.264900207519531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,5120,32,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,5120,32,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,5120,4.028450012207031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,5120,2.6843137105305988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,5120,5.255065409342448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,65536,0.5558272043863932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,65536,2.1735081990559895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,16384,0.09591466585795085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,16384,0.22715627352396645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,65536,57.556819661458334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,16384,0.5351423899332682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,12288,0.14714879989624025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,12288,0.19469653765360515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,4096,1.7531562805175782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,65536,4.165256500244141
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,4096,3.2103424072265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,4096,4.440951538085938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,12288,0.08308053016662598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,10240,0.10646186669667561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,3584,3.2737622578938805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,3584,2.5592491149902346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,10240,0.06369280020395915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,10240,0.09768746693929037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,3584,1.5068501790364583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,8192,0.09024853706359863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,8192,0.07860906918843588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,8192,0.0500053326288859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,3072,2.927001698811849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,7168,0.1984512011210124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,7168,0.0695637305577596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,7168,0.04659200112024943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,3072,1.495244852701823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,2560,2.270344543457031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,3072,2.1997225443522135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,65536,70.37989908854166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,6144,0.19237546920776366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,6144,0.06868053277333577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,2560,2.043153127034505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,2560,1.0951680501302083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,6144,0.040447998046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,2048,1.6063147226969403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,5120,0.06987093289693197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,4096,0.05809493462244669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,5120,0.034577067693074545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,5120,0.05120533307393392
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,4096,0.04195199807484944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,2048,0.871116828918457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,2048,1.465002695719401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,1536,1.0145792007446288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,1024,0.6454613367716472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,1536,1.1044181823730468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,4096,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,3584,0.06628693342208862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,3584,0.037512532869974774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,1536,0.7336277643839518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,1024,0.8616277058919272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,3584,0.030001066128412884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,3072,0.048913065592447916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,1024,0.5469184239705404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,3072,0.0233130673567454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,2560,0.04399786790211995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,3072,0.03467520078023274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,768,0.4770815849304199
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,768,0.5102250734965007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,768,0.6500693639119466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,512,0.3773098627726237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,2560,0.03051519989967346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,2560,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,512,0.4544853210449219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,2048,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,512,0.35382614135742185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,256,0.30607360204060874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,2048,0.024029866854349772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,1536,0.03331306576728821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,256,0.2684927940368652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,1536,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,256,0.3180885314941406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,1536,0.013755733768145243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,128,0.24268800417582193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,1024,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,1024,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,128,0.2877098719278971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,128,0.294707202911377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,1024,0.010376532872517902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,64,0.3098282814025879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,2048,0.03956053256988525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,768,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,64,0.325767453511556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,768,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,64,0.23821652730305992
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,10240,32,0.30979413986206056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,768,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,10240,32,0.24255146980285644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,512,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,10240,32,0.3395242691040039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,512,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,512,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,256,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,256,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,128,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,128,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,128,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,64,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,64,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,64,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,4096,32,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,32,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,4096,32,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,16384,8.301465352376301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,4096,256,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,12288,8.658875528971354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,16384,0.45455360412597656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,65536,1.1477333068847657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,65536,1.9473066965738934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,16384,0.48793598810831706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,12288,11.243553670247396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,16384,0.31812267303466796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,65536,2.930517323811849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,12288,0.126255997021993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,16384,13.776964314778647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,16384,15.632588704427084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,12288,0.15404373804728191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,10240,0.09188693364461263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,10240,0.09775786399841309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,10240,0.15158613522847492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,8192,0.20735999743143715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,8192,0.22005759874979652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,8192,0.0449834664662679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,12288,5.707225545247396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,7168,0.18394452730814617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,7168,0.04157439867655437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,7168,0.17445546785990398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,10240,3.65308583577474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,6144,0.20483412742614746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,10240,6.785399373372395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,6144,0.07765333652496338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,10240,8.971502685546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,12288,0.1390933354695638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,6144,0.05928959846496582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,5120,0.08297812938690186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,8192,7.4180948893229175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,5120,0.05553493499755859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,4096,0.05573973258336386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,5120,0.03197866678237915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,65536,32.255828857421875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,4096,0.060108800729115806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,4096,0.03239253362019857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,8192,5.262336222330729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,3584,0.04676266511281331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,3584,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,8192,3.0978731791178387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,3072,0.048571733633677165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,3072,0.031778132915496825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,3072,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,7168,6.200934346516927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,7168,4.571067810058594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,2560,0.06007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,7168,2.553343963623047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,2560,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,2560,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,2048,0.03607893387476603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,3584,0.05181119839350382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,2048,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,6144,2.1072213490804037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,2048,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,6144,3.8564521789550783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,1536,0.032255999247233075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,65536,45.83430989583333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,6144,5.77792002360026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,1536,0.01986560026804606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,1536,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,1024,0.027135999997456868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,5120,1.7790602366129558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,5120,3.146649678548177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,1024,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,5120,4.420368957519531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,768,0.024234666426976522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,768,0.011025066177050274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,768,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,512,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,512,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,4096,1.5283541361490884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,4096,2.4061609903971357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,4096,3.4895530700683595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,256,0.01853440006573995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,512,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,1024,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,3584,3.1465131123860677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,3584,1.1777706146240234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,3584,2.0163584391276044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,256,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,256,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,65536,58.413431803385414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,128,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,3072,2.5778859456380205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,128,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,128,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,64,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3584,32,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3584,32,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,3072,1.2437845865885415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,3072,1.8773333231608074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,64,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,2048,1.3530794779459634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,2560,0.8915285110473633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,2560,1.4824447631835938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,2560,2.0955477396647137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3584,32,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,2048,0.6993920008341472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,2048,1.1668139139811198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,16384,0.22367572784423828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,16384,0.07782399654388428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,65536,1.208115259806315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,65536,0.750216547648112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,12288,0.10632533232371014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,16384,0.187391996383667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,65536,0.44960425694783523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,1536,0.5829290390014649
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,1536,0.8362325032552084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,1536,1.0144426981608072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,1024,0.5580458958943685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,12288,0.09144319693247477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,1024,0.687991460164388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,768,0.436736011505127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,1024,0.4268373489379883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,768,0.5234005292256673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,12288,0.06082559823989868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,10240,0.08796160221099854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,10240,0.05232640107472738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,768,0.3300693194071452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,10240,0.0756053368250529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,8192,0.07372799714406332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,8192,0.06829546292622884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,8192,0.0400383989016215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,512,0.3220117251078288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,256,0.2151082674662272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,512,0.2643285274505615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,7168,0.08550399939219157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,256,0.2542933305104574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,512,0.3667626698811849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,256,0.24016213417053223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,128,0.22715733846028646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,7168,0.05618346532185873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,7168,0.03894613186518352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,6144,0.07417173385620117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,128,0.19537919362386066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,6144,0.05044906536738077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,5120,0.06437546809514363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,6144,0.03362133502960205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,5120,0.04276906649271647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,64,0.22606186866760253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,128,0.22432106335957847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,5120,0.02815999984741211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,64,0.20681385993957518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,64,0.24702293078104653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,8192,32,0.1959935983022054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,8192,32,0.2299498716990153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,4096,0.05307733217875162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,8192,32,0.3306495984395345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,4096,0.03556693394978841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,4096,0.023005867004394533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,3584,0.07154346307118734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,3584,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,3584,0.0335530678431193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,3072,0.045533867677052815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,3072,0.028600533803304035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,3072,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,2560,0.04106239875157674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,2560,0.03816106716791789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,2560,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,2048,0.04174506664276123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,2048,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,2048,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,1536,0.03068586587905884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,1536,0.02362026572227478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,1536,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,16384,7.17318827311198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,1024,0.03819520076115926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,16384,9.94696553548177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,12288,7.111611938476562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,1024,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,1024,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,768,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,768,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,12288,10.638608805338542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,768,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,512,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,512,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,16384,13.98309529622396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,512,0.0075082664688428235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,256,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,256,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,128,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,256,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,128,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,12288,4.876151529947917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,64,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,128,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,64,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,10240,3.57215576171875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,64,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,10240,6.988014729817708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,3072,32,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,3072,32,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,3072,32,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,10240,9.067588297526042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,65536,1.400320053100586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,8192,7.255279032389323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,65536,1.120699691772461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,65536,29.34599609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,65536,1.13974609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,16384,0.1773226737976074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,16384,0.3459413210550944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,8192,4.843690490722656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,16384,0.19739306767781575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,12288,0.20582399368286133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,12288,0.10171733697255451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,8192,2.9824681599934895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,12288,0.05379413366317749
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,10240,0.0834218660990397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,7168,3.805047353108724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,10240,0.06417066653569539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,10240,0.17370452880859374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,7168,6.21673838297526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,8192,0.07031146685282388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,8192,0.0609279990196228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,7168,0.18612906138102214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,7168,2.264029947916667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,65536,38.645145670572916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,8192,0.05239466826121012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,6144,3.3256790161132814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,7168,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,7168,0.046011734008789065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,6144,5.4549504597981775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,6144,1.9375445048014324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,6144,0.07089493274688721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,6144,0.04184746742248535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,6144,0.02959360082944234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,5120,0.062327468395233156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,5120,0.037444265683492024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,5120,0.024985599517822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,5120,4.456550598144531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,5120,1.6889514923095703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,5120,2.7304959615071613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,4096,0.053384534517923986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,4096,0.031982932488123575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,4096,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,4096,3.533516693115234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,3584,0.048844798405965166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,4096,1.2463104248046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,3584,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,4096,2.1722452799479166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,3584,0.01972800095876058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,3072,0.04392960071563721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,3072,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,3584,3.099955240885417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,3072,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,2560,0.04041386842727661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,3584,1.0868053436279297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,3584,1.8282496134440105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,65536,56.763936360677086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,2560,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,3072,2.476714579264323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,2560,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,2048,0.03553280035654704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,2048,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,3072,1.5875072479248047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,3072,0.8643242518107096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,2048,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,1536,0.03089066743850708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,1536,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,2560,1.9933172861735027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,1536,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,2560,1.259485880533854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,2560,0.7714474360148112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,2048,1.3807957967122397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,1024,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,2048,0.994986661275228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,1024,0.010752000411351522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,2048,0.6230357487996419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,1024,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,768,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,768,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,768,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,512,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,1536,0.7847594579060873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,512,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,1536,0.8360618591308594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,1536,0.4990976015726726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,1024,0.5254485448201497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,1024,0.38068908055623374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,512,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,1024,0.6073002497355143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,768,0.4148906707763672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,256,0.01928106745084127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,256,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,256,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,768,0.45349547068277996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,128,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,64,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,128,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,128,0.005660800139109293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,768,0.3126954714457194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,512,0.2960725466410319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,256,0.22913707097371422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,64,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,512,0.32088747024536135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,256,0.21449066797892252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,512,0.23500800132751465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,256,0.18585599263509114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,64,0.005251200000445048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2560,32,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2560,32,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2560,32,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,128,0.2063360055287679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,65536,1.1253419240315756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,65536,0.3729738553365072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,128,0.16745813687642414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,128,0.19837867418924968
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,64,0.20456107457478842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,65536,0.5130240122477214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,16384,0.36177921295166016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,64,0.21315946578979492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,64,0.17162240346272786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,7168,32,0.2098837375640869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,7168,32,0.20145492553710936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,16384,0.06341973145802816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,16384,0.15605759620666504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,12288,0.0955733299255371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,12288,0.058806399504343666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,7168,32,0.21940906842549643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,12288,0.048878931999206544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,10240,0.08215893109639485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,10240,0.0501418670018514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,10240,0.0527018666267395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,8192,0.06970026493072509
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,8192,0.04242240190505982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,8192,0.03430399894714355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,7168,0.16008532842000325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,7168,0.03829760154088338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,7168,0.030242133140563964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,6144,0.19548160235087078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,6144,0.042871467272440594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,6144,0.026862933238347368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,16384,6.323916625976563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,5120,0.16261119842529298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,5120,0.028808534145355225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,5120,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,4096,0.16110933621724446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,12288,6.16287587483724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,16384,8.302590942382812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,4096,0.02491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,4096,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,3584,0.1534293333689372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,12288,10.3081298828125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,3584,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,16384,13.45652058919271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,3584,0.026589866479237872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,3072,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,3072,0.043178665637969973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,3072,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,2560,0.03952639897664388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,2560,0.022016000747680665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,2560,0.021542400121688843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,12288,4.284757486979166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,2048,0.03601066668828328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,2048,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,2048,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,10240,4.931003824869792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,10240,2.813678995768229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,1536,0.03204480012257894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,1536,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,65536,23.450248209635415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,1536,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,10240,8.320750935872395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,1024,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,1024,0.0362496018409729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,1024,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,768,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,8192,6.646647644042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,768,0.023586134115854897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,768,0.011025066177050274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,512,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,8192,3.7857279459635413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,512,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,8192,2.3076833089192705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,512,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,256,0.028637866179148357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,65536,32.97542928059896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,7168,3.1333717346191405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,256,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,7168,2.0267691294352215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,256,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,7168,5.8578943888346355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,128,0.01884160041809082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,128,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,128,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,64,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,64,0.005355733136336008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,64,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,2048,32,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,2048,32,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,6144,2.7205291748046876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,6144,1.883716328938802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,6144,5.688729349772135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,2048,32,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,5120,1.2982271830240886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,5120,2.3589886983235675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,5120,4.251545715332031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,65536,1.0679295857747397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,65536,0.38488747278849283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,16384,0.14585173924763997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,16384,0.061269334952036535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,65536,0.713045310974121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,12288,0.0964949369430542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,16384,0.06393173138300577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,4096,3.8612650553385413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,12288,0.045499734083811444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,4096,1.047381337483724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,3584,1.0174805323282876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,4096,1.8010111490885419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,12288,0.05092693169911703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,3584,1.5270570119222007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,10240,0.17500160535176595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,3584,3.058244323730469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,3072,2.3798784891764324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,10240,0.03952639897664388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,10240,0.04130133390426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,8192,0.0425983985265096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,8192,0.035257601737976076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,8192,0.06959786415100097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,3072,1.301367441813151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,65536,54.12942708333334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,7168,0.07881386280059814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,3072,0.7526058832804362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,7168,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,2560,1.0726058959960938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,6144,0.07123626867930094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,2560,1.8879435221354168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,6144,0.027135999997456868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,7168,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,6144,0.026897066831588747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,2560,0.7780352274576823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,2048,0.9616725285847982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,5120,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,5120,0.06382933457692465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,2048,1.2170581817626953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,2048,0.5622101465861002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,5120,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,1536,0.7735978444417317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,4096,0.05188266833623251
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,4096,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,1536,0.728985595703125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,4096,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,1536,0.429260794321696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,3584,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,3584,0.04833279848098755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,1024,0.5184170722961425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,1024,0.3295232137044271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,1024,0.4737023989359538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,3584,0.017917867501576742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,3072,0.04382719993591309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,768,0.3695274670918783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,768,0.38720852533976235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,3072,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,2560,0.03870720068613688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,2560,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,768,0.3552255948384603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,3072,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,512,0.20520960489908852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,2560,0.014472533265749613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,512,0.268285878499349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,2048,0.03512320121129354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,512,0.2769578615824381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,256,0.2085205396016439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,2048,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,2048,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,256,0.15404373804728191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,256,0.18442239761352539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,128,0.17988266944885253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,1536,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,1536,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,1536,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,128,0.1673898696899414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,128,0.14079999923706055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,1024,0.025838933388392132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,1024,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,1024,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,64,0.18001920382181805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,64,0.14021973609924315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,64,0.18769920667012532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,768,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,768,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,6144,32,0.14086826642354328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,6144,32,0.17216426531473797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,768,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,6144,32,0.18831253051757812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,512,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,512,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,512,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,256,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,256,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,128,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,128,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,128,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,64,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,64,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,64,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1536,32,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1536,32,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1536,32,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,16384,5.52970225016276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,16384,7.460761515299479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,12288,5.1663874308268225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,65536,0.6159360249837239
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,65536,0.6732458750406901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,65536,0.9901738484700522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,16384,0.34682881037394203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,16384,0.040243200461069745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,16384,0.0630741318066915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,12288,0.09642666975657145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,12288,9.86053975423177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,16384,14.121505737304688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,12288,0.046728531519571945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,12288,0.048230401674906415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,10240,0.08191786607106527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,10240,0.029388799269994097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,10240,0.06761813163757324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,12288,3.348991902669271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,8192,0.025326933463414508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,8192,0.184661340713501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,8192,0.05505706469217936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,10240,4.133717346191406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,7168,0.17216854095458983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,65536,20.76545613606771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,10240,7.975492350260417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,7168,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,7168,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,6144,0.06963199774424235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,10240,2.478899129231771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,6144,0.026555732885996504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,6144,0.026999467611312868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,5120,0.06519466638565063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,5120,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,65536,27.181260172526045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,5120,0.023209599653879802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,8192,6.516701761881511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,8192,1.9920895894368489
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,8192,3.309565989176432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,4096,0.05239466826121012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,4096,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,4096,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,3584,0.04836053450902303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,3584,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,3584,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,7168,2.8010836283365887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,3072,0.04355413516362508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,7168,1.7052331288655598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,3072,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,7168,5.815531921386719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,3072,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,2560,0.05085866848627726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,6144,1.4794410705566405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,2560,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,6144,2.447632090250651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,2560,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,2048,0.03426986535390218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,2048,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,2048,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,5120,3.9837641398111976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,6144,5.142933146158854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,1536,0.030481066306432086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,5120,1.9796651204427083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,5120,1.1413845062255858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,1536,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,1024,0.03054933349291484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,1536,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,1024,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,1024,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,4096,3.100091807047526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,768,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,4096,0.8958293279012045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,4096,1.547537104288737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,768,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,768,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,3584,1.3560831705729166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,512,0.021367466449737547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,512,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,3584,0.8303616205851236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,3584,2.6818219502766927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,512,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,65536,51.84119466145834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,256,0.0042303999265035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,3072,2.2862505594889324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,256,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,3072,0.6993237177530924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,128,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,256,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,3072,1.1587242126464843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,2560,1.884979248046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,128,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,64,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,128,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,2560,0.5957632064819336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,2560,0.9498624165852865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,64,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,2048,1.134762700398763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,1024,32,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,64,0.005011199911435445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,2048,0.9004693349202475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,1024,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,2048,0.48523839314778644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,1536,0.6679210662841797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,1024,32,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,1536,0.5880149205525715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,1536,0.387720521291097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,65536,0.9721173604329427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,65536,0.22476800282796225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,65536,0.24521387418111168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,1024,0.44417705535888674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,16384,0.12447893619537354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,1024,0.4471466700236003
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,16384,0.0338261326154073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,16384,0.06301013231277466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,768,0.44926293690999347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,1024,0.27931305567423503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,12288,0.09608533382415771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,768,0.22685012817382813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,768,0.3274751981099447
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,512,0.24169813791910805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,12288,0.048401065667470294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,10240,0.0831829309463501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,512,0.22920533816019692
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,12288,0.02754559914271037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,10240,0.024200532833735147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,512,0.1586176077524821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,10240,0.040755200386047366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,8192,0.06976853211720785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,256,0.18169172604878742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,256,0.14312106768290203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,8192,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,256,0.14335892995198568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,8192,0.03362133502960205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,128,0.1511082649230957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,7168,0.07837013403574625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,7168,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,128,0.12772693634033203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,128,0.11786239941914875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,64,0.14714879989624025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,6144,0.06877866586049398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,64,0.11656533082326252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,7168,0.03041279911994934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,6144,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,5120,32,0.1265664021174113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,5120,0.059903999169667564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,6144,0.026653865973154705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,64,0.1346901257832845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,5120,32,0.11666773160298664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,5120,0.020307199160257975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,5120,0.023108265797297158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,4096,0.051950931549072266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,5120,32,0.14458880424499512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,3584,0.04782079855600993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,4096,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,3584,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,3584,0.017709867159525553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,4096,0.014165332913398743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,3072,0.055637331803639736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,3072,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,3072,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,16384,4.47808837890625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,16384,5.411703491210938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,2560,0.04369066556294759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,2560,0.010410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,2560,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,2048,0.04212053219477336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,12288,3.8979242960611975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,2048,0.010711466272672016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,1536,0.04194986820220947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,2048,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,1536,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,12288,9.120972696940104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,16384,12.09019775390625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,1536,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,1024,0.006856533388296763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,1024,0.02491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,1024,0.008463999629020691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,768,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,512,0.021160533030827842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,12288,2.8653910319010416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,768,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,768,0.027135999997456868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,65536,17.158109537760417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,10240,3.18566411336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,512,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,512,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,256,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,256,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,65536,22.00637410481771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,256,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,10240,7.796018981933594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,10240,2.174976094563802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,128,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,128,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,128,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,8192,2.82241694132487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,8192,6.127684020996094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,64,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,64,0.0021845333278179167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,768,32,0.0020138667275508243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,768,32,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,64,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,768,32,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,8192,1.793499755859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,7168,2.2337877909342447
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,7168,2.1864107767740886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,7168,5.498606872558594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,65536,2.163507080078125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,65536,0.15035732587178546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,65536,0.5047978719075521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,16384,0.12458666960398357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,6144,4.882875569661459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,16384,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,6144,1.2264106750488282
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,16384,0.06263466676076254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,6144,1.85886713663737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,12288,0.198963197072347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,12288,0.023483733336130776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,5120,1.545898691813151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,12288,0.048093867301940915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,5120,1.2455594380696615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,5120,3.7131604512532554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,10240,0.08215893109639485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,10240,0.04109653234481812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,10240,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,8192,0.07144107023874918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,8192,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,7168,0.078165332476298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,8192,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,4096,0.7579647699991863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,4096,1.2126548767089844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,4096,3.097634124755859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,3584,2.5441622416178387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,7168,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,3584,1.063486925760905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,3584,0.613375981648763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,6144,0.01365226705869039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,6144,0.06983679930369059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,5120,0.06079146862030029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,6144,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,5120,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,7168,0.030374399820963544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,5120,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,65536,49.44810791015625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,3072,2.13937505086263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,3072,0.5791402816772461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,3072,0.9123146692911783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,2560,1.6754656473795575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,4096,0.05249706506729126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,2560,0.489847469329834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,2560,0.75011412302653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,3584,0.04761600097020467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,4096,0.011941333611806233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,4096,0.01914560000101725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,2048,1.0435242970784506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,3584,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,3584,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,3072,0.043383467197418216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,3072,0.009898666540781658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,1536,0.6060373306274414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,2048,0.3964927991231283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,3072,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,2048,0.6199637095133463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,1024,0.3975850741068522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,1536,0.46608638763427734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,1536,0.31412906646728517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,1024,0.3176447868347168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,2560,0.039185067017873124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,2560,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,2048,0.034884266058603924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,2560,0.013892267147699991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,1024,0.20865707397460936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,2048,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,1536,0.03027626673380534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,2048,0.011843199531237285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,512,0.2113877296447754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,768,0.29870080947875977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,768,0.15182506243387858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,512,0.18496853510538738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,1536,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,256,0.15404373804728191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,512,0.11922773520151775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,768,0.2596522649129232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,1024,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,1024,0.025770666201909383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,1024,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,256,0.10298026402791341
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,768,0.023005867004394533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,1536,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,768,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,256,0.07942826747894287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,768,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,512,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,128,0.11502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,128,0.05932373205820719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,128,0.08721066315968831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,64,0.0726698637008667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,64,0.04918613433837891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,512,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,64,0.04833279848098755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,512,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,256,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,4096,32,0.06942719618479411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,256,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,4096,32,0.04143679936726888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,128,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,128,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,256,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,128,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,64,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,4096,32,0.04990293184916179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,64,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,64,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,512,32,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,512,32,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,512,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,65536,0.045909333229064944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,16384,4.762999471028646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,16384,4.354867045084635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,16384,0.12468907038370769
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,16384,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,65536,0.5292714754740397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,16384,0.062054399649302164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,12288,0.09550506273905436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,12288,3.4606763203938806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,12288,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,12288,0.04782079855600993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,12288,9.046800740559895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,16384,11.696158854166667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,10240,0.08560640017191569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,10240,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,8192,0.06847146352132162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,10240,0.046011734008789065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,12288,2.594338226318359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,65536,15.79208984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,8192,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,65536,1.3076821645100911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,8192,0.05533013343811035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,7168,0.16168959935506183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,7168,0.012663466731707254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,65536,19.188981119791666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,10240,2.86747309366862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,10240,7.416422526041667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,7168,0.02955946723620097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,6144,0.06905173460642497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,10240,2.077081553141276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,6144,0.010069333513577779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,6144,0.026555732885996504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,5120,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,5120,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,8192,1.5436458587646484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,8192,2.193032582600911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,4096,0.05218986670176188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,8192,5.86792958577474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,4096,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,4096,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,3584,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,7168,1.2423508961995444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,7168,1.933687464396159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,3584,0.04774826765060425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,3584,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,3072,0.04235946734746297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,7168,5.629917907714844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,3072,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,6144,4.368247477213542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,6144,1.0454357147216797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,6144,1.6010922749837238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,5120,0.08331946531931558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,2560,0.038701868057250975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,3072,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,2560,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,5120,0.9095850626627604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,5120,1.306555684407552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,2048,0.033587201436360674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,2560,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,2048,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,5120,3.638886515299479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,2048,0.01197760005791982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,4096,1.0715819040934245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,1536,0.012117333213488261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,4096,0.6779221216837565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,1536,0.010922666390736897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,1024,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,4096,2.8906155904134114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,3584,2.467737579345703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,1024,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,3584,0.5584896087646485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,3584,0.9091413497924805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,1024,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,768,0.022937599817911783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,768,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,3072,0.7879679997762044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,512,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,768,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,512,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,65536,47.03487955729167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,3072,0.47697919209798173
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,3072,2.1703338623046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,512,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,2560,0.657476298014323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,1536,0.02908160090446472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,2560,1.6589141845703126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,2560,0.425164794921875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,256,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,256,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,256,0.018602667252222697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,2048,1.0289493560791017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,128,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,2048,0.5093023935953777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,128,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,128,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,2048,0.36505600611368816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,64,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,1536,0.3913045247395833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,1536,0.6005418777465821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,1536,0.24449706077575684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,64,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,1024,0.461243724822998
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,256,32,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,1024,0.30231892267862953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,1024,0.15786666870117189
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,256,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,768,0.28067839940389
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,256,32,0.005016533533732096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,65536,0.03290453354517619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,65536,0.9787392298380533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,768,0.23050239880879722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,65536,0.25391786893208823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,768,0.14161920547485352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,512,0.20124799410502114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,512,0.16097280184427898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,16384,0.06260053316752115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,16384,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,512,0.12677120367685954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,256,0.13710932731628417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,256,0.06864213148752848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,256,0.08980480035146078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,16384,0.12458666960398357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,12288,0.009724799791971843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,12288,0.09642666975657145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,12288,0.04819519917170207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,10240,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,128,0.0818176031112671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,10240,0.08209066390991211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,128,0.08130559921264649
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,64,0.066594131787618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,128,0.05205333232879639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3584,32,0.0626688003540039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,10240,0.04089173475901286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,8192,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,8192,0.0683690627415975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,64,0.15943679809570313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3584,32,0.07987199624379476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,64,0.04307626485824585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,8192,0.03293866713841756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,7168,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,7168,0.21975040435791016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,7168,0.030139732360839843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3584,32,0.04379306634267171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,6144,0.06928640206654867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,6144,0.026555732885996504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,6144,0.007439999779065449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,5120,0.0602453351020813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,5120,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,4096,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,5120,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,4096,0.052019198735555015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,4096,0.018500266472498576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,16384,3.5140950520833334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,16384,3.9206570943196617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,3584,0.05758293469746908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,3584,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,3584,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,12288,2.8506795247395833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,3072,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,3072,0.05935786565144857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,3072,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,12288,8.78817138671875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,2560,0.03778560161590576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,2560,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,2560,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,16384,11.79122314453125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,2048,0.03413333495457967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,2048,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,12288,2.4042144775390626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,2048,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,65536,13.773380533854166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,1536,0.040447998046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,10240,2.421896616617839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,65536,15.603575642903646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,1536,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,1024,0.025600000222524004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,1536,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,1024,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,10240,1.7160874684651692
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,1024,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,10240,7.220258076985677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,768,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,768,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,768,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,512,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,8192,1.4400511423746745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,8192,1.8480810801188152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,256,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,8192,5.7582249959309895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,512,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,7168,1.5888703664143882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,256,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,512,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,256,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,128,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,7168,1.300377655029297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,7168,5.08238525390625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,128,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,128,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,64,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,6144,4.515601094563802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,6144,1.3538986206054688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,6144,0.9983658472696939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,64,0.0019797333826621372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,64,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,128,32,0.004947199920813243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,128,32,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,128,32,0.0024874667326609296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,5120,1.1212800343831382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,5120,3.6559529622395837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,5120,0.7922688166300456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,65536,0.032255999247233075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,65536,0.9677824020385742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,65536,0.23975253105163574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,4096,2.789137013753255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,16384,0.12458666960398357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,4096,0.5804373423258464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,4096,0.8872959772745768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,16384,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,16384,0.06273706754048666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,12288,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,3584,2.5001983642578125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,3584,0.4890965461730957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,12288,0.19739306767781575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,12288,0.04795733292897542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,3584,0.8416191736857096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,10240,0.08195412953694661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,10240,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,3072,2.013457107543945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,3072,0.652834129333496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,10240,0.04115840196609497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,8192,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,3072,0.4248575846354167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,8192,0.06935893694559733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,65536,46.321834309895834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,8192,0.0335530678431193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,2560,1.6030719757080079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,2560,0.38434133529663084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,2560,0.5439488093058269
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,7168,0.030105600754419964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,6144,0.06939307053883871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,7168,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,7168,0.07758506933848063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,2048,0.9767936070760092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,2048,0.42618878682454425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,6144,0.025838933388392132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,6144,0.007098666826883952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,2048,0.3129685401916504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,5120,0.06038186550140381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,1536,0.5176661491394043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,1024,0.35392853418986003
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,1536,0.33621333440144857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,1536,0.20848639806111655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,5120,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,4096,0.05174613396326701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,5120,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,4096,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,1024,0.22357333501180016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,4096,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,3584,0.0474453330039978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,768,0.2700629234313965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,1024,0.14312106768290203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,3584,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,768,0.19473066329956054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,3584,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,768,0.11728213628133137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,512,0.18746026357014972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,512,0.13666987419128418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,3072,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,512,0.08557226657867431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,3072,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,3072,0.042871467272440594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,2560,0.03874133427937825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,256,0.11936426957448323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,2560,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,2560,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,256,0.06048426628112793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,256,0.07608319918314616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,2048,0.034508800506591795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,128,0.07205546696980794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,2048,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,128,0.04519253174463908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,128,0.045090134938557944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,2048,0.01181013286113739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,64,0.059665067990620935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,1536,0.0289792001247406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,1536,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,64,0.035224533081054686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,1536,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,64,0.03696639935175578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,1024,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,1024,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,3072,32,0.05584213336308798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,1024,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,3072,32,0.03805866638819377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,3072,32,0.02635093331336975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,768,0.022835199038187662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,768,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,768,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,512,0.020680532852808634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,512,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,512,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,256,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,256,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,256,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,128,0.02775040070215861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,128,0.0022869333624839783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,16384,3.3643519083658853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,128,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,16384,3.3078954060872396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,12288,2.4168446858723955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,64,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,64,0.0019797333826621372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,32,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,64,32,0.016756266355514526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,64,64,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,64,32,0.0019456000377734502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,16384,11.421661376953125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,12288,8.434005228678386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,65536,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,16384,0.12339200178782146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,65536,0.24026452700297035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,65536,14.462361653645834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,65536,2.741350301106771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,12288,2.106709289550781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,65536,16.06417032877604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,16384,0.06386026541392008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,16384,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,10240,1.97205327351888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,12288,0.20363945960998536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,12288,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,12288,0.04747946659723918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,10240,0.08140799999237061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,10240,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,10240,0.04109653234481812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,10240,1.6142336527506511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,10240,7.16550801595052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,8192,0.17933653195699056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,8192,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,8192,1.7070762634277343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,8192,1.2783616383870444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,8192,0.03290453354517619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,7168,0.07806293169657388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,8192,5.621965026855468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,7168,1.2910592397054037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,7168,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,7168,0.03007146716117859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,6144,0.08229546546936035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,6144,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,7168,5.080302937825521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,7168,0.998570696512858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,5120,0.06038186550140381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,6144,0.02645333409309387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,6144,1.0916180928548178
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,5120,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,6144,0.8824149449666342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,5120,0.02286506692568461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,4096,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,6144,4.25000966389974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,4096,0.0506880005200704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,3584,0.048162134488423665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,4096,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,5120,0.9544021606445312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,3584,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,5120,0.7656789143880208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,3584,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,5120,3.485047403971354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,3072,0.04174506664276123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,3072,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,4096,0.7256064097086589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,4096,0.5268479983011882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,3072,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,4096,2.903722635904948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,2560,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,2560,0.03877546787261963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,2560,0.013585066795349121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,2048,0.03420159816741943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,3584,2.387046305338542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,3584,0.6171989440917969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,2048,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,3584,0.4436309178670247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,65536,45.8739013671875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,2048,0.013140267133712769
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,1536,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,3072,2.065339787801107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,1536,0.029730133215586346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,3072,0.3745792071024577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,1536,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,3072,0.53066987991333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,1024,0.025116799275080363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,1024,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,1024,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,2560,1.5626922607421876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,768,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,2560,0.33491627375284827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,2560,0.4301141421000163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,2048,0.9460351943969727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,768,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,2048,0.35297279357910155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,768,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,512,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,2048,0.24634025891621908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,512,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,1536,0.4761941274007161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,256,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,1536,0.2680490811665853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,512,0.006378666559855144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,256,0.01843093236287435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,256,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,1536,0.19541333516438802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,1024,0.3266559918721517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,128,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,128,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,1024,0.18213334083557128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,1024,0.1285802682240804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,768,0.6447104136149089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,64,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,768,0.14390613238016764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,128,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,64,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,768,0.20316160519917809
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,64,0.004878933231035868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,256,32,32,0.01658453345298767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,512,0.17472853660583496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,256,32,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,512,0.11683839956919354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,256,32,32,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,256,0.05300693511962891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,256,0.06949546337127685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,512,0.07635626792907715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,256,0.18339840571085614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,128,0.06498986482620239
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,128,0.034645334879557295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,128,0.03887786865234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,16384,1.7302186330159504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,64,0.053657599290212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,64,0.04215466578801473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2560,32,0.05120000044504801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,64,0.03338239987691243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,16384,2.96994145711263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2560,32,0.03368959824244182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,12288,3.0014806111653645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2560,32,0.037205334504445395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,12288,2.5313621520996095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,10240,1.783944574991862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,10240,2.689809163411458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,16384,7.571729024251302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,12288,8.19623260498047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,10240,1.146231460571289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,16384,2.844091796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,16384,3.752823384602865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,8192,2.762581380208333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,8192,2.0927146911621093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,7168,2.280072530110677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,12288,3.8816393534342444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,7168,1.905766423543294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,8192,2.5700352986653643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,7168,4.033262888590494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,6144,1.396326446533203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,6144,2.0956501007080077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,12288,10.947720336914063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,12288,1.8458965301513672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,6144,1.6625663757324218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,5120,0.6906197230021159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,65536,16.070110066731772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,5120,1.1353429158528647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,5120,1.5071914672851563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,16384,15.352248128255209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,10240,2.040456517537435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,4096,2.3110313415527344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,4096,0.5694122950236002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,65536,15.565653483072916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,10240,9.31048075358073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,4096,0.845038922627767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,10240,1.3390165964762368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,3584,1.1741525014241536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,3584,2.0590250651041666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,3584,0.4339712142944336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,3072,0.433732255299886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,8192,1.2405418395996093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,8192,1.0783743540445963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,3072,0.6510592142740885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,3072,0.6936576207478841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,2560,0.37942612965901695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,7168,1.2524543762207032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,7168,1.4804650624593099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,2560,1.1631956736246745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,8192,8.241117858886719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,7168,4.736443583170573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,6144,4.063300323486328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,2048,0.6639616012573242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,2560,0.7384746551513672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,2048,0.27275946935017903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,6144,0.7857152303059896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,1536,0.2041162649790446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,2048,0.3079498608907064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,1536,0.3676159858703613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,1536,0.4698794682820638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,6144,1.3551615397135417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,1024,0.11011412938435872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,5120,0.6432426452636719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,5120,0.7188138961791992
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,5120,3.3549652099609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,1024,0.19445759455362957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,4096,0.794214375813802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,1024,0.722056516011556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,768,0.09000746409098306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,4096,0.48680960337320967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,768,0.17203200658162435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,768,0.2045952002207438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,512,0.06867626508076986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,4096,2.848221842447917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,512,0.09349119663238525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,512,0.0731818675994873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,3584,0.5079349199930827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,3584,2.4410113016764323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,3584,0.3869695981343587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,256,0.05191680192947388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,256,0.047650134563446044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,3072,0.3230698585510254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,3072,0.43595094680786134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,256,0.04836693207422892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,128,0.04348586797714234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,3072,2.167125447591146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,128,0.03310933311780294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,128,0.03915093342463176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,64,0.03932160139083862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,64,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,2560,1.5193770090738932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,2560,0.3483989397684733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,2560,0.5123071988423665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,64,0.03805866638819377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,2048,0.9071957270304362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,65536,32,0.03894613186518352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,2048,0.19537919362386066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,65536,32,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,65536,32,0.037478399276733396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,2048,0.4076544125874837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,1536,0.5483519872029622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,1536,0.21940906842549643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,1024,0.2653525352478027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,1536,0.3107839902242025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,65536,54.81857503255209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,16384,0.6191786448160808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,1024,0.1868117332458496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,16384,0.4562944094340007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,65536,2.1332650502522785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,1024,0.14762667020161946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,65536,3.925299326578776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,768,0.5111466725667317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,16384,1.2670293172200522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,768,0.11912533442179363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,12288,0.7590549468994141
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,65536,2.356121571858724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,12288,0.47827625274658203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,768,0.08506027062733969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,512,0.06297599871953329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,512,0.16551253000895183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,512,0.0884394645690918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,256,0.08031573295593261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,12288,0.37023998896280924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,10240,0.3995306650797526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,256,0.05785599946975708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,256,0.047991466522216794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,10240,0.5669546763102213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,10240,0.30481065114339195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,8192,0.3264512062072754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,128,0.0579584002494812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,8192,0.2389674663543701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,8192,0.4938709259033203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,7168,0.29559466044108074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,128,0.03242666721343994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,64,0.027170133590698243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,64,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,128,0.03242666721343994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,64,0.048469332853953044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,2048,32,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,7168,0.21647359530131022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,2048,32,0.04369066556294759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,6144,0.2525866667429606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,6144,0.155238405863444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,7168,0.6686037063598633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,2048,32,0.02802346746126811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,6144,0.3521877288818359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,5120,0.3232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,5120,0.09659732977549235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,5120,0.27962026596069334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,4096,0.10007893244425456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,3584,0.21947733561197916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,4096,0.48868694305419924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,3584,0.4651349385579427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,4096,0.07714133262634278
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,16384,2.2679893493652346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,16384,2.7920384724934895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,3072,0.3494229316711426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,3584,0.06683306694030762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,12288,1.3541717529296875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,3072,0.36584107081095374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,3072,0.05867520173390707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,2560,0.09120426972707113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,2560,0.04969813426335652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,2048,0.06154239972432455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,2560,0.1589248021443685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,2048,0.15667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,12288,1.8558976491292316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,65536,10.07633056640625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,2048,0.06597973505655924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,65536,10.436368815104167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,12288,8.184012858072917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,1536,0.046830932299296066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,10240,1.1100480397542318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,16384,11.324004109700521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,1536,0.057070934772491456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,1536,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,1024,0.03815679947535197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,1024,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,1024,0.03973120053609212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,10240,6.780177307128906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,768,0.03259733319282532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,768,0.035396265983581546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,10240,1.1171839396158854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,768,0.02198186715443929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,8192,0.906069310506185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,512,0.028706133365631104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,8192,0.9217365264892579
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,7168,0.7694677352905274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,512,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,512,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,256,0.024098134040832518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,256,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,8192,5.45798594156901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,256,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,128,0.02228906750679016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,7168,4.731972249348958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,7168,0.7994368235270183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,128,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,128,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,6144,0.6397951761881511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,6144,0.6514346440633138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,64,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,6144,3.968852233886719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,64,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,64,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,16384,32,0.020138667027155558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,16384,32,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,16384,32,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,5120,3.263248952229818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,5120,0.5522080103556315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,5120,0.5255509376525879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,4096,2.5619455973307295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,4096,0.37423785527547204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,4096,0.4281002680460612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,65536,1.6615423838297527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,3584,2.69844487508138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,16384,0.5084159851074219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,16384,0.40942932764689133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,65536,1.8909525553385418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,12288,0.3931136131286621
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,16384,1.0624341328938802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,65536,2.7887616475423176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,3584,0.6162773132324219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,3584,0.30310398737589517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,12288,1.272934341430664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,3072,1.8562389373779298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,3072,0.30870186487833656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,3072,0.22865920066833495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,2560,1.481181844075521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,12288,0.28897279103597007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,2560,0.2536448001861572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,10240,0.24944639205932617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,8192,0.37379414240519204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,8192,0.2558293342590332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,10240,0.3262122790018717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,10240,0.46035626729329426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,65536,44.54806315104167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,8192,0.2032298723856608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,2560,0.34505386352539064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,2048,0.9585322697957357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,2048,0.1426431973775228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,2048,0.21073919932047525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,7168,0.20497066179911294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,1536,0.44881919225056965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,1536,0.11083093484242756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,7168,0.31020374298095704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,1536,0.16032320658365887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,7168,0.13369919459025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,6144,0.13663573265075685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,6144,0.10017600059509277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,5120,0.11383466720581055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,6144,0.2888703982035319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,5120,0.23306239446004234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,1024,0.20118187268575033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,1024,0.28460267384847004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,1024,0.07963306903839111
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,768,0.18298880259195965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,768,0.09693866570790609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,768,0.06601386864980062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,5120,0.08744959831237793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,512,0.13383679389953612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,4096,0.09222826957702637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,4096,0.1497429370880127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,4096,0.06567253271738688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,512,0.067958402633667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,3584,0.0846506675084432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,3584,0.12257280349731445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,3584,0.05932373205820719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,512,0.048264535268147786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,3072,0.07891626358032226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,256,0.039867734909057616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,256,0.07154346307118734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,256,0.034713598092397054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,128,0.026009599367777508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,128,0.05123413403828939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,128,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,3072,0.10318506558736165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,3072,0.051677866776784265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,2560,0.09222826957702637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,2560,0.06427306731541951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,64,0.04225706656773885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,2560,0.04426986773808797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,2048,0.05546666781107584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,2048,0.07099733352661133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,64,0.01966080069541931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,2048,0.0369322657585144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,64,0.03126613299051921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1536,32,0.014916266997655234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1536,32,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1536,32,0.03928426504135132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,1536,0.054306133588155114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,1536,0.0466261347134908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,1536,0.03031040032704671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,1024,0.03682880004247029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,1024,0.03833173513412476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,1024,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,768,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,768,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,16384,1.1802965799967446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,768,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,512,0.027611732482910156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,16384,1.8514602661132813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,512,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,512,0.02222080032030741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,12288,0.8867136001586914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,65536,5.7992192586263025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,256,0.02215253313382467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,65536,7.499397277832031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,256,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,256,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,128,0.019933867454528808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,128,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,12288,7.905928548177083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,128,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,64,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,64,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,12288,1.9735552469889324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,16384,10.292599487304688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,64,0.009004799524943034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,10240,1.896994145711263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,10240,1.8098859151204425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,12288,32,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,12288,32,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,12288,32,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,8192,0.6037504196166992
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,10240,7.10823262532552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,8192,0.7182336171468099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,8192,5.218201700846354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,16384,0.5764095942179362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,65536,1.3250559488932292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,16384,0.4430506706237793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,7168,0.6989824295043945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,16384,0.2889045397440592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,12288,0.3414698600769043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,7168,1.576789347330729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,12288,1.0888533274332681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,65536,3.4299563090006506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,7168,5.85628916422526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,6144,0.44776105880737305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,6144,0.5344597498575847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,12288,0.5344597498575847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,10240,0.27985919316609703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,65536,4.2379605611165365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,10240,0.3578879992167155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,6144,5.080029805501303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,10240,0.18875734011332196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,5120,0.36085761388142906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,5120,0.41666345596313475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,8192,0.5238101323445638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,8192,0.3144362767537435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,5120,4.016332753499349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,8192,0.09806079864501953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,4096,0.271016534169515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,4096,0.2758655865987142
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,4096,2.5172991434733074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,7168,0.08652799924214681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,6144,0.12895572980244954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,7168,0.3741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,7168,0.7197013219197591
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,6144,0.2673664093017578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,3584,0.23476907412211098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,3584,0.19483307202657063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,6144,0.07693653106689453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,5120,0.11018239657084147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,5120,0.21927253405253092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,3584,3.2244394938151038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,3072,1.8062676747639972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,3072,0.20077226956685384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,3072,0.14305280049641927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,5120,0.06263466676076254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,4096,0.09151146411895753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,4096,0.18933760325113932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,4096,0.05021013418833414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,2560,0.16858453750610353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,3584,0.08256853421529134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,2560,1.4275925954182944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,3584,0.0457045316696167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,2560,0.12267519632975261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,3584,0.08611839612325033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,3072,0.07304533322652182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,2048,0.8426154454549154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,2048,0.1400810718536377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,2048,0.1022976001103719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,3072,0.1707690715789795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,1536,0.60808531443278
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,3072,0.04041386842727661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,2560,0.06331733465194703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,65536,52.37275390625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,2560,0.03536213239034017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,1536,0.10888533592224121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,2560,0.0618837316830953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,2048,0.05519359906514486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,2048,0.0475818673769633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,1536,0.07884693145751953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,1024,0.25197226206461587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,2048,0.029252266883850096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,1024,0.07690239747365316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,1024,0.056251732508341465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,768,0.18916692733764648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,1536,0.04608000119527181
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,768,0.07369386355082194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,1536,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,1536,0.03826346794764201
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,1024,0.03642026583353679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,1024,0.02624853253364563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,768,0.04730879863103231
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,512,0.09390079975128174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,1024,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,768,0.031880533695220946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,512,0.05003946622212728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,768,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,512,0.03505493402481079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,256,0.060446933905283606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,256,0.027409066756566364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,256,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,768,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,128,0.04382719993591309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,512,0.02754559914271037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,512,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,256,0.023040000597635904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,512,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,128,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,128,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,256,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,256,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,64,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,64,0.034884266058603924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,128,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,64,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,1024,32,0.03239253362019857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,1024,32,0.010749866565068562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,128,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,1024,32,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,128,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,64,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,64,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,64,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,10240,32,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,10240,32,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,10240,32,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,16384,0.9265834808349609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,16384,1.6066218058268231
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,65536,4.452210998535156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,12288,0.700381851196289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,65536,0.932317860921224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,16384,0.8360618591308594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,16384,0.5211135864257812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,16384,1.228049087524414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,65536,1.396872584025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,12288,0.28351147969563806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,12288,0.7994368235270183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,65536,7.039215087890625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,65536,5.027362060546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,12288,1.1146923065185548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,10240,0.5905066808064778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,16384,11.492214965820313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,12288,9.308091227213541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,12288,0.1772874673207601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,10240,0.889958381652832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,10240,0.2257919947306315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,10240,0.2795178731282552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,10240,0.116974933942159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,8192,0.08775466283162435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,8192,0.11219627062479656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,8192,0.5529258728027344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,7168,0.22459732691446938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,8192,0.6382240295410156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,10240,7.282483418782552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,7168,0.211080535252889
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,7168,0.5578378677368164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,7168,0.4686165491739909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,6144,0.1853098710378011
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,7168,0.08362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,8192,0.7456437428792317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,8192,5.129793294270834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,6144,0.1296725352605184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,6144,0.07096319993336996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,5120,0.09253546396891275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,5120,0.09212586879730225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,7168,5.17215576171875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,5120,0.06004053354263306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,6144,0.47028907140096027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,6144,0.4050944010416667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,4096,0.0740010658899943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,4096,0.07516160011291503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,4096,0.048264535268147786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,3584,0.0606549342473348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,5120,0.33129812876383463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,3584,0.06270293394724527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,5120,0.5644970575968424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,3072,0.054306133588155114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,3584,0.04392960071563721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,6144,3.988514200846354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,5120,3.353497568766276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,3072,0.057139201958974206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,4096,0.2346325397491455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,3072,0.038809601465861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,2560,0.04693333307902019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,2560,0.048742401599884036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,2560,0.0338261326154073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,4096,0.2521087964375814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,4096,2.7597142537434896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,2048,0.04242773453394572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,3584,0.21917014122009276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,2048,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,2048,0.03863893349965413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,3584,0.1457152048746745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,3584,2.1076992034912108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,1536,0.03549866676330567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,3072,1.7770837148030598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,3072,0.19012266794840496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,1536,0.030446932713190718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,1024,0.021741867065429688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,1024,0.030037333567937214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,1536,0.022899200518925987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,3072,0.11622400283813476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,1024,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,2560,1.3954048156738281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,2560,0.16011946996053059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,768,0.027170133590698243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,2560,0.10018133322397868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,768,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,768,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,65536,46.47676595052083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,2048,0.8292693456013998
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,2048,0.08174933592478434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,512,0.024132267634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,512,0.01242453356583913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,512,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,2048,0.13305173714955648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,256,0.020989867051442464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,1536,0.4144810676574707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,256,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,1536,0.09553919633229574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,1536,0.06304426590601603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,256,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,1024,0.23985493977864586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,128,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,128,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,128,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,768,0.17650346755981444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,64,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,64,0.00525546669960022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,1024,0.07338666915893555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,1024,0.04440746704737346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,64,0.007095466554164887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,768,0.04829866488774617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,768,0.037988265355428055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,8192,32,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,8192,32,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,512,0.08577706813812255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,512,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,512,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,8192,32,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,256,0.055637331803639736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,256,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,16384,0.3463850657145182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,65536,0.8638421376546225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,256,0.01986560026804606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,65536,1.6495615641276042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,128,0.03949226538340251
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,128,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,16384,0.408403205871582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,65536,1.4286506652832032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,16384,0.21244586308797203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,64,0.03287039995193482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,128,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,64,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,64,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,12288,0.31662079493204753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,12288,0.5582165400187175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,12288,0.12134400208791096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,768,32,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,10240,0.19831466674804688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,768,32,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,10240,0.2703359921773275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,768,32,0.014472533265749613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,10240,0.1059157371520996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,8192,0.20077226956685384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,8192,0.09973759651184082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,8192,0.0843775987625122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,16384,0.6455637613932292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,7168,0.3158357302347819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,7168,0.11171840031941731
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,6144,0.0997717301050822
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,65536,2.9584383646647137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,7168,0.18810879389444987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,16384,1.3807615915934244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,6144,0.09622186819712321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,6144,0.06615039904912313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,12288,0.4788874626159668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,5120,0.08932373523712159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,65536,5.829939270019532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,5120,0.18418347040812175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,5120,0.054135465621948244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,12288,0.9750527699788412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,4096,0.06529706716537476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,16384,10.346291097005208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,4096,0.04345173438390096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,4096,0.06758399804433188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,12288,7.963033548990886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,3584,0.059869865576426186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,3584,0.05840213298797607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,10240,0.3933183987935384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,3584,0.04113066593805949
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,10240,1.0731178283691407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,3072,0.053998935222625735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,3072,0.050653866926829015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,8192,0.317303466796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,3072,0.03618133465449015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,2560,0.04741119941075643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,2560,0.04123306671778361
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,10240,6.478609212239583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,8192,0.8004949569702149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,2560,0.030818132559458415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,2048,0.04109653234481812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,8192,5.179665120442708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,7168,0.27163305282592776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,2048,0.03566933472951253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,2048,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,7168,0.4800511995951335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,1536,0.03614720106124878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,1536,0.033587201436360674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,7168,4.787814331054688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,1024,0.034577067693074545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,1536,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,6144,0.26419199307759605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,1024,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,6144,0.4294656117757161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,1024,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,768,0.026726400852203368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,6144,3.7155156453450524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,768,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,768,0.014233600099881491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,5120,0.17336319287618002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,5120,0.30757548014322916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,512,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,5120,3.206348673502604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,512,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,4096,0.1372159957885742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,512,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,256,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,256,0.008531199892361958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,4096,0.17517226537068684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,4096,2.458555857340495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,128,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,256,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,128,0.006243200103441874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,3584,0.11506346861521403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,128,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,3584,0.10280959606170655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,64,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,3584,2.8548095703125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,3072,0.1897813320159912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,64,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,3072,1.7458175659179687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,7168,32,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,64,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,7168,32,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,3072,0.08802986939748128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,7168,32,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,65536,41.501765950520834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,2560,1.3652309417724608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,2560,0.09082346757253011
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,2560,0.07488853136698405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,65536,1.1977386474609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,2048,0.07137280305226644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,65536,1.3979989369710286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,2048,0.06167893409729004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,2048,1.7840789794921874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,16384,0.3455317179361979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,65536,1.6543744405110676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,12288,0.20558506647745767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,16384,0.1805301348368327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,16384,0.307370662689209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,1536,0.3913386662801107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,1536,0.054067198435465494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,12288,0.27241811752319334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,1536,0.0475818673769633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,1024,0.23999147415161132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,12288,0.1210367997487386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,1024,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,1024,0.04437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,10240,0.131549866994222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,10240,0.09973759651184082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,10240,0.23094612757364907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,768,0.1712469259897868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,768,0.03505493402481079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,8192,0.09410560131072998
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,8192,0.0851253350575765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,8192,0.14042773246765136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,768,0.028569600979487103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,512,0.08038400014241537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,7168,0.10427733262379964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,7168,0.10304853121439617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,512,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,256,0.05021013418833414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,512,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,256,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,7168,0.08232959906260172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,6144,0.08864426612854004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,6144,0.0892586628595988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,128,0.03648853302001953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,256,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,6144,0.06270293394724527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,5120,0.07615040143330892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,5120,0.07424000104268393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,128,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,64,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,5120,0.051712000370025636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,128,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,64,0.029149866104125975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,4096,0.06396373510360717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,64,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,512,32,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,4096,0.05939199924468994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,4096,0.04215466578801473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,3584,0.05881173213322958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,512,32,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,512,32,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,3584,0.05239466826121012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,3584,0.0392192006111145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,3072,0.053179732958475744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,3072,0.046250665187835695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,3072,0.0340992013613383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,16384,0.47254187266031905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,2560,0.04669440189997355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,2560,0.029661866029103597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,2560,0.03945813179016113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,65536,1.815176518758138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,16384,1.1322709401448567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,2048,0.04164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,2048,0.03778560161590576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,12288,0.3711999893188477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,2048,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,1536,0.034850132465362546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,65536,4.988416035970053
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,1536,0.03020799954732259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,1536,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,1024,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,12288,0.8243200302124023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,1024,0.028535467386245728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,16384,10.564096069335937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,10240,0.304808521270752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,1024,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,12288,7.462809753417969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,768,0.02648746569951375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,768,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,768,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,10240,0.6654634475708008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,512,0.023515733083089192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,512,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,512,0.011497599879900615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,256,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,10240,6.504855346679688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,8192,0.5205333391825359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,8192,0.25248427391052247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,256,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,8192,5.26028798421224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,256,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,7168,0.22043306032816568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,128,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,128,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,128,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,7168,0.40714238484700516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,64,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,64,0.005865600208441416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,64,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,6144,0.33508691787719724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,6144,0.15762772560119628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,6144,32,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,6144,32,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,7168,7.2774312337239575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,6144,32,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,5120,0.08731306393941243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,6144,3.7085525512695314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,5120,0.24661332766215005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,5120,3.044659169514974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,65536,1.0597375869750976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,16384,0.24582826296488441
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,65536,0.7877632141113281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,65536,1.1600213368733725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,16384,0.316484260559082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,16384,0.15947093963623046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,4096,0.07041599750518798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,4096,0.11506026585896809
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,12288,0.1473536014556885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,4096,2.3968427022298178
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,12288,0.546338144938151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,3584,2.0347904205322265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,3584,0.06689813137054443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,3584,0.06734506289164224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,10240,0.11113813718159993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,12288,0.12004693349202473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,10240,0.1060863971710205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,10240,0.5309781392415365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,3072,2.7796480814615885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,8192,0.10021546681722004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,8192,0.09250133037567139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,3072,0.056900266806284586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,3072,0.0592192014058431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,8192,0.12424533367156983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,7168,0.20026027361551918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,2560,1.3323946634928385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,2560,0.050756267706553136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,2560,0.04631893237431844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,65536,41.70198974609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,2048,0.7746218363444011
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,7168,0.08178346951802572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,6144,0.08700586954752604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,7168,0.08096426328023275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,6144,0.07208960056304932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,2048,0.037033601601918535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,6144,0.061166934172312414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,5120,0.07543466885884603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,2048,0.040753066539764404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,5120,0.05519359906514486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,5120,0.05181440114974976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,1536,0.35891199111938477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,1536,0.02935466567675273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,1536,0.031880533695220946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,1024,0.23511039415995277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,1024,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,4096,0.06485333442687988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,1024,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,4096,0.04427093267440796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,3584,0.05874346494674683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,4096,0.04143786827723185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,3584,0.040550398826599124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,768,0.1669119993845622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,512,0.07270399729410806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,3072,0.034747731685638425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,3584,0.03799039920171102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,768,0.019797333081563315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,3072,0.051507198810577394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,768,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,512,0.014643200238545737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,512,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,3072,0.03324586749076843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,256,0.04519253174463908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,2560,0.04754773378372192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,256,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,2560,0.03078719973564148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,2560,0.028706133365631104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,128,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,2048,0.040686933199564616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,2048,0.02379093368848165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,256,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,2048,0.024473599592844644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,1536,0.03519146839777629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,128,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,128,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,64,0.026349866390228273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,1536,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,1536,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,64,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,64,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,1024,0.029149866104125975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,256,32,0.022900267442067464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,256,32,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,1024,0.014472533265749613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,768,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,1024,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,256,32,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,768,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,768,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,512,0.024234666426976522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,512,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,512,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,16384,0.4506965319315593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,256,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,65536,1.7784832000732422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,256,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,16384,1.1313152313232422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,256,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,128,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,128,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,12288,0.3486037254333496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,128,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,64,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,64,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,64,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,65536,7.951803588867188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,5120,32,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,12288,0.8338773091634115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,5120,32,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,10240,0.2678442637125651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,5120,32,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,16384,10.446506754557292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,12288,9.370039876302084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,10240,0.6541311899820964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,65536,0.5300565401713053
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,65536,2.187946573893229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,16384,0.2041855971018473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,16384,0.24053759574890138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,10240,7.992115275065105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,65536,0.892245356241862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,16384,0.08308053016662598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,12288,0.09943040211995444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,8192,0.22067200342814125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,7168,0.2193749268849691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,8192,0.48145065307617185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,12288,0.3520170529683431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,7168,0.39461545944213866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,12288,0.06751573085784912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,10240,0.05751466751098633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,10240,0.09464853604634603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,10240,0.08973653316497802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,8192,5.162666829427083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,7168,4.394561258951823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,8192,0.08014506498972575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,8192,0.046353065967559816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,8192,0.06908586819966635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,6144,0.31614294052124026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,6144,0.09512960116068522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,7168,0.16346453030904134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,7168,0.0681984027226766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,5120,0.056251732508341465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,6144,3.6744875590006507
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,7168,0.042496001720428465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,6144,0.07120213508605958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,6144,0.06294186512629191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,5120,3.009194691975911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,5120,0.23606613477071128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,4096,0.07447893619537353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,6144,0.03758080005645752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,4096,0.04870826800664266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,5120,0.0602453351020813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,5120,0.0527018666267395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,4096,0.051609599590301515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,4096,2.6727060953776043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,5120,0.031231999397277832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,4096,0.04334933360417684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,3584,0.0475818673769633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,4096,0.025628799200057985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,3584,0.04638719956080119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,3584,2.0224683125813803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,3584,0.05925546487172445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,3072,0.03805866638819377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,3584,0.03935573498408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,3072,1.687448501586914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,3072,0.028328533967336016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,3072,0.05102933247884115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,65536,40.22456461588542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,3584,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,3072,0.043656531969706217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,2560,0.03935573498408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,3072,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,2560,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,2560,0.024678399165471397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,2560,1.334988784790039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,2560,0.032631466786066696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,2560,0.04304213523864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,2048,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,2048,0.7725738525390625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,2048,0.0343722661336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,2048,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,2048,0.03543039957682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,2048,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,1536,0.3784021377563477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,1536,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,1536,0.030139732360839843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,1536,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,1024,0.025667200485865276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,1024,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,1536,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,1536,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,1024,0.22258346875508628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,1024,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,1024,0.018634666999181114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,768,0.02379093368848165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,768,0.17728853225708008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,1024,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,768,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,768,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,768,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,768,0.016587733229001363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,512,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,512,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,512,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,512,0.07335253556569418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,256,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,512,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,256,0.04413439830144246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,512,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,256,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,256,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,256,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,256,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,128,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,128,0.031846400101979574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,128,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,128,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,64,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,64,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,128,0.005727999905745188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,64,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,128,0.013206400473912559
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,64,0.004704000055789947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,64,0.025838933388392132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,64,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,4096,32,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,128,32,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,4096,32,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,4096,32,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,128,32,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,128,32,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,16384,0.18616320292154948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,65536,0.4781397183736165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,65536,0.8302591959635416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,16384,0.2179072062174479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,65536,0.8651775995890298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,16384,0.4431872049967448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,16384,0.07891626358032226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,65536,1.7554091135660808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,12288,0.0979967991511027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,12288,0.09048426946004232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,16384,1.1188223520914713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,65536,4.62912851969401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,12288,0.34973014195760094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,12288,0.0632149338722229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,10240,0.05324800014495849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,10240,0.08355840047200522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,10240,0.07970133622487387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,8192,0.06563839912414551
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,8192,0.06840319633483886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,12288,0.8211797078450521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,8192,0.04307626485824585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,7168,0.16011946996053059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,16384,10.189994303385417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,10240,0.2636799971262614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,12288,7.547596740722656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,7168,0.06079146862030029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,10240,0.6937941233317058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,7168,0.03925333420435588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,6144,0.06649173498153686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,6144,0.05447680155436198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,6144,0.03423573176066081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,5120,0.059562667210896814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,8192,0.21562026341756185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,10240,6.3343958536783855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,5120,0.04597760041554769
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,5120,0.02945706645647685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,8192,5.163554382324219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,4096,0.051336534818013514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,7168,0.3916799863179525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,8192,0.5088586807250977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,4096,0.03993599812189738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,7168,0.20469759305318197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,4096,0.023756800095240276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,3584,0.046825599670410153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,3584,0.03573653300603231
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,3584,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,3072,0.04205226500829061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,6144,0.07010986804962158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,7168,4.5473790486653645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,3072,0.025668267409006757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,3072,0.018739199638366698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,6144,0.3363840103149414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,2560,0.03860479990641276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,5120,0.057207465171813965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,6144,3.78787841796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,5120,0.21346987088521324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,2560,0.02211839954058329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,2560,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,2048,0.034167468547821045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,2048,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,5120,3.225258636474609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,2048,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,4096,0.043110398451487224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,1536,0.02986666758855184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,1536,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,4096,0.07041707038879394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,1536,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,4096,2.34246826171875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,3584,0.03874133427937825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,1024,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,3584,0.05812906821568807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,768,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,1024,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,1024,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,3584,2.1871274312337237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,768,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,3072,1.8217301686604817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,512,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,3072,0.03170986572901408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,768,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,512,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,3072,0.04829866488774617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,65536,40.36983642578125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,2560,0.026793599128723145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,256,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,512,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,2560,0.04081919987996419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,256,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,2560,1.430084228515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,256,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,2048,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,128,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,2048,0.7634944279988607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,128,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,128,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,2048,0.033587201436360674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,64,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,64,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,1536,0.3537919998168945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,1536,0.028091732660929365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,1536,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,64,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,1024,0.21234347025553385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,1024,0.018465065956115724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,1024,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3584,32,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3584,32,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3584,32,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,768,0.1693013350168864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,768,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,65536,0.7971839904785156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,65536,0.4297013282775879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,65536,0.7145472208658854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,16384,0.14199466705322267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,512,0.07161173025767008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,768,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,256,0.04440746704737346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,512,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,256,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,16384,0.2016597270965576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,12288,0.08721066315968831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,512,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,16384,0.07133866945902506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,12288,0.09130667050679525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,256,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,12288,0.05642240047454834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,128,0.032017066081364946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,10240,0.07314773400624594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,10240,0.048230401674906415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,64,0.025803732872009277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,10240,0.07645866870880128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,128,0.004537599782148996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,128,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,64,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,8192,0.06202026605606079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,64,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,8192,0.06420480012893677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,64,32,0.022459733486175536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,8192,0.039185067017873124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,7168,0.08058773676554362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,64,32,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,7168,0.053896534442901614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,7168,0.03543039957682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,64,32,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,6144,0.0672426700592041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,6144,0.04925440152486165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,6144,0.03092479904492696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,16384,0.4274175961812337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,5120,0.04116479953130086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,5120,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,5120,0.058538667360941564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,65536,1.7584810892740887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,16384,1.1177300771077474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,4096,0.04969813426335652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,4096,0.03495253324508667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,4096,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,12288,0.35024213790893555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,65536,4.843997701009115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,3584,0.04638719956080119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,3584,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,3584,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,12288,0.8043829600016276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,3072,0.04154026508331299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,3072,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,10240,0.2616650740305583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,3072,0.024778666098912557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,2560,0.03829760154088338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,16384,10.050730387369793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,2560,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,12288,7.586474609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,10240,0.6495914459228516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,2560,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,2048,0.0340992013613383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,10240,6.187281290690104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,2048,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,1536,0.028842665751775104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,2048,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,8192,0.21299200057983397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,1536,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,8192,0.4762965202331543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,1024,0.024268800020217897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,8192,5.232776387532552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,1536,0.01437013347943624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,7168,0.18725546201070148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,1024,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,1024,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,7168,0.38976853688557944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,768,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,768,0.023238400618235268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,7168,4.660770161946615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,768,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,6144,0.05010773340861002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,512,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,512,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,6144,0.324232546488444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,512,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,256,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,256,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,5120,0.03642026583353679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,6144,3.6778666178385415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,5120,0.2125823974609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,5120,3.0284459431966146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,256,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,128,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,4096,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,128,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,128,0.004162133236726125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,64,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,4096,0.06785386403401693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,64,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,3072,32,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,4096,2.934579213460286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,3584,0.02648746569951375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,64,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,3584,0.056012801329294835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,3072,32,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,3584,2.0187136332194013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,3072,32,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,3072,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,3072,1.6905899047851562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,65536,0.6155263900756835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,65536,0.7413077036539714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,16384,0.12011520067850749
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,65536,0.8811861038208008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,3072,0.049420801798502605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,16384,0.09461759726206462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,2560,0.022357332706451415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,65536,41.92293701171875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,2560,0.042973868052164715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,16384,0.06498986482620239
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,12288,0.08785920143127442
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,2560,1.3256703694661458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,12288,0.07420586744944255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,2048,0.7588501612345377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,2048,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,2048,0.035327998797098796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,10240,0.042530135313669844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,12288,0.050312534968058265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,10240,0.054579198360443115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,10240,0.07669760386149088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,1536,0.4059136072794597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,1536,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,8192,0.06471680005391439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,1536,0.02723840077718099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,1024,0.21906773249308267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,8192,0.04481706619262695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,1024,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,7168,0.07536640167236328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,8192,0.03549866676330567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,1024,0.01849493384361267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,768,0.168721071879069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,768,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,7168,0.04679679870605469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,6144,0.06649173498153686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,7168,0.031027199824651082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,512,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,6144,0.04092586835225423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,512,0.07181653181711832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,768,0.017061332861582436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,6144,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,5120,0.05874346494674683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,512,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,5120,0.03652266661326091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,256,0.0443391998608907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,256,0.004845866560935974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,5120,0.023654399315516154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,256,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,4096,0.0500053326288859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,128,0.032017066081364946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,4096,0.02955946723620097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,4096,0.01993173360824585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,128,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,3584,0.04580693244934082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,3584,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,64,0.025663999716440837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,3584,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,64,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,3072,0.04123306671778361
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,64,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8192,32,32,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8192,32,32,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,3072,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,3072,0.021572266022364298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8192,32,32,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,2560,0.045602134863535565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,2560,0.024473599592844644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,2048,0.03293866713841756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,2560,0.014569600423177084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,2048,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,2048,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,1536,0.029730133215586346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,1536,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,1536,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,1024,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,1024,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,768,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,1024,0.025256532430648803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,768,0.022323199113210044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,768,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,512,0.028228267033894854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,512,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,256,0.028228267033894854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,256,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,256,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,512,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,128,0.02611200014750163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,128,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,128,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,64,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,64,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2560,32,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2560,32,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2560,32,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,12288,16.018534342447914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,10240,15.064576212565104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,12288,18.4672607421875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,16384,21.976371256510415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,65536,1.6471040089925129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,65536,0.3565567970275879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,16384,0.3668992042541504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,16384,0.19473066329956054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,12288,0.3168256123860677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,65536,1.1662335713704428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,16384,0.1995434602101644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,12288,0.050892798105875645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,16384,26.6492919921875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,12288,0.04990293184916179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,10240,0.07577599684397379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,10240,0.0630783994992574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,8192,0.0370688001314799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,10240,0.15738879839579265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,8192,0.03717120091120402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,8192,0.19292160669962566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,10240,28.428664143880205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,7168,0.18107733726501465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,7168,0.03601066668828328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,7168,0.050722134113311765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,6144,0.19565226236979166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,6144,0.043144532044728595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,6144,0.04522666533788045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,12288,35.02547607421875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,5120,0.04652373393376668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,5120,0.05758293469746908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,10240,13.520140584309896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,5120,0.027784534295399982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,8192,11.64066162109375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,4096,0.05283840099970499
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,4096,0.03280213276545207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,4096,0.035327998797098796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,8192,11.720737711588542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,3584,0.15230293273925782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,3072,0.052292267481486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,3584,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,3584,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,16384,46.89254557291667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,7168,10.584131876627605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,3072,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,3072,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,2560,0.037956265608469646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,2560,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,2560,0.02362026572227478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,2048,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,2048,0.04201813141504924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,2048,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,8192,22.63900146484375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,1536,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,7168,10.53532206217448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,1536,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,1536,0.03700053294499715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,1024,0.024268800020217897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,1024,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,1024,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,768,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,6144,8.7876953125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,768,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,6144,8.565418497721355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,7168,19.31820271809896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,768,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,512,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,5120,7.2727208455403645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,256,0.025873066981633504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,512,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,512,0.0059690664211908976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,256,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,256,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,128,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,6144,16.44223124186198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,5120,7.228518168131511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,5120,13.370572916666665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,128,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,128,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,64,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,4096,5.782289123535156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,64,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,2048,32,0.022425599892934165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,2048,32,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,64,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,2048,32,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,4096,5.970807393391928
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,3584,5.049377950032552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,65536,0.9001984278361002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,65536,0.6646783828735352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,65536,0.6337535858154297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,16384,0.34423465728759767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,16384,0.06154239972432455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,4096,10.586795043945312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,16384,0.06263466676076254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,3584,5.861410013834635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,12288,0.04149973392486572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,12288,0.09086080392201742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,3072,4.621824137369791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,3584,9.170841471354167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,3072,4.621824137369791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,12288,0.04918613433837891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,2560,3.51290028889974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,3072,8.038604736328125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,10240,0.16138240496317546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,10240,0.03563520113627116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,10240,0.04276906649271647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,8192,0.15602347056070964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,8192,0.03037866751352946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,8192,0.056081068515777585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,7168,0.18937172889709472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,2560,3.828526814778646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,2560,6.564317830403645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,2048,2.8731051127115887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,6144,0.16895999908447265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,7168,0.03399680058161418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,6144,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,7168,0.03078826665878296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,6144,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,2048,2.9342025756835937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,1536,2.1805397033691407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,5120,0.17575252850850423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,5120,0.02177706758181254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,1536,2.2761812845865887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,2048,5.268377685546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,1536,3.986841583251953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,5120,0.0233130673567454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,4096,0.052121599515279136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,4096,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,3584,0.045841066042582194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,4096,0.020138667027155558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,1024,1.4522368113199868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,3584,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,1024,1.7335978190104169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,3584,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,1024,3.1411542256673175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,768,1.1967488606770833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,3072,0.04556800127029419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,512,0.9362431844075522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,768,1.437115732828776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,768,2.4343210856119795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,512,1.6158719380696613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,3072,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,3072,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,2560,0.03850239912668864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,2560,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,2560,0.015252266327540079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,2048,0.03453866640726726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,512,1.1204607645670572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,2048,0.011565867066383361
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,256,0.8453119913736978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,2048,0.011980799833933513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,256,1.0929835001627604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,256,0.9503402709960938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,1536,0.028569600979487103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,128,0.917845344543457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,64,0.9619455973307292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,1536,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,1536,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,128,1.2560384114583334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,128,0.9568255742390951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,1024,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,1024,0.02501973311106364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,1024,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,768,0.02228906750679016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,64,1.07325439453125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,768,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,64,1.0851669311523438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,65536,32,0.9788074493408203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,768,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,512,0.02379093368848165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,65536,32,1.173367436726888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,512,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,512,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,65536,32,1.0720255533854166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,256,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,256,0.018870399395624796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,256,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,128,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,128,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,64,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,128,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,64,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,64,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1536,32,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1536,32,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1536,32,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,16384,5.806284586588541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,65536,0.6252543767293294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,16384,0.20101119677225748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,65536,0.9126229604085287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,65536,1.5545003255208334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,16384,9.44551289876302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,12288,7.081540425618489
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,16384,0.03563520113627116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,16384,0.18885973294576008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,12288,8.292079162597656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,12288,0.029525333642959596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,16384,11.555567423502605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,12288,0.08639146486918131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,12288,0.054681599140167236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,10240,0.026862933238347368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,10240,0.197324800491333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,10240,0.06004053354263306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,8192,0.06382933457692465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,12288,3.946116383870443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,8192,0.032153600454330446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,8192,0.0400383989016215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,7168,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,7168,0.029969066381454468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,7168,0.15796906153361004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,6144,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,6144,0.18012159665425617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,6144,0.0451583981513977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,10240,3.573145548502604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,10240,5.957529703776041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,10240,6.769766235351563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,5120,0.16981333096822102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,8192,4.676642354329427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,5120,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,5120,0.0233130673567454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,4096,0.04993706544240316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,4096,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,4096,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,3584,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,3584,0.04539733330408732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,65536,29.962888590494792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,8192,5.25134531656901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,3584,0.03061760067939758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,8192,2.585497538248698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,3072,0.04498773415883382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,3072,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,3072,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,2560,0.037478399276733396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,7168,2.3394304911295576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,2560,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,7168,3.9642453511555993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,2560,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,7168,4.580829874674479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,2048,0.03334826628367106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,2048,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,65536,39.74772135416667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,1536,0.030242133140563964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,2048,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,1536,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,1536,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,1024,0.026520532369613648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,6144,3.5918164571126305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,6144,2.113467788696289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,6144,3.822723134358724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,65536,45.335689290364584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,1024,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,5120,1.7562624613444011
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,5120,2.814771270751953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,1024,0.011707733074824016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,768,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,768,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,5120,3.1225855509440104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,768,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,512,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,512,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,512,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,4096,2.1016576131184896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,256,0.018598399559656777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,4096,1.3927424112955729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,3584,1.5819435119628906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,4096,2.400221761067708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,256,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,3584,2.070113118489583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,128,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,128,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,128,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,3584,1.3492907206217448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,64,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,256,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,3072,1.2542634328206381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,64,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,3072,1.9035818735758465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,3072,0.9826304117838541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,64,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,1024,32,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,1024,32,0.0020138667275508243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,2560,1.035537083943685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,2560,0.9192448298136393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,2560,1.4512128194173177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,1024,32,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,16384,0.11226452986399334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,2048,0.9627989451090494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,65536,0.6089728037516277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,65536,0.24562346140543617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,2048,1.1486890157063803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,2048,0.756053352355957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,65536,0.45783039728800456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,1536,0.8938837051391602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,1536,0.8609450658162435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,16384,0.030173865954081218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,16384,0.06280533472696939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,12288,0.08690133094787597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,12288,0.04908373355865479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,10240,0.07563947041829427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,1536,0.5954890569051107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,1024,0.45062828063964844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,12288,0.02474453250567118
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,1024,0.6938965479532878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,1024,0.43444372812906906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,10240,0.021329067150751748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,768,0.3549525260925293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,768,0.3640320142110189
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,8192,0.06335146824518839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,768,0.6199978510538737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,512,0.27521705627441406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,8192,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,10240,0.04130133390426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,8192,0.03334826628367106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,7168,0.07447786331176758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,7168,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,7168,0.03007146716117859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,512,0.3732138633728027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,512,0.2896213213602702
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,128,0.21278719902038573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,256,0.23176533381144204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,6144,0.06560426553090414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,256,0.25668160120646155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,256,0.23084373474121095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,6144,0.02696533401807149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,6144,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,128,0.2007040023803711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,5120,0.014230400323867798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,128,0.23551999727884926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,5120,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,5120,0.05659306844075521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,4096,0.04942506551742554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,4096,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,64,0.2510848045349121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,64,0.23582720756530762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,64,0.18855253855387372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,16384,32,0.2459989388783773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,4096,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,3584,0.0443391998608907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,16384,32,0.1904640038808187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,16384,32,0.2627583980560303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,3584,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,3584,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,3072,0.05287253459294637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,3072,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,3072,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,2560,0.037273601690928145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,2560,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,2048,0.03310719927151998
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,2560,0.013892267147699991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,2048,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,2048,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,1536,0.028330665826797486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,16384,4.289877319335938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,1536,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,1024,0.037478399276733396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,1536,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,1024,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,1024,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,12288,6.266947428385416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,12288,5.924591064453125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,768,0.03242666721343994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,768,0.011707733074824016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,16384,8.303684488932292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,768,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,16384,8.330752054850262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,512,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,512,0.020717867215474448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,256,0.018602667252222697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,256,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,512,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,256,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,128,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,128,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,12288,3.139720408121745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,128,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,64,0.0020821332931518556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,64,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,10240,3.0423723856608076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,10240,4.925405883789063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,32,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,768,32,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,10240,5.393510437011718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,768,64,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,65536,21.93489786783854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,8192,4.277282206217448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,768,32,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,65536,0.626585578918457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,65536,0.07669760386149088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,16384,0.11270826657613117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,65536,0.23794347445170083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,8192,3.830779774983724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,12288,0.08727893034617105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,16384,0.02693120042483012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,16384,0.10943146546681722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,8192,1.9964244842529297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,12288,0.05618346532185873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,7168,1.9030698140462239
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,12288,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,10240,0.17056427001953126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,10240,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,7168,3.3659220377604164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,65536,33.56549072265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,7168,3.4855936686197913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,65536,32.9459716796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,6144,3.1300267537434894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,10240,0.05246293147404989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,8192,0.06406826575597127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,8192,0.03502080043156942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,8192,0.013585066795349121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,7168,0.07502506573994955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,6144,2.810777537027995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,6144,1.4229845682779947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,7168,0.030037333567937214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,6144,0.07802879810333252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,7168,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,5120,2.3950335184733076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,6144,0.02611200014750163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,6144,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,5120,1.4018559773763022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,4096,1.1439786275227866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,4096,1.6697685241699218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,5120,0.05908480087916056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,4096,1.8023082733154296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,5120,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,5120,2.169477335611979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,5120,0.02307093342145284
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,4096,0.0501418670018514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,4096,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,3584,1.3597012837727864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,3584,1.4451370239257812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,3584,0.045602134863535565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,3584,0.9056938807169596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,4096,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,3584,0.009553066889444987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,3584,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,3072,1.03481814066569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,3072,0.04140373468399048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,3072,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,3072,0.8126805623372396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,3072,1.2617045084635417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,3072,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,2560,0.8520021438598633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,2560,1.032806396484375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,2560,0.03829760154088338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,2560,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,2048,0.8662687937418619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,2048,0.6857727686564128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,2560,0.6999722798665364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,2048,0.03321173389752706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,2560,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,2048,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,1536,0.03170986572901408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,2048,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,2048,0.567739741007487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,1536,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,1536,0.5871615727742513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,1536,0.6610602696736654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,1536,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,1024,0.35235840479532876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,1536,0.45370025634765626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,1024,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,1024,0.33757867813110354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,1024,0.441924254099528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,768,0.2867487907409668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,1024,0.008259200056393941
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,768,0.022796799739201866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,768,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,768,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,1024,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,512,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,768,0.4038997332255046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,512,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,512,0.22473386128743492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,768,0.2853205362955729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,512,0.27989333470662436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,512,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,512,0.21913599967956543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,256,0.1810090700785319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,256,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,256,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,256,0.15370240211486816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,256,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,256,0.18391040166219075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,128,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,128,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,64,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,128,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,128,0.168448003133138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,128,0.1764693260192871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,128,0.17012054125467937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,64,0.1734997272491455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,64,0.13987840016682943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,64,0.18810879389444987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,512,32,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,64,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,512,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,12288,32,0.17332906723022462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,12288,32,0.16097280184427898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,512,32,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,65536,0.5869226455688477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,65536,0.03809279998143514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,12288,32,0.19486719767252605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,16384,0.19445759455362957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,65536,0.24671573638916017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,16384,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,12288,0.08652799924214681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,16384,0.062293334801991784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,12288,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,10240,0.07563947041829427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,12288,0.04747733275095622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,10240,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,10240,0.06679893334706624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,16384,3.3345194498697914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,8192,0.06352213223775229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,16384,8.114858500162761
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,16384,6.966920471191406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,8192,0.04993706544240316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,8192,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,7168,0.030373332897822063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,7168,0.16855039596557617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,7168,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,12288,5.862638854980469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,12288,5.05340576171875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,6144,0.175820795694987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,6144,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,6144,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,5120,0.16411306063334147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,5120,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,4096,0.04928853511810303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,4096,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,5120,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,12288,2.8759722391764324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,4096,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,10240,2.274303944905599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,10240,4.106001027425131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,10240,4.803242492675781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,65536,19.3726806640625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,3584,0.04427093267440796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,8192,3.7700948079427086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,3584,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,3072,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,3072,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,3584,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,3072,0.04171093304951985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,2560,0.03696639935175578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,2560,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,2560,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,65536,27.233860270182294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,8192,1.7582079569498699
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,2048,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,2048,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,8192,3.2937983194986975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,7168,1.5477760314941407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,2048,0.032290132840474446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,1536,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,1024,0.023586134115854897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,1536,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,1536,0.029525333642959596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,7168,3.3985877990722657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,7168,2.9638997395833333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,6144,2.825420888264974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,1024,0.0056650668382644655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,65536,32.000921630859374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,768,0.022528000672658286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,1024,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,768,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,512,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,6144,1.2052820841471354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,768,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,512,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,5120,1.9269290924072267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,6144,2.384349822998047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,512,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,5120,2.3218175252278646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,5120,1.0754389444986978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,256,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,4096,1.475549825032552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,4096,1.4759253184000651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,256,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,256,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,128,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,4096,0.9110528310139975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,128,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,128,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,64,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,3584,0.747929573059082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,3584,1.2286635080973307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,64,0.005899733304977417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,64,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,3584,1.3129386901855469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,3072,0.9048063913981119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,256,32,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,256,32,0.0020138667275508243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,3072,1.166165288289388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,3072,0.7026005427042643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,2560,0.7582719802856446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,256,32,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,65536,0.5857621510823567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,2560,0.9501354853312174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,65536,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,2560,0.5417301177978515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,65536,0.31546026865641275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,16384,0.11219627062479656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,2048,0.6100309371948243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,12288,0.08673280080159505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,2048,0.7573845545450847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,16384,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,16384,0.062498132387797035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,2048,0.48452266057332355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,12288,0.04727466503779094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,1536,0.4610730806986491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,1536,0.45943466822306317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,1536,0.5987317403157552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,12288,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,1024,0.40399786631266277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,1024,0.31187626520792644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,10240,0.07546666463216146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,10240,0.04072106679280599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,10240,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,8192,0.06410239934921265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,8192,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,768,0.24832000732421874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,1024,0.27132587432861327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,8192,0.03345066706339518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,7168,0.07342080275217691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,768,0.3325269381205241
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,768,0.20650666554768882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,7168,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,512,0.19401386578877766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,512,0.14052586555480956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,512,0.23162879943847656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,6144,0.06632106701533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,7168,0.029798400402069092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,256,0.15073280334472655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,6144,0.0070250665148099255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,6144,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,5120,0.05829973220825195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,256,0.1339392026265462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,5120,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,256,0.12888747056325275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,128,0.13847893079121906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,5120,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,64,0.14714879989624025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,128,0.11745279630025227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,4096,0.049356798330942794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,128,0.12827306588490803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,4096,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,64,0.11618986924489338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,4096,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,3584,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,64,0.1364309310913086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,3584,0.017132800817489625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,3584,0.04505600134531657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,10240,32,0.1288863976796468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,3072,0.04993706544240316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,10240,32,0.118886399269104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,10240,32,0.1358143965403239
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,3072,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,2560,0.03778560161590576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,3072,0.006311466793219249
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,2560,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,2560,0.013585066795349121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,2048,0.0338261326154073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,2048,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,2048,0.011844266454378765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,1536,0.02805759906768799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,1536,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,16384,2.874163309733073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,1024,0.034645334879557295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,1536,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,1024,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,1024,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,16384,5.471607462565104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,768,0.02693120042483012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,768,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,16384,7.195746358235676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,12288,4.041454823811849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,12288,5.414058430989583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,768,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,512,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,512,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,256,0.019215999046961467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,512,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,256,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,12288,2.307171122233073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,128,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,65536,15.178547159830728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,128,0.0021845333278179167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,10240,1.7146539052327472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,128,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,64,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,64,0.0019797333826621372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,10240,4.395653279622396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,10240,3.366980234781901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,128,32,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,8192,3.658888498942057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,64,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,128,32,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,128,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,65536,21.56325480143229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,8192,2.4917332967122396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,8192,1.4379348754882812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,65536,0.02481493353843689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,65536,0.24081066449483238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,16384,0.11106986999511718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,65536,0.5790719985961914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,12288,0.09314986864725748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,16384,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,12288,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,16384,0.06229013204574585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,7168,3.130163319905599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,7168,1.1839115142822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,7168,2.2882985432942706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,65536,29.378932698567706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,6144,2.556006368001302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,6144,1.857433573404948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,12288,0.048740267753601074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,10240,0.14144852956136067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,10240,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,6144,0.960750961303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,10240,0.040990932782491045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,5120,2.0084395090738933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,8192,0.06328320105870565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,8192,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,8192,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,7168,0.07321600119272867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,5120,0.8315221150716147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,5120,1.4962346394856771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,7168,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,7168,0.029934932788213093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,4096,1.3201749165852865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,6144,0.06420480012893677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,4096,1.2969983418782554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,4096,0.8425472259521485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,6144,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,3584,1.0784768422444662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,6144,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,3584,1.0029013315836588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,3584,0.5942954381306966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,5120,0.05734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,5120,0.006006399790445963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,5120,0.022797866662343343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,4096,0.005560533205668131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,4096,0.019011199474334717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,4096,0.05003946622212728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,3072,0.7990271886189778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,2560,0.6983338673909505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,3072,0.5511850357055664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,3584,0.04450986782709758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,3072,0.9525589625040689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,3584,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,2560,0.7368021647135417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,2560,0.45250558853149414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,2048,0.522649606068929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,3584,0.017128533124923705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,2048,0.5925546646118164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,3072,0.04109653234481812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,3072,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,2560,0.03805866638819377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,2560,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,3072,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,1536,0.40669867197672527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,2048,0.3747157414754232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,2048,0.03314346671104431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,2560,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,1536,0.2792426745096842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,1536,0.4614144007364909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,1024,0.264738130569458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,1024,0.30955521265665686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,2048,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,1024,0.18691412607828778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,768,0.21531306902567543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,1536,0.028706133365631104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,2048,0.011707733074824016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,1536,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,1536,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,1024,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,768,0.2663424015045166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,1024,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,768,0.14745599428812664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,1024,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,512,0.16616106033325195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,512,0.11543893019358317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,256,0.11622186501820882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,512,0.1854464054107666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,768,0.022357332706451415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,256,0.11530240376790364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,768,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,256,0.07799466451009116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,512,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,512,0.004568533102671305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,768,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,512,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,128,0.0876202662785848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,256,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,128,0.06475093364715576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,256,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,64,0.04754773378372192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,64,0.06826666990915933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,128,0.05748053391774496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,64,0.04764906565348308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,128,0.017132800817489625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,256,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,128,0.0020479999482631685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,8192,32,0.06529599825541178
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,128,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,8192,32,0.04150613149007161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,64,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,64,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,8192,32,0.04798826773961385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,64,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,64,32,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,64,32,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,64,32,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,65536,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,65536,0.5836437225341797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,16384,0.1947648048400879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,65536,0.23627093633015953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,16384,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,16384,2.816477966308594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,16384,0.06137173175811768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,16384,4.890180460611979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,12288,3.5670697530110678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,12288,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,12288,0.17595733006795247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,12288,0.21483519872029624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,10240,0.1958570639292399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,12288,5.260697428385416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,16384,6.9432627360026045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,10240,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,10240,0.040243200461069745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,8192,0.18688000043233235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,8192,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,12288,1.9988138834635418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,7168,0.07389866511027018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,8192,0.03310933311780294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,65536,12.287623087565105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,7168,0.050892798105875645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,7168,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,6144,0.06840319633483886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,6144,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,10240,2.931268310546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,10240,1.6545791625976562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,6144,0.02723840077718099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,5120,0.06782293319702148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,10240,4.2034858703613285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,8192,3.336772155761719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,5120,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,5120,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,4096,0.15677439371744792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,65536,19.689131673177084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,4096,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,4096,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,8192,1.2458975474039713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,8192,2.2603775024414063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,3584,0.044544001420338944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,3584,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,3584,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,7168,1.9144704182942707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,3072,0.040072532494862874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,7168,1.036834144592285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,7168,3.082786051432292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,3072,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,3072,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,2560,0.03717120091120402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,65536,27.847816975911456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,6144,2.5138516743977863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,2560,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,2560,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,6144,1.6120831807454425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,6144,0.8696085611979166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,2048,0.032085333267847696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,2048,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,5120,1.9340618133544922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,2048,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,1536,0.02986666758855184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,1536,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,5120,0.7522645314534505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,5120,1.2626261393229166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,1536,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,4096,1.2360010782877604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,1024,0.0040224000811576845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,768,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,1024,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,1024,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,4096,0.6495552062988281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,4096,1.0164565404256185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,3584,1.0120192209879557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,3584,0.8907434463500976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,3584,0.5592405319213867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,768,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,768,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,512,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,512,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,3072,0.7637333552042643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,3072,0.729702377319336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,3072,0.47165441513061523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,256,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,512,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,256,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,2560,0.6112255732218425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,2560,0.6374368031819662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,128,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,2560,0.3980629285176595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,2048,0.49346561431884767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,128,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,2048,0.505241584777832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,64,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,128,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,2048,0.3071957270304362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,1536,0.36744534174601234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,64,0.0019456000377734502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,1536,0.5077333450317383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,64,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,192,32,32,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,192,32,32,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,192,32,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,1536,0.5097813288370768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,1024,0.2473973274230957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,1024,0.2667861302693685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,1024,0.1573205312093099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,768,0.19602773984273275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,768,0.23193599383036295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,768,0.13230079809824627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,512,0.1491967995961507
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,16384,1.850094985961914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,16384,2.0995413462320967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,512,0.09621760050455729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,512,0.3433471997578939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,12288,1.2967936197916665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,256,0.21118292808532715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,16384,4.036812845865885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,256,0.06574079990386963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,256,0.17605973879496256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,128,0.16701439221700032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,12288,1.549311955769857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,128,0.05420373280843099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,12288,3.2366569519042967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,10240,1.2714324951171876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,128,0.05058559974034628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,64,0.04092586835225423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,32,0.04198400179545085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,64,0.06164480050404867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,7168,64,0.040584532419840495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,7168,32,0.05918720165888468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,7168,32,0.0362496018409729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,10240,5.436108907063803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,10240,1.9424938201904296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,8192,1.2611925760904947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,8192,1.0263210932413735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,8192,2.644753011067708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,7168,0.9192448298136393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,7168,0.8131242752075195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,7168,1.715985107421875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,16384,2.3534591674804686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,12288,2.945672607421875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,6144,1.6966314951578778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,5120,0.6742698669433593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,12288,4.885231018066406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,5120,1.554193115234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,16384,5.891413370768229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,6144,3.3874603271484376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,4096,0.9184586842854818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,5120,2.837811279296875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,6144,1.7217877705891926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,4096,1.2365824381510415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,16384,10.325128173828125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,12288,1.7169748942057292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,4096,0.4798464139302571
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,3584,0.6335829416910808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,65536,13.272165934244791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,3584,0.9481215794881186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,3072,0.376145076751709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,3584,2.000213368733724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,3072,0.413046391805013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,10240,2.9008555094401043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,3072,1.6674474080403645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,10240,5.668317667643229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,10240,3.1348052978515626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,2560,0.7594666798909505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,2560,0.6261760075887044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,8192,3.2322560628255212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,2560,0.5694122950236002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,8192,2.788591003417969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,65536,23.45229695638021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,8192,1.083904012044271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,2048,0.27415892283121746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,2048,0.9641984303792318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,1536,0.3601066589355469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,2048,0.2473301410675049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,1536,0.48148479461669924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,7168,1.3894997914632161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,7168,2.7351722717285156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,7168,1.580953598022461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,1536,0.3642709414164225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,1024,0.10758826732635499
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,1024,0.25398613611857096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,6144,2.343931833902995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,768,0.08110079765319825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,6144,1.3238272349039713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,1024,0.10707626342773438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,6144,0.7825408299763997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,768,0.18653866449991863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,768,0.20872534116109215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,512,0.06628693342208862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,5120,2.2245717366536457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,65536,30.393412272135418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,512,0.07522986729939779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,256,0.049459199110666915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,5120,0.6553248087565104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,5120,1.2156927744547525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,512,0.08502506415049235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,4096,1.1882155100504557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,256,0.04235946734746297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,4096,0.9644373575846353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,256,0.04833279848098755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,4096,0.5360981623331706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,128,0.04198400179545085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,3584,0.9215317408243815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,64,0.03819520076115926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,128,0.03952533404032389
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,128,0.02908160090446472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,64,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,3584,0.7424341201782226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,3584,0.4753749211629231
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,64,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,3072,1.5288661956787108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,3072,0.3990528106689453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,65536,32,0.037649067242940266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,3072,1.62696533203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,65536,32,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,2560,1.3933226267496743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,2560,0.5277354558308919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,65536,32,0.03717120091120402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,2560,0.310749880472819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,2048,0.4135253270467122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,2048,0.44960425694783523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,2048,0.2541226704915365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,1536,0.3406847953796387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,16384,0.9181866963704428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,1536,0.19971413612365724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,1536,0.667579714457194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,16384,0.6095872243245443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,65536,2.2466219584147136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,12288,0.4672831853230794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,16384,0.48769706090291337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,1024,0.5174613316853841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,65536,2.615500895182292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,65536,3.650696563720703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,1024,0.13885439236958821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,12288,0.7562922795613607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,1024,0.21964799563090004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,768,0.18394452730814617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,768,0.1980415980021159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,768,0.11110400358835856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,512,0.13332479794820148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,12288,0.36949332555135095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,512,0.1346560001373291
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,10240,0.3021823883056641
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,512,0.08297812938690186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,10240,0.582378641764323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,10240,0.37966505686442054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,8192,0.28262399037679037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,256,0.08676693439483643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,256,0.07485439777374267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,8192,0.2445653279622396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,8192,0.48779945373535155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,7168,0.2843306541442871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,256,0.05867520173390707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,128,0.04447466532389323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,128,0.064955735206604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,128,0.04252586762110393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,7168,0.3968000094095866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,64,0.05502293507258097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,7168,0.5038079897562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,64,0.03570133447647095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,6144,0.4950357437133789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,64,0.03686399857203166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,6144,32,0.05287040074666342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,5120,0.12803413073221842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,5120,0.277128537495931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,6144,0.6036138534545898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,5120,0.09390079975128174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,6144,32,0.02945386568705241
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,6144,32,0.03604480028152466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,6144,0.22824959754943847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,4096,0.09311573505401612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,4096,0.22838613192240395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,4096,0.07127040227254232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,3584,0.18196479479471844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,3584,0.06580906709035238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,3072,0.18565120697021484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,3584,0.3664896011352539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,3072,0.15011839866638182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,16384,2.399095408121745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,3072,0.3306154568990072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,16384,3.47688954671224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,12288,2.7850069681803387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,16384,6.575410970052083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,2560,0.21189972559611
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,12288,4.6882476806640625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,2560,0.19541333516438802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,2048,0.05645653406778971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,2560,0.0509610652923584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,2048,0.18626559575398763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,65536,9.233305867513021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,2048,0.04160853226979573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,1536,0.04720640182495117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,1536,0.05826239983240763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,12288,1.5191380818684896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,1536,0.05307733217875162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,1024,0.03870720068613688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,1024,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,10240,1.9781631469726562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,1024,0.03874133427937825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,10240,1.3766656239827475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,10240,3.8650538126627603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,768,0.031675734122594196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,768,0.03399680058161418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,65536,16.75717976888021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,768,0.022016000747680665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,8192,3.2512341817220056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,8192,1.5339519500732421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,512,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,512,0.029149866104125975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,8192,0.9055573145548502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,512,0.024268800020217897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,256,0.02379093368848165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,256,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,256,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,7168,1.2630357106526693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,128,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,128,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,7168,0.8260608037312827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,7168,2.663799540201823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,65536,26.669976806640626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,6144,0.7088127772013346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,6144,1.0902186075846354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,6144,2.394043731689453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,128,0.011228799819946289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,64,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,64,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,64,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,16384,32,0.020002132654190062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,5120,1.785582987467448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,16384,32,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,16384,32,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,5120,0.885486920674642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,4096,1.1314858754475912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,5120,0.6060373306274414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,4096,0.48421548207600906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,4096,0.7222613016764323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,3584,0.6065834681193034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,3584,0.9157632191975912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,65536,1.7645226796468099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,3584,0.38341973622639974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,65536,1.494664510091146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,16384,0.4005887985229492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,16384,0.7319210688273112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,16384,0.49404586156209307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,65536,2.7487574259440106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,12288,0.378060785929362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,3072,0.34887679417928064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,3072,0.6007808049519856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,3072,0.5270528157552083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,12288,0.5655221303304037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,2560,0.4734634717305501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,2560,0.42632532119750977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,2560,0.2874709447224935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,2048,0.4177237192789714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,12288,0.3097599983215332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,2048,0.3584341367085775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,10240,0.30764373143514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,10240,0.4192319869995117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,10240,0.24381440480550132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,2048,0.21763413747151694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,8192,0.18305706977844238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,8192,0.3362474759419759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,8192,0.24767146110534669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,1536,0.31187626520792644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,1536,0.2711210568745931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,7168,0.20432213147481285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,1536,0.17667412757873535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,1024,0.20630186398824057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,1024,0.18704959551493328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,1024,0.12062719662984211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,7168,0.296891721089681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,768,0.16817493438720704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,7168,0.11383466720581055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,6144,0.12960426807403563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,768,0.14820693333943685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,6144,0.10393599669138591
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,6144,0.26463573773701987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,768,0.09936213493347168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,5120,0.11182080109914143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,512,0.11574613253275554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,5120,0.21770240465799967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,5120,0.08512852986653646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,512,0.12646186351776123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,512,0.07499093214670817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,256,0.07645866870880128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,256,0.07096319993336996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,4096,0.09014613628387451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,256,0.051609599590301515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,4096,0.14035520553588868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,4096,0.07444480260213217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,128,0.05768533150355021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,3584,0.08198826313018799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,3584,0.10403520266215008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,128,0.03853653271993001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,3584,0.06652586857477824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,128,0.038024532794952395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,3072,0.08830293019612631
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,64,0.049561599890391037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,64,0.045363199710845944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,64,0.032153600454330446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,3072,0.07314773400624594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,3072,0.050892798105875645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,5120,32,0.04734293222427368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,5120,32,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,2560,0.06352213223775229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,2560,0.07420586744944255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,5120,32,0.03242666721343994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,2560,0.04389546712239583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,2048,0.05461333195368449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,2048,0.060279464721679686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,2048,0.036761601765950516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,1536,0.05382826725641886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,1536,0.046214401721954346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,1536,0.028945066531499225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,1024,0.03631786505381267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,1024,0.051780267556508386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,16384,1.7665365854899089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,16384,2.6434900919596354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,1024,0.02532373269399007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,12288,1.8987691243489582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,768,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,16384,6.248243204752604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,768,0.03150506615638733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,768,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,12288,4.53935801188151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,512,0.02263039946556091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,512,0.028433066606521607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,512,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,65536,8.212343343098958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,256,0.023995733261108397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,256,0.013892267147699991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,12288,1.3269332885742187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,256,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,10240,1.5142229715983073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,128,0.034065067768096924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,65536,13.191030883789063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,128,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,10240,3.68541005452474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,128,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,10240,1.1085098266601563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,64,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,64,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,64,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,12288,32,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,12288,32,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,8192,1.9001684824625653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,8192,0.7790250778198242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,12288,32,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,8192,6.310946146647135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,7168,2.52586669921875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,7168,1.013486925760905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,7168,0.6831445058186849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,65536,1.1223040262858073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,6144,2.1329920450846354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,16384,0.5755541483561198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,16384,0.4357461293538411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,16384,0.28446292877197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,6144,0.5337088267008464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,12288,0.4531882603963216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,65536,3.4379094441731772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,6144,1.9111595153808594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,12288,0.7827456156412761
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,5120,0.7088127772013346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,65536,29.29264729817708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,5120,0.49933652877807616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,4096,1.0624682744344076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,65536,3.211639404296875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,12288,0.233949867884318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,5120,4.003123219807943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,4096,0.544324239095052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,10240,0.2671274820963542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,10240,0.1773567994435628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,10240,0.3627680142720541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,4096,0.3594922701517741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,8192,0.18008747100830078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,8192,0.18916692733764648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,8192,0.3189759890238444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,3584,0.4708693186442058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,3584,1.4697813669840494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,3584,0.3146410624186198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,3072,0.5983914693196615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,7168,0.50360533396403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,3072,0.7369728088378906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,3072,0.26801493962605794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,7168,0.17339733441670735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,2560,0.4312405268351237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,7168,0.08861013253529867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,6144,0.1271466652552287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,6144,0.07260159651438394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,2560,0.3434154510498047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,6144,0.21701973279317222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,2560,0.3354933420817057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,5120,0.15663785934448243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,5120,0.10820159912109376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,2048,0.33471145629882815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,2048,0.28549121220906576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,2048,0.17411413192749023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,5120,0.05925546487172445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,1536,0.2830677350362142
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,1536,0.2196821371714274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,1536,0.1396714687347412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,4096,0.08983893394470215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,4096,0.08980480035146078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,3584,0.0808618704477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,4096,0.049356798330942794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,1024,0.14714879989624025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,1024,0.18445653915405275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,3584,0.04485119978586833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,3584,0.08000853061676025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,3072,0.07195306619008382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,768,0.1307637294133504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,1024,0.09936213493347168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,3072,0.07123626867930094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,768,0.11912533442179363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,768,0.08256853421529134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,512,0.09775786399841309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,512,0.09512960116068522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,3072,0.039185067017873124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,2560,0.03413333495457967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,2560,0.06280533472696939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,2560,0.06082559823989868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,512,0.061229864756266274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,2048,0.05420373280843099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,256,0.06662826538085938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,2048,0.048503466447194415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,2048,0.028569600979487103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,1536,0.045533867677052815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,256,0.042837333679199216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,256,0.05819733142852783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,128,0.05140479803085327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,128,0.031880533695220946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,128,0.032529066006342575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,1536,0.03945813179016113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,64,0.04485119978586833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,1536,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,1024,0.035735468069712326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,1024,0.025702399015426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,64,0.029998934268951415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,1024,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,64,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,768,0.03166933258374532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,768,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,768,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,4096,32,0.04194986820220947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,4096,32,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,4096,32,0.026897066831588747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,512,0.027340799570083618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,512,0.015014400084813436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,256,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,256,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,512,0.01269760032494863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,256,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,128,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,128,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,128,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,16384,2.1646336873372394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,16384,1.5607466379801431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,64,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,64,0.020887466271718343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,64,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,10240,32,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,65536,7.314602661132812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,10240,32,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,12288,2.976494852701823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,10240,32,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,12288,5.131568908691406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,16384,10.84081522623698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,65536,11.211605834960938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,12288,1.10295041402181
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,16384,0.36259838740030925
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,65536,2.190540822347005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,16384,0.24064000447591147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,16384,0.969762102762858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,12288,0.8261973063151041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,12288,0.2800640106201172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,65536,2.596387227376302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,10240,1.839581807454427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,10240,2.4508074442545573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,10240,5.908206685384115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,8192,0.6866261164347331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,8192,2.4563028971354166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,8192,2.8595199584960938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,65536,2.7938474019368487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,12288,0.32942078908284506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,7168,3.386845906575521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,10240,0.20660907427469888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,10240,0.27794561386108396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,8192,0.08622079690297445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,8192,0.10226240158081054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,10240,0.19394559860229493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,8192,0.22906986872355142
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,65536,25.243375651041667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,7168,0.594432004292806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,7168,0.987784512837728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,6144,0.85753173828125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,6144,2.067251205444336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,6144,0.5165055910746257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,7168,0.1160533348719279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,5120,1.6559445699055988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,7168,0.07901866436004638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,7168,0.1987925370534261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,6144,0.1240063985188802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,6144,0.06847146352132162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,6144,0.10171519915262858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,5120,0.6977877298990885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,5120,0.09134079615275065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,5120,0.08973653316497802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,5120,0.424345588684082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,4096,0.5480789184570313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,4096,1.017514673868815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,3584,0.4679679870605469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,4096,0.3038837432861328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,3584,0.8235690434773764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,5120,0.05867520173390707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,3584,0.250709327061971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,4096,0.06567253271738688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,4096,0.04676266511281331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,4096,0.07202133337656656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,3584,0.06085973183314005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,3584,0.06683306694030762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,3584,0.04242773453394572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,3072,0.5740885416666667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,3072,0.053350400924682614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,3072,0.22446079254150392
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,3072,0.4029439926147461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,2560,0.4473855972290039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,2560,0.1914197285970052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,2560,0.3433098793029785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,2048,0.320580259958903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,2560,0.04734293222427368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,3072,0.037614933649698895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,3072,0.056251732508341465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,2560,0.04669440189997355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,2048,0.2816341400146484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,2560,0.03280213276545207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,2048,0.04133546749750773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,2048,0.03839999834696452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,2048,0.15564160346984862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,2048,0.02696533401807149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,1536,0.26722987492879235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,1024,0.1707658608754476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,1536,0.22196906407674155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,1536,0.1223680019378662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,1024,0.15209813117980958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,1024,0.08823466300964355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,1536,0.03573760191599528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,1536,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,1024,0.029661866029103597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,1536,0.029896533489227294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,768,0.12260693709055584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,1024,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,1024,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,768,0.02672533392906189
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,768,0.07239680290222168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,768,0.1265664021174113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,768,0.019591466585795084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,512,0.09226240317026774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,512,0.08188587029774984
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,512,0.05420373280843099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,256,0.062054399649302164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,768,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,256,0.04638719956080119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,512,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,512,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,512,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,256,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,256,0.03805866638819377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,256,0.008835200468699138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,256,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,128,0.028568534056345622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,128,0.04836693207422892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,128,0.02733653386433919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,128,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,64,0.04092586835225423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,64,0.022835199038187662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,64,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,128,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,128,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,64,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3584,32,0.038365864753723146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,64,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,8192,32,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,64,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3584,32,0.023552000522613525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,8192,32,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3584,32,0.024098134040832518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,8192,32,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,16384,1.8359636942545574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,16384,1.734485371907552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,65536,2.1687978108723955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,16384,0.20759894053141276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,16384,0.8138069152832031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,12288,0.2481152057647705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,16384,0.9091754913330078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,65536,3.7401257832845056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,65536,2.478557840983073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,12288,2.281915791829427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,12288,0.668398920694987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,65536,7.338222757975261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,16384,7.667234293619792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,12288,4.270011901855469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,12288,1.0565631866455079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,12288,0.12451840241750081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,65536,10.632874552408854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,10240,0.13608959515889485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,8192,0.23913812637329102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,10240,0.24487147331237794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,8192,0.08956586519877116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,10240,0.09983999729156494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,8192,0.17667412757873535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,10240,1.500979232788086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,7168,0.18913280169169108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,10240,4.1103360493977865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,10240,0.9026901245117187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,8192,0.85480105082194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,7168,0.22360746065775552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,8192,0.6024533589680989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,7168,0.08492373625437419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,6144,0.06133759816487631
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,8192,3.460778554280599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,6144,0.09618773460388183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,6144,0.09659732977549235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,5120,0.07830186684926352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,7168,2.7402923583984373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,5120,0.1679360071818034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,7168,0.8721408208211263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,5120,0.05191680192947388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,7168,0.49571841557820634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,65536,24.10301513671875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,4096,0.06399893363316854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,6144,2.014344533284505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,6144,0.6236842473347981
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,3584,0.058265598615010586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,6144,0.43625812530517577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,4096,0.042700799306233723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,4096,0.06761600176493326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,3584,0.03915093342463176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,3584,0.059938132762908936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,5120,1.5934463500976563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,5120,0.33675947189331057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,5120,0.5052757263183594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,3072,0.053452801704406736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,3072,0.05181440114974976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,3072,0.0339626669883728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,4096,0.9754624048868814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,4096,0.401032543182373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,2560,0.04696746667226155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,4096,0.2500607967376709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,2560,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,2560,0.04386133352915446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,3584,0.347545591990153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,3584,0.9420448303222656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,2048,0.03566933472951253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,2048,0.04096000194549561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,3072,0.554973856608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,3584,0.2219007968902588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,3072,0.30491307576497395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,2048,0.025600000222524004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,1536,0.03546453317006429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,1536,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,2560,0.41925973892211915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,3072,0.19281919797261554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,1536,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,1024,0.030242133140563964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,1024,0.019796266158421835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,2560,0.2549077351888021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,2560,0.16517012914021809
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,2048,0.3309568087259928
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,1024,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,768,0.026282666126887004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,2048,0.20848639806111655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,2048,0.1374890645345052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,1536,0.21623466809590658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,512,0.023754666248957314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,768,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,768,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,512,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,512,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,1536,0.10386666456858318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,1536,0.16278187433878583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,256,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,1024,0.11356159845987956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,1024,0.1619935989379883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,256,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,256,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,1024,0.0764245351155599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,768,0.11472213268280029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,768,0.08973653316497802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,128,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,768,0.06294186512629191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,128,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,128,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,64,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,64,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,512,0.1721002737681071
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,512,0.06642346779505412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,64,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,512,0.04720640182495117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,256,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,256,0.056866133213043214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,7168,32,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,7168,32,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,256,0.03884373505910237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,7168,32,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,128,0.04433813492457072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,128,0.025326933463414508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,128,0.025156267484029132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,64,0.03700053294499715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,65536,1.8877098083496093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,16384,0.2943999926249186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,64,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,65536,2.2080169677734376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,16384,0.3433471997578939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,64,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,3072,32,0.03594239950180054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,16384,0.17960747083028156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,3072,32,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,65536,1.7970858256022137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,12288,0.20077226956685384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,12288,0.27013120651245115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,3072,32,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,12288,0.11673599878946941
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,10240,0.2098858674367269
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,10240,0.10004479885101318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,10240,0.21261653900146485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,8192,0.20585813522338867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,7168,0.09768959681193033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,16384,1.578152592976888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,16384,1.6126293182373046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,8192,0.17111040751139323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,8192,0.3192490577697754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,12288,1.0714453379313151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,65536,5.671423848470052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,12288,4.131596883138021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,16384,6.3531005859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,7168,0.20889600118001303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,6144,0.3017045338948568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,7168,0.06925653616587321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,65536,7.1472127278645825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,5120,0.07464853127797445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,6144,0.06017706791559855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,6144,0.09082880020141601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,5120,0.05116586685180664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,5120,0.0736255963643392
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,12288,0.928494962056478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,10240,0.885043207804362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,10240,3.4204330444335938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,4096,0.06219093402226766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,10240,0.7618218739827474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,4096,0.06031359831492106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,4096,0.04160853226979573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,8192,0.7333503723144531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,3584,0.06085973183314005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,3584,0.05379413366317749
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,8192,0.5607423782348633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,3584,0.03799039920171102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,8192,2.7423744201660156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,3072,0.045772798856099445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,3072,0.0519157330195109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,7168,2.330316925048828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,7168,0.7098709106445312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,3072,0.03273386756579082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,2560,0.0466261347134908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,2560,0.04631893237431844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,2560,0.0291157325108846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,6144,0.5939200083414714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,7168,0.4443818728129069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,2048,0.04106239875157674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,6144,1.9582293192545575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,6144,0.3769002596537272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,65536,23.32381795247396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,2048,0.024132267634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,2048,0.0347818652788798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,1536,0.0348469336827596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,5120,1.5493802388509115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,1536,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,1024,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,1536,0.02754559914271037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,5120,0.26555733680725097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,5120,0.48721920649210615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,1024,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,4096,0.9467562357584635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,1024,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,4096,0.21367467244466148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,768,0.026521599292755126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,3584,0.7505919774373372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,768,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,4096,0.4020213445027669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,768,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,3584,0.35095891952514646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,3584,0.1845919926961263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,512,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,512,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,512,0.011878400047620138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,256,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,3072,0.29549226760864256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,3072,0.5108031908671061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,3072,0.16110933621724446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,256,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,2560,0.3974485397338867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,256,0.008631466825803121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,2560,0.25234773953755696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,128,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,128,0.0056618665655454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,2048,0.3125589370727539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,2560,0.31726932525634766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,128,0.01269760032494863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,64,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,64,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,2048,0.20790613492329918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,64,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4096,2560,2048,0.11321600278218587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4096,2560,1536,0.16561493873596192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4096,2560,1536,0.20558506647745767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,6144,32,0.017879466215769448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,6144,32,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,6144,32,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,16384,0.248251740137736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,65536,0.7369696299235027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,16384,0.2945023854573568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,65536,0.95010134379069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,65536,1.1620352427164713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,16384,0.15138133366902667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,12288,0.13458773295084636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,12288,0.22562133471171059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,12288,0.11775999863942463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,10240,0.10728106498718262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,8192,0.0886783997217814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,10240,0.18475947380065919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,10240,0.10417493184407552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,8192,0.09427626927693686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,8192,0.08765439987182617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,7168,0.09703786373138427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,7168,0.06761813163757324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,7168,0.08308053016662598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,6144,0.08471893469492595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,6144,0.07348906993865967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,6144,0.06004053354263306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,5120,0.061713067690531406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,5120,0.07287466526031494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,5120,0.04942506551742554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,4096,0.04997119903564453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,4096,0.06331733465194703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,4096,0.040584532419840495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,3584,0.05785599946975708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,3584,0.03614720106124878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,3584,0.037205334504445395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,3072,0.051370668411254886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,3072,0.03259733319282532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,3072,0.03218773404757182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,2560,0.046284798781077066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,2560,0.028194133440653486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,2560,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,2048,0.040209066867828366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,2048,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,1536,0.03546453317006429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,2048,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,1536,0.01832853356997172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,1536,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,1024,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,1024,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,1024,0.028326400121053058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,768,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,768,0.025770666201909383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,512,0.023108265797297158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,512,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,768,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,512,0.010956799983978272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,256,0.020309333006540933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,256,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,256,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,128,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,128,0.0047082667549451195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,128,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,64,0.004845866560935974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,64,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,64,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,5120,32,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,5120,32,0.00443200021982193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,5120,32,0.006550399959087372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,65536,0.7989248275756836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,65536,0.9325909296671548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,65536,0.49817600250244143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,12288,0.09321386814117431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,16384,0.23729279836018882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,16384,0.08478720188140869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,12288,0.17087146441141765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,8192,0.06331733465194703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,16384,0.1799850622812907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,8192,0.07860906918843588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,10240,0.05618346532185873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,10240,0.07731200059254964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,10240,0.09335467020670572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,12288,0.06512639919916788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,7168,0.041230932871500654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,6144,0.058538667360941564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,8192,0.04471466541290283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,6144,0.035327998797098796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,6144,0.0686079978942871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,7168,0.07652693589528402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,7168,0.06748159726460776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,5120,0.05840213298797607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,5120,0.051774934927622474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,5120,0.03051519989967346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,4096,0.043110398451487224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,4096,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,4096,0.04922026793162028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,3584,0.037956265608469646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,3584,0.04556800127029419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,3584,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,3072,0.04171093304951985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,3072,0.03348480065663655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,3072,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,2560,0.037717334429423016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,2560,0.022562134265899658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,2560,0.017509333292643228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,2048,0.03331306576728821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,2048,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,2048,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,1536,0.03020159999529521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,1536,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,1536,0.012628266215324402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,1024,0.025156267484029132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,1024,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,768,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,1024,0.009796266754468281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,768,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,768,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,512,0.007713066538174947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,512,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,512,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,256,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,256,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,128,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,128,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,256,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,128,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,64,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,64,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,64,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,4096,32,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,4096,32,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,4096,32,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,16384,0.15332694053649903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,65536,0.4710741360982259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,65536,0.8262655893961588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,16384,0.21176320711771646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,65536,0.7449258804321289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,16384,0.07829866409301758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,12288,0.09622186819712321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,12288,0.08441173235575358
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,12288,0.0605183998743693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,10240,0.07423786322275797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,10240,0.0807253360748291
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,10240,0.050312534968058265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,8192,0.062054399649302164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,8192,0.06669653256734212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,8192,0.04174506664276123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,7168,0.05918720165888468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,7168,0.03860479990641276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,6144,0.051643733183542886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,6144,0.064955735206604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,7168,0.07226026852925618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,6144,0.03314346671104431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,5120,0.056524801254272464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,5120,0.04468053181966146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,5120,0.028296534220377607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,4096,0.04997119903564453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,4096,0.03679573138554891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,4096,0.02321066657702128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,3584,0.04539733330408732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,3584,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,3072,0.04242453177769979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,3584,0.03430399894714355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,3072,0.02501973311106364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,3072,0.01884160041809082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,2560,0.037956265608469646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,2560,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,2560,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,2048,0.03310933311780294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,2048,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,2048,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,1536,0.028842665751775104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,1536,0.01181013286113739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,1536,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,1024,0.024951465924580894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,1024,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,768,0.023040000597635904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,1024,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,768,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,768,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,512,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,512,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,512,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,256,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,256,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,256,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,128,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,128,0.0056309332450230915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,64,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,128,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,64,0.0033781332274278007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,64,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3584,32,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3584,32,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3584,32,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,16384,0.12106986840565999
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,65536,0.41434453328450516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,65536,0.7164234797159831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,65536,0.6857727686564128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,16384,0.07051946322123209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,16384,0.16698026657104492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,12288,0.0859818696975708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,12288,0.08253440062204996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,12288,0.05420373280843099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,10240,0.07260159651438394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,10240,0.062054399649302164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,10240,0.046114134788513186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,8192,0.06157653331756592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,8192,0.05198506514231364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,8192,0.037067735195159913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,7168,0.07161173025767008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,7168,0.0529749313990275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,7168,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,6144,0.06498986482620239
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,6144,0.04829546610514323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,6144,0.03020799954732259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,5120,0.040106666088104245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,5120,0.056524801254272464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,5120,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,4096,0.04901546637217204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,4096,0.03389439980189006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,4096,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,3584,0.04485119978586833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,3584,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,3584,0.018972800175348917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,3072,0.024166399240493776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,3072,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,3072,0.04092586835225423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,2560,0.037062398592631024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,2560,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,2560,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,2048,0.03304106593132019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,2048,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,2048,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,1536,0.028706133365631104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,1536,0.010956799983978272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,1024,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,1536,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,1024,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,1024,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,768,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,768,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,768,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,512,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,512,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,512,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,256,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,256,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,256,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,128,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,128,0.005594666798909505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,64,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,64,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,64,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,3072,32,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,3072,32,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,3072,32,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,65536,0.3667968114217123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,65536,0.6028287887573243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,65536,0.6500352223714192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,16384,0.10410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,16384,0.09212586879730225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,16384,0.06345386505126953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,12288,0.08246613343556722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,12288,0.07079253196716309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,12288,0.04911786715189616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,10240,0.07232853571573893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,10240,0.04997119903564453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,10240,0.04174400170644124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,8192,0.04293973445892334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,8192,0.06130346854527792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,8192,0.0340992013613383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,7168,0.0714741309483846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,7168,0.040140799681345624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,7168,0.030853333075841267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,6144,0.06382933457692465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,6144,0.0347818652788798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,6144,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,5120,0.05611519813537598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,5120,0.03037866751352946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,5120,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,4096,0.04962986707687378
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,4096,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,4096,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,3584,0.04468053181966146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,3584,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,3072,0.040994131565093996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,3584,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,3072,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,3072,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,2560,0.037205334504445395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,2560,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,2560,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,2048,0.03290453354517619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,2048,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,1536,0.028535467386245728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,2048,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,1536,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,1536,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,1024,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,1024,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,768,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,1024,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,768,0.007507200042406718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,768,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,512,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,512,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,512,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,256,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,256,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,256,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,128,0.017334399620691936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,128,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,128,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,64,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,64,0.0032032000521818793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2560,32,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,64,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2560,32,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2560,32,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,65536,0.6228309631347656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,65536,0.48971093495686846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,16384,0.10356053511301677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,65536,0.31402241388956703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,16384,0.06314666668574015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,16384,0.06434133450190226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,12288,0.04915200074513753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,12288,0.08235733509063721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,10240,0.0729087988535563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,10240,0.042905600865681966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,12288,0.04836693207422892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,10240,0.04102826515833537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,8192,0.06130346854527792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,8192,0.03648746808369954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,8192,0.03375786542892456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,7168,0.0713045358657837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,7168,0.03293866713841756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,7168,0.03160746693611145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,6144,0.06379520098368327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,6144,0.026658133665720625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,6144,0.02877440055211385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,5120,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,5120,0.05577386617660522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,5120,0.023102933168411256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,4096,0.048401065667470294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,4096,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,3584,0.04474879900614421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,3584,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,4096,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,3584,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,3072,0.040516265233357746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,3072,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,3072,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,2560,0.03778560161590576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,2560,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,2560,0.01402453382809957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,2048,0.03283626635869344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,2048,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,2048,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,1536,0.028569600979487103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,1536,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,1536,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,1024,0.02461013396581014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,1024,0.007707733412583668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,768,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,1024,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,768,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,768,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,512,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,512,0.005595733225345611
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,512,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,256,0.018769067525863648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,256,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,256,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,128,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,128,0.0035103999078273775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,128,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,64,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,64,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,2048,32,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,2048,32,0.0028319999575614927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,2048,32,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,65536,0.5838848114013672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,16384,0.1033898671468099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,65536,0.36601174672444664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,65536,0.2721791903177897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,16384,0.05000213384628296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,16384,0.06280533472696939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,12288,0.040072532494862874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,12288,0.08174933592478434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,12288,0.048059733708699544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,10240,0.07174826463063558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,10240,0.04092586835225423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,10240,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,8192,0.060787200927734375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,8192,0.0291157325108846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,8192,0.0338261326154073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,7168,0.07099733352661133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,7168,0.028051199515660603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,7168,0.030242133140563964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,6144,0.0636245330174764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,6144,0.024098134040832518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,6144,0.0265173335870107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,5120,0.055910400549570714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,5120,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,5120,0.023108265797297158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,4096,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,4096,0.04843519926071167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,4096,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,3584,0.0443391998608907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,3584,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,3584,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,3072,0.040584532419840495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,3072,0.014847999811172486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,3072,0.015871999661127727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,2560,0.03662506739298503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,2560,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,2560,0.013926399747530618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,2048,0.03273386756579082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,2048,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,2048,0.011844266454378765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,1536,0.01242453356583913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,1536,0.028535467386245728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,1536,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,1024,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,1024,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,1024,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,768,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,768,0.007573333382606506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,768,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,512,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,512,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,512,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,256,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,256,0.004365866879622141
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,256,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,128,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,128,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,128,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,64,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,64,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,64,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1536,32,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1536,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1536,32,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,65536,0.23756799697875977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,65536,0.5142186800638835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,65536,0.25442986488342284
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,16384,0.06324906746546427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,16384,0.03508906761805217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,16384,0.10342400074005127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,12288,0.08209066390991211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,12288,0.02805759906768799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,12288,0.048264535268147786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,10240,0.025804799795150758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,10240,0.0717141310373942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,10240,0.04078933397928874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,8192,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,8192,0.03372373183568318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,8192,0.06072320143381754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,7168,0.07109866937001547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,7168,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,7168,0.030100266138712566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,6144,0.017744000752766928
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,6144,0.0634879986445109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,5120,0.05546666781107584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,6144,0.026316799720128375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,5120,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,5120,0.023040000597635904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,4096,0.0491157333056132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,4096,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,4096,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,3584,0.044100264708201095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,3584,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,3584,0.012761599818865457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,3072,0.040243200461069745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,3072,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,3072,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,2560,0.03659093379974365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,2560,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,2048,0.03246080080668132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,2560,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,2048,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,2048,0.01181013286113739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,1536,0.029320534070332843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,1536,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,1536,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,1024,0.02430293361345927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,1024,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,768,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,1024,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,768,0.006040533383687338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,512,0.020753065745035805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,768,0.007541333138942718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,512,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,512,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,256,0.01853440006573995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,256,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,256,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,128,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,128,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,64,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,64,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,64,0.004810666541258494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,1024,32,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,1024,32,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,1024,32,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,65536,0.24135680198669435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,16384,0.10311679840087891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,65536,0.19961172739664715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,65536,0.5051050821940104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,16384,0.028257066011428834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,16384,0.06287360191345215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,12288,0.08205440044403076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,12288,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,12288,0.04802560011545817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,10240,0.07157759666442871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,8192,0.06079146862030029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,10240,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,10240,0.04116479953130086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,8192,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,7168,0.07075839837392171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,8192,0.033787735303243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,7168,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,7168,0.030173865954081218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,6144,0.0634879986445109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,6144,0.014131200313568116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,6144,0.026521599292755126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,5120,0.0553984006245931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,4096,0.04809066851933797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,4096,0.012181333700815837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,5120,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,4096,0.019348265727361043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,3584,0.04485119978586833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,5120,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,3584,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,3584,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,3072,0.04031146764755249
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,2560,0.03635199864705403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,3072,0.015595733126004537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,3072,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,2560,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,2048,0.03256319959958394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,2048,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,2560,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,2048,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,1536,0.028194133440653486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,1536,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,1024,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,1024,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,1024,0.008086400230725606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,1536,0.010102400183677673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,768,0.022425599892934165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,768,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,768,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,512,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,512,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,512,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,256,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,256,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,256,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,128,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,128,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,128,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,64,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,64,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,64,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,768,32,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,768,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,768,32,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,65536,0.07550293604532878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,65536,0.5102933247884114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,16384,0.10321919918060303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,65536,0.236680539449056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,16384,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,16384,0.06301013231277466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,12288,0.08198826313018799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,12288,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,12288,0.04860586722691854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,10240,0.07154346307118734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,10240,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,10240,0.040651734670003256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,8192,0.06150720119476318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,8192,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,7168,0.07075839837392171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,8192,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,7168,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,7168,0.03020799954732259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,6144,0.06331733465194703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,6144,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,6144,0.026180267333984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,5120,0.05526080131530762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,5120,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,4096,0.04788906574249267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,4096,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,4096,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,3584,0.04508800109227498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,3584,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,5120,0.022662399212519328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,3584,0.017474132776260375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,3072,0.04003520011901855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,3072,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,3072,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,2560,0.03607893387476603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,2560,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,2560,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,2048,0.03214933276176453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,2048,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,2048,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,1536,0.02805759906768799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,1536,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,1536,0.010376532872517902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,1024,0.024064000447591147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,1024,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,1024,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,768,0.022323199113210044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,768,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,768,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,512,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,512,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,512,0.006347733239332835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,256,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,256,0.0033386667569478357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,128,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,256,0.0056650668382644655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,128,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,128,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,64,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,64,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,512,32,0.016724266608556113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,512,32,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,512,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,65536,0.03645439942677815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,65536,0.4941823959350586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,16384,0.10301439762115479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,65536,0.2350762685139974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,16384,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,16384,0.061610666910807285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,10240,0.07174826463063558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,12288,0.08202239672342936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,12288,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,10240,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,10240,0.04041386842727661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,12288,0.04747946659723918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,8192,0.06154239972432455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,8192,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,7168,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,7168,0.07069013118743897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,6144,0.06331520080566407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,7168,0.029627732435862225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,8192,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,6144,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,6144,0.02614293297131856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,5120,0.0553984006245931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,5120,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,4096,0.04778666496276855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,4096,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,3584,0.04478293259938558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,5120,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,4096,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,3584,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,3584,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,3072,0.04010453224182129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,3072,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,3072,0.015460266669591268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,2560,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,2560,0.03621546824773152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,2560,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,2048,0.032255999247233075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,2048,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,2048,0.011707733074824016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,1536,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,1536,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,1536,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,1024,0.024268800020217897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,1024,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,1024,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,768,0.02228906750679016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,768,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,768,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,512,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,512,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,512,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,256,0.018225065867106118
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,256,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,256,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,128,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,128,0.002621866762638092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,128,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,64,0.016960000991821288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,64,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,64,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,256,32,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,256,32,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,65536,0.035598933696746826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,65536,0.496674124399821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,256,32,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,65536,0.2349397341410319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,16384,0.10335573355356853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,16384,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,12288,0.08209066390991211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,16384,0.061610666910807285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,12288,0.0474453330039978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,12288,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,10240,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,10240,0.07147520383199056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,10240,0.04041386842727661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,8192,0.0606549342473348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,8192,0.008738133311271667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,7168,0.07154346307118734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,7168,0.008254933357238769
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,7168,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,8192,0.03293866713841756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,6144,0.06314666668574015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,6144,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,6144,0.026180267333984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,5120,0.0553984006245931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,5120,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,5120,0.007165866593519847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,4096,0.04778666496276855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,4096,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,3584,0.04386133352915446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,4096,0.018602667252222697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,3584,0.017303466796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,3584,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,3072,0.04089173475901286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,3072,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,3072,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,2560,0.03601066668828328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,2560,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,2560,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,2048,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,2048,0.005629866818586985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,2048,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,1536,0.02768213351567586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,1536,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,1024,0.024166399240493776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,1536,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,1024,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,768,0.02235413392384847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,1024,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,768,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,768,0.008360532919565837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,512,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,256,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,512,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,256,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,512,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,128,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,128,0.0027637332677841187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,256,0.005454933146635691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,128,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,64,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,64,0.0048767998814582825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,128,32,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,128,32,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,128,32,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,65536,0.020309333006540933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,65536,0.4997802734375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,65536,0.23524692853291831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,16384,0.10284372965494792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,16384,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,16384,0.061713067690531406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,12288,0.00784853349129359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,12288,0.08192000389099122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,12288,0.047441065311431885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,10240,0.0714026689529419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,10240,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,10240,0.04037973483403524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,8192,0.06051733493804932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,8192,0.006820266445477803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,8192,0.03334826628367106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,7168,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,7168,0.07051946322123209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,6144,0.06314666668574015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,7168,0.02955946723620097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,6144,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,6144,0.026180267333984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,5120,0.0552618662516276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,5120,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,5120,0.022459733486175536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,4096,0.0475818673769633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,4096,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,4096,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,3584,0.04361706574757894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,3584,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,3584,0.017029333114624023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,3072,0.03979946772257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,3072,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,2560,0.03621546824773152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,3072,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,2560,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,2560,0.01365226705869039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,2048,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,2048,0.031914667288462324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,2048,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,1536,0.027818665901819868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,1536,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,1024,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,1024,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,1536,0.009858133395512898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,1024,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,768,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,768,0.02208426594734192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,768,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,512,0.020411733786265054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,512,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,512,0.006445866823196411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,256,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,256,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,128,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,256,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,128,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,128,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,64,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,64,0.0026911998788515727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,64,0.004773333172003428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,64,32,0.016860800981521606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,64,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,65536,0.4801173210144043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,65536,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,64,32,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,65536,0.23988906542460123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,16384,0.10280959606170655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,16384,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,16384,0.06290773153305054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,12288,0.0816810687383016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,12288,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,12288,0.048230401674906415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,10240,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,10240,0.04109653234481812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,8192,0.06068906784057617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,10240,0.07144107023874918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,8192,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,8192,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,7168,0.07048532962799073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,7168,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,7168,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,6144,0.06294186512629191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,6144,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,5120,0.05502293507258097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,6144,0.026316799720128375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,5120,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,5120,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,4096,0.04778666496276855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,4096,0.0049792001644770306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,4096,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,3584,0.04369066556294759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,3584,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,3072,0.040072532494862874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,3072,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,3072,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,2560,0.035973334312438966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,3584,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,2560,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,2560,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,2048,0.031948800881703696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,1536,0.027852799495061236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,2048,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,1536,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,2048,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,1024,0.024027733008066814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,1024,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,1024,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,1536,0.009898666540781658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,768,0.02208426594734192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,768,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,768,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,512,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,512,0.006206933160622915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,256,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,512,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,256,0.002963199963172277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,256,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,128,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,128,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,128,0.005525333185990652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,64,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,64,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,160,32,32,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,160,32,32,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,160,32,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,16384,1.5975765228271483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,16384,3.5280896504720056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,16384,1.8259967803955077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,12288,1.2131275177001952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,12288,1.3602442423502603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,12288,2.9565269470214846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,10240,1.135103988647461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,10240,2.247031402587891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,10240,1.073425038655599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,8192,0.8641194661458333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,8192,0.905079460144043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,7168,0.8152405420939127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,8192,1.85350824991862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,7168,0.7620266596476237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,6144,1.4091264088948567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,6144,0.6108501434326172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,6144,0.7010986963907878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,7168,1.5996927897135416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,5120,0.5902677536010742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,5120,0.545962651570638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,5120,1.0560170491536458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,4096,0.8537770589192709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,4096,0.4756479899088542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,3584,0.42093226114908855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,3584,0.828004264831543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,4096,0.4156021436055501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,3072,0.3431040128072103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,3584,0.39546880722045896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,3072,0.7244458516438802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,3072,0.31969280242919923
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,2560,0.2709504127502441
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,2560,0.5509749094645182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,2048,0.4482730547587077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,2560,0.3127978642781576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,2048,0.26610347429911296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,2048,0.22193066279093424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,1536,0.19568640391031902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,1536,0.34508800506591797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,1536,0.17527467409769695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,1024,0.0905898650487264
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,1024,0.0985088030497233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,1024,0.24101546605428062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,768,0.07744853496551514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,768,0.18995200792948405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,768,0.08734719753265381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,512,0.07062186400095621
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,512,0.04471466541290283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,512,0.06813013553619385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,256,0.045363199710845944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,256,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,128,0.039867734909057616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,256,0.04460800091425578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,128,0.038809601465861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,64,0.03863893349965413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,64,0.014233600099881491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,128,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,64,0.037614933649698895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,65536,32,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,65536,32,0.037922132015228274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,65536,32,0.036556800206502274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,16384,0.5689343770345052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,16384,0.9501696268717448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,16384,0.40212478637695315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,65536,3.561198933919271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,65536,2.113086954752604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,65536,1.6004095713297528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,12288,0.4360191980997722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,12288,0.7497045516967773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,12288,0.3358720143636068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,10240,0.3295232137044271
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,10240,0.6002346674601238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,10240,0.2775039990743001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,8192,0.28832426071166994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,8192,0.22343680063883462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,8192,0.482372252146403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,7168,0.18213547070821126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,7168,0.2690389315287272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,7168,0.37232640584309895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,6144,0.21599574089050294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,6144,0.12086613178253174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,6144,0.33058134714762366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,5120,0.1033898671468099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,5120,0.080076797803243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,5120,0.27214508056640624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,4096,0.21951146125793458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,4096,0.0604479988416036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,3584,0.0602453351020813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,4096,0.07611733277638753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,3584,0.19288746515909833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,3584,0.05928959846496582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,3072,0.053179732958475744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,3072,0.12410879929860433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,2560,0.04737600088119507
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,3072,0.0504149317741394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,2560,0.04358826478322347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,2560,0.04566933314005534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,2048,0.04078933397928874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,2048,0.03672746817270915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,1536,0.03543039957682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,2048,0.035396265983581546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,1024,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,1536,0.028569600979487103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,1536,0.029013333717981975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,1024,0.023483733336130776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,1024,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,768,0.026692267258961993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,768,0.01966080069541931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,768,0.019966934124628702
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,512,0.023893332481384276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,512,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,256,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,512,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,256,0.011844266454378765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,256,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,128,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,128,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,64,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,128,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,64,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,16384,32,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,16384,32,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,32,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,16384,64,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,16384,0.46103785832722977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,65536,1.3491541544596353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,65536,1.6855040232340497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,16384,0.6767616271972656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,65536,2.678852335611979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,12288,0.34908161163330076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,16384,0.30641492207845056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,12288,0.5508437474568685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,12288,0.2587989330291748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,10240,0.2652479966481527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,10240,0.41171627044677733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,10240,0.21142187118530273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,8192,0.10253653526306153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,8192,0.22920533816019692
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,8192,0.3227648099263509
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,7168,0.14127786954243976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,7168,0.08782506783803304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,7168,0.31600640614827474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,6144,0.08697173595428467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,6144,0.26026666959126793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,6144,0.0718506654103597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,5120,0.08000853061676025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,5120,0.22920533816019692
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,5120,0.06010666688283285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,4096,0.062225067615509035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,4096,0.09908906618754068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,4096,0.04724053144454956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,3584,0.057070934772491456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,3584,0.05577280124028524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,3584,0.04498773415883382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,3072,0.05120000044504801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,3072,0.04669440189997355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,3072,0.03843413194020589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,2560,0.04645546674728394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,2560,0.03956053256988525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,2048,0.0400383989016215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,2560,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,2048,0.03850239912668864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,1536,0.030446932713190718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,1536,0.034406399726867674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,2048,0.02788693308830261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,1024,0.02887679934501648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,1536,0.02263039946556091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,1024,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,1024,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,768,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,768,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,768,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,512,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,512,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,256,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,256,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,512,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,128,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,256,0.010035199920336406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,128,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,128,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,64,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,64,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,64,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,12288,32,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,12288,32,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,12288,32,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,65536,1.0532864252726237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,16384,0.39915520350138345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,16384,0.5712554931640625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,65536,2.235157267252604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,16384,0.26647893587748206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,65536,1.4587562561035157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,12288,0.3005098660786947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,12288,0.4504917462666829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,12288,0.2237781365712484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,10240,0.22719146410624186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,10240,0.17943894068400065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,10240,0.36536213556925456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,8192,0.08765439987182617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,8192,0.12045653661092122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,8192,0.30818986892700195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,7168,0.30539093017578123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,7168,0.08140693505605062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,7168,0.09997653166453044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,6144,0.0842357317606608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,6144,0.26688852310180666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,6144,0.0667242685953776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,5120,0.156496000289917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,5120,0.07331840197245279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,5120,0.05546666781107584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,4096,0.06181546847025553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,4096,0.04426773389180501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,4096,0.06488746802012126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,3584,0.05649066766103109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,3584,0.05679359833399454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,3584,0.04140373468399048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,3072,0.05089066823323568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,3072,0.05027840137481689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,3072,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,2560,0.04573866526285807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,2560,0.04403200149536133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,2048,0.03996693293253581
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,2560,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,2048,0.03433813254038493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,2048,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,1536,0.03430399894714355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,1536,0.027509333690007527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,1536,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,1024,0.016858667135238647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,1024,0.01925119956334432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,1024,0.028842665751775104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,768,0.025941334168116253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,768,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,768,0.015595733126004537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,512,0.023483733336130776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,512,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,512,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,256,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,256,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,256,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,128,0.018972800175348917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,128,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,128,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,64,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,64,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,64,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,10240,32,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,10240,32,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,10240,32,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,16384,0.3414016087849935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,65536,0.9426944096883139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,65536,1.7910784403483073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,65536,1.2023125966389974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,16384,0.22704960505167643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,16384,0.4392266591389974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,12288,0.3594581286112467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,12288,0.24552106857299805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,12288,0.12861440181732178
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,10240,0.16660480499267577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,10240,0.29487787882486977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,10240,0.08093012968699137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,8192,0.0851967970530192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,8192,0.22719146410624186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,8192,0.061810131867726645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,7168,0.09608533382415771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,7168,0.19288746515909833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,6144,0.07990612983703613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,6144,0.07618559996287028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,7168,0.05997226635615031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,6144,0.04918613433837891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,5120,0.062259201208750406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,5120,0.05092693169911703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,5120,0.042461868127187094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,4096,0.051780267556508386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,4096,0.046011734008789065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,4096,0.03269973397254944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,3584,0.04904959996541341
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,3584,0.04123306671778361
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,3072,0.04375893274943034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,3584,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,3072,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,3072,0.03662506739298503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,2560,0.03979946772257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,2560,0.023859200874964397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,2560,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,2048,0.03389439980189006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,2048,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,2048,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,1536,0.02986133297284444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,1536,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,1536,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,1024,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,1024,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,768,0.023108265797297158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,1024,0.012862933675448099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,768,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,768,0.011400533715883891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,512,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,512,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,256,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,256,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,256,0.007880533238252004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,512,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,128,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,128,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,128,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,64,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,64,0.006448000172773997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,64,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,8192,32,0.016926934321721397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,8192,32,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,8192,32,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,65536,1.0900821685791016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,16384,0.41850881576538085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,65536,0.7576213200887044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,16384,0.19531092643737794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,65536,1.573307673136393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,16384,0.30866772333780923
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,12288,0.2174293359120687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,12288,0.3158357302347819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,12288,0.10212266445159912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,10240,0.09843733310699462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,10240,0.07666347026824952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,10240,0.23811413447062174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,8192,0.07492266496022543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,8192,0.06038186550140381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,8192,0.20391252835591636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,7168,0.08969813187917074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,7168,0.09331626892089843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,7168,0.05768533150355021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,6144,0.07502506573994955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,6144,0.06662826538085938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,6144,0.04948906501134236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,5120,0.06021120150883993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,5120,0.0568992018699646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,4096,0.04966400067011516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,5120,0.04085760116577149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,4096,0.04419840176900228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,4096,0.03078826665878296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,3584,0.04724053144454956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,3584,0.042490665117899576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,3584,0.03031040032704671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,3072,0.04266666571299235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,2560,0.037956265608469646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,2560,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,2560,0.02379093368848165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,3072,0.026658133665720625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,3072,0.03659093379974365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,2048,0.03304106593132019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,2048,0.02218666672706604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,2048,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,1536,0.0289792001247406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,1536,0.01955733299255371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,1536,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,1024,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,1024,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,1024,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,768,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,768,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,512,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,768,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,512,0.008738133311271667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,256,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,512,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,256,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,256,0.006140799820423126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,128,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,128,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,128,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,64,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,7168,32,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,64,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,7168,32,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,7168,32,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,65536,0.7324330647786458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,65536,0.9647786458333334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,65536,1.3518848419189453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,16384,0.2657279968261719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,16384,0.13257386684417724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,16384,0.33935359319051106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,12288,0.11956906318664551
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,12288,0.2746026674906413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,12288,0.09728000164031983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,10240,0.0831829309463501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,10240,0.23377920786539713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,10240,0.0818176031112671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,8192,0.060893865426381436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,8192,0.10953386624654134
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,8192,0.05676373243331909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,7168,0.07376213073730468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,7168,0.0762880007425944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,7168,0.0552618662516276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,6144,0.06550186475118001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,6144,0.05928959846496582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,6144,0.047650134563446044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,5120,0.05737066666285197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,5120,0.039662933349609374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,5120,0.048981332778930665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,4096,0.039867734909057616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,4096,0.04840000073115031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,4096,0.029798400402069092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,3584,0.04505600134531657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,3584,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,3584,0.028228267033894854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,3072,0.04072106679280599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,3072,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,2560,0.028091732660929365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,3072,0.032358400026957196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,2560,0.037478399276733396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,2560,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,2048,0.032085333267847696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,2048,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,2048,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,1536,0.0281877338886261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,1536,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,1024,0.024778666098912557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,1024,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,1536,0.014916266997655234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,1024,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,768,0.02228906750679016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,768,0.010513066252072652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,512,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,768,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,512,0.00904319981733958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,512,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,256,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,256,0.018360533316930137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,256,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,128,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,128,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,128,0.00654720018307368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,64,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,64,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,6144,32,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,6144,32,0.016481066743532814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,6144,32,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,65536,0.8444245020548502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,16384,0.10622293154398602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,65536,1.121139144897461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,16384,0.2227541287740072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,16384,0.2749781290690104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,65536,0.6849194844563802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,12288,0.218451197942098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,12288,0.08239680131276449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,12288,0.09106773535410563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,10240,0.06737919648488364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,10240,0.13526612917582195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,8192,0.057275732358296715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,10240,0.06512639919916788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,8192,0.058097068468729654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,8192,0.053282133738199865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,7168,0.06966613133748373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,7168,0.07307946681976318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,7168,0.049491198857625325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,6144,0.06154239972432455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,6144,0.04177920023600261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,5120,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,6144,0.0413696010907491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,5120,0.05471573273340861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,5120,0.03508373498916626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,4096,0.04734293222427368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,4096,0.030408533414204915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,4096,0.029286400477091475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,3584,0.04355413516362508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,3584,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,3584,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,3072,0.03979946772257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,3072,0.024405332406361897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,2560,0.03580586512883504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,3072,0.027131734291712443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,2560,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,2560,0.023893332481384276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,2048,0.03160746693611145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,2048,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,2048,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,1536,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,1536,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,1536,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,1024,0.024064000447591147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,1024,0.010956799983978272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,1024,0.01159999966621399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,768,0.02228906750679016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,768,0.010032000144322713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,768,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,512,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,512,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,256,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,256,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,512,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,256,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,128,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,128,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,128,0.004261333247025808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,64,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,64,0.003786666691303253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,64,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,5120,32,0.01641493340333303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,5120,32,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,5120,32,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,65536,0.45325654347737626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,65536,0.729088020324707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,16384,0.15377066930135092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,16384,0.22879573504130044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,65536,0.8931306838989258
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,16384,0.05734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,12288,0.15124479929606122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,12288,0.07618559996287028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,12288,0.05034666856129964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,10240,0.06679893334706624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,10240,0.06369280020395915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,10240,0.039867734909057616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,8192,0.05649066766103109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,8192,0.032051199674606325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,8192,0.04860586722691854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,7168,0.06826666990915933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,7168,0.042871467272440594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,7168,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,6144,0.06079146862030029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,6144,0.033928533395131424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,6144,0.02635093331336975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,5120,0.05399466753005981
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,5120,0.0293503999710083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,5120,0.022425599892934165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,4096,0.04642133315404256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,4096,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,4096,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,3584,0.042973868052164715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,3584,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,3584,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,3072,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,3072,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,2560,0.03519146839777629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,3072,0.039082666238149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,2560,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,2560,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,2048,0.031231999397277832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,2048,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,2048,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,1536,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,1024,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,1536,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,1536,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,1024,0.009796266754468281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,1024,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,768,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,768,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,768,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,512,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,512,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,256,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,512,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,256,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,256,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,128,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,128,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,64,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,64,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,4096,32,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,64,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,4096,32,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,4096,32,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,65536,0.6758058547973633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,65536,0.7941802978515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,16384,0.19206825892130536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,16384,0.1094655990600586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,65536,0.4344490687052409
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,16384,0.05242773294448853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,12288,0.07570772965749105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,12288,0.06294186512629191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,12288,0.045499734083811444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,10240,0.06720853646596273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,10240,0.037239468097686766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,10240,0.056012801329294835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,8192,0.056797866026560465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,8192,0.04829759995142619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,8192,0.0291157325108846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,7168,0.06806186834971109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,7168,0.02867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,7168,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,6144,0.06085866689682007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,6144,0.032153600454330446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,6144,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,5120,0.028018132845560713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,5120,0.05376000006993612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,5120,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,4096,0.046830932299296066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,4096,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,4096,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,3584,0.042734932899475095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,3584,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,3584,0.015871999661127727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,3072,0.03901013135910034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,3072,0.02187946637471517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,3072,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,2560,0.03546453317006429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,2560,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,2560,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,2048,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,2048,0.031402667363484696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,1536,0.027272532383600872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,2048,0.010683733224868774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,1536,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,1024,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,1536,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,1024,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,1024,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,768,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,768,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,512,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,512,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,512,0.006517333288987477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,768,0.007303466896216075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,256,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,256,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,256,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,128,0.01669013301531474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,64,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,128,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,128,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,64,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3584,32,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3584,32,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3584,32,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,65536,0.6853631973266602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,65536,0.38174721399943035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,65536,0.6193440119425456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,16384,0.1006282647450765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,16384,0.09622186819712321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,16384,0.04686400095621745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,12288,0.050619733333587644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,12288,0.07574186325073243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,12288,0.0392192006111145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,10240,0.06608213186264038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,10240,0.0423253337542216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,10240,0.03338239987691243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,8192,0.037546666463216145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,8192,0.0562175989151001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,8192,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,7168,0.06799360116322836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,7168,0.03310933311780294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,7168,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,6144,0.06082559823989868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,6144,0.029832533995310467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,6144,0.02173759937286377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,5120,0.05369173288345337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,5120,0.025770666201909383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,5120,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,4096,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,4096,0.014916266997655234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,4096,0.04618240197499593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,3584,0.042530135313669844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,3584,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,3584,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,3072,0.03887786865234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,3072,0.012930132945378623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,3072,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,2560,0.03546453317006429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,2560,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,2560,0.011571199695269267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,2048,0.031027199824651082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,2048,0.013346133629480996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,2048,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,1536,0.027170133590698243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,1536,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,1536,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,1024,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,1024,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,1024,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,768,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,768,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,768,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,512,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,512,0.0060362666845321655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,512,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,256,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,256,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,256,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,128,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,128,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,64,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,128,0.01665173371632894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,64,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,3072,32,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,64,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,3072,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,3072,32,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,65536,0.3227306683858236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,65536,0.5792426427205404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,65536,0.5656576156616211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,16384,0.0950272003809611
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,16384,0.03932160139083862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,16384,0.06143999894460043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,12288,0.07550293604532878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,12288,0.04573866526285807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,12288,0.03317546645800273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,10240,0.06580906709035238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,10240,0.02792106668154399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,8192,0.056012801329294835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,10240,0.04194986820220947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,8192,0.03273386756579082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,8192,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,7168,0.06830080350240073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,7168,0.030105600754419964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,7168,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,6144,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,6144,0.06062080065409342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,5120,0.053725866476694736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,6144,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,5120,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,5120,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,4096,0.046044798692067464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,4096,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,4096,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,3584,0.04238826831181844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,3584,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,3584,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,3072,0.03877546787261963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,3072,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,3072,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,2560,0.034918399651845296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,2560,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,2560,0.0109525332848231
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,2048,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,2048,0.03126613299051921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,2048,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,1536,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,1536,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,1536,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,1024,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,1024,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,1024,0.007030400137106578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,768,0.02198186715443929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,768,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,768,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,512,0.019797333081563315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,512,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,512,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,256,0.018016000588734947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,256,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,256,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,128,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,128,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,128,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,64,0.01627840002377828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,64,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2560,32,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2560,32,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2560,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,65536,0.5581823984781902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,65536,0.266376527150472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,65536,0.45878292719523117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,16384,0.09441280364990234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,16384,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,16384,0.04679679870605469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,12288,0.07546772956848144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,12288,0.03659093379974365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,12288,0.029491200049718218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,10240,0.06563839912414551
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,10240,0.025600000222524004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,8192,0.05587626695632934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,10240,0.032051199674606325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,8192,0.026316799720128375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,8192,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,7168,0.067686398824056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,7168,0.024951465924580894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,7168,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,6144,0.06045013268788656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,6144,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,5120,0.053179732958475744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,6144,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,5120,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,5120,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,4096,0.046114134788513186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,4096,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,3584,0.0421887993812561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,4096,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,3584,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,3584,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,3072,0.03870720068613688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,3072,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,3072,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,2560,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,2560,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,2560,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,2048,0.03078826665878296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,2048,0.00883840024471283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,2048,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,1536,0.027306665976842243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,1536,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,1536,0.013346133629480996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,1024,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,1024,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,1024,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,768,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,768,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,768,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,512,0.0197269340356191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,512,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,512,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,256,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,256,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,128,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,128,0.0036821333070596062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,128,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,64,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,64,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,64,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,2048,32,0.01607039968172709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,2048,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,2048,32,0.004739200075467428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,65536,0.49943892161051434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,65536,0.20445866584777833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,65536,0.34563093185424804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,16384,0.03556693394978841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,16384,0.0945151964823405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,16384,0.037546666463216145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,12288,0.07529813448588053
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,12288,0.030101333061854047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,10240,0.06553599834442139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,12288,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,10240,0.024132267634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,8192,0.02177706758181254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,8192,0.055910400549570714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,10240,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,8192,0.020309333006540933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,7168,0.06727680365244547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,7168,0.020309333006540933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,7168,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,6144,0.06017706791559855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,6144,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,6144,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,5120,0.05311146577199301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,5120,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,5120,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,4096,0.04580693244934082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,4096,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,4096,0.01242453356583913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,3584,0.04215466578801473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,3584,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,3584,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,3072,0.03901439905166626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,3072,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,3072,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,2560,0.034918399651845296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,2560,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,2560,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,2048,0.03092479904492696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,2048,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,2048,0.010338133573532105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,1536,0.027033599217732747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,1536,0.008701866865158081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,1536,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,1024,0.023379200696945192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,1024,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,1024,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,768,0.021673599878946938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,768,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,768,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,512,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,512,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,512,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,256,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,256,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,256,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,128,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,128,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,128,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,64,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,64,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1536,32,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1536,32,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1536,32,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,16384,0.09444586435953775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,65536,0.4459520022074382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,65536,0.1322325309117635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,65536,0.2394048055013021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,16384,0.026282666126887004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,16384,0.03546453317006429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,12288,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,12288,0.07553706963857015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,12288,0.027613866329193115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,10240,0.06566720008850098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,10240,0.018871466318766274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,10240,0.023890133698781332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,8192,0.05625066757202148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,8192,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,8192,0.019933867454528808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,7168,0.06748159726460776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,7168,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,7168,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,6144,0.06017706791559855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,6144,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,6144,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,5120,0.05324800014495849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,5120,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,5120,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,4096,0.045602134863535565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,4096,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,4096,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,3584,0.04215146700541179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,3584,0.010410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,3584,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,3072,0.03829760154088338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,3072,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,3072,0.010376532872517902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,2560,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,2560,0.0347818652788798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,2560,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,2048,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,2048,0.03061760067939758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,2048,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,1536,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,1536,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,1536,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,1024,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,1024,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,1024,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,768,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,768,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,512,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,512,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,768,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,256,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,256,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,256,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,128,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,128,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,64,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,128,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,64,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,64,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,1024,32,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,1024,32,0.0024565334121386213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,1024,32,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,65536,0.18582186698913575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,65536,0.436462942759196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,65536,0.129911470413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,16384,0.09461759726206462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,16384,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,16384,0.03536213239034017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,12288,0.07519573370615641
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,12288,0.027955200274785357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,12288,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,10240,0.06557013193766276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,10240,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,10240,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,8192,0.055978667736053464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,8192,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,8192,0.019797333081563315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,7168,0.06737919648488364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,7168,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,6144,0.06004053354263306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,7168,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,6144,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,6144,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,5120,0.05304319858551025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,5120,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,5120,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,4096,0.045499734083811444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,4096,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,4096,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,3584,0.0421887993812561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,3584,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,3584,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,3072,0.038262399037679036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,3072,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,3072,0.010408533612887065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,2560,0.034577067693074545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,2560,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,2560,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,2048,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,2048,0.030719999472300214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,2048,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,1536,0.02682773272196452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,1536,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,1536,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,1024,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,1024,0.006551466882228851
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,768,0.021640533208847047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,1024,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,768,0.00518506666024526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,768,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,512,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,512,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,256,0.0176746666431427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,512,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,256,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,128,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,256,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,128,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,128,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,64,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,64,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,768,32,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,768,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,64,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,768,32,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,65536,0.04573866526285807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,65536,0.13017919858296711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,65536,0.43287893931070964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,16384,0.09417386849721274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,16384,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,12288,0.07550293604532878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,16384,0.03508906761805217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,12288,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,12288,0.027509333690007527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,10240,0.06584320068359376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,10240,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,10240,0.02379093368848165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,8192,0.0559445341428121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,8192,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,8192,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,7168,0.06737919648488364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,7168,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,7168,0.01808746655782064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,6144,0.059938132762908936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,6144,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,6144,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,5120,0.05307733217875162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,5120,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,5120,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,4096,0.045602134863535565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,4096,0.007946666578451793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,4096,0.012151466806729634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,3584,0.04194986820220947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,3584,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,3584,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,3072,0.03822933435440064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,3072,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,3072,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,2560,0.03453866640726726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,2560,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,2560,0.009348266323407491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,2048,0.03065173427263896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,2048,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,2048,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,1536,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,1536,0.02696533401807149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,1536,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,1024,0.023586134115854897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,1024,0.005596800148487091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,1024,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,768,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,768,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,768,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,512,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,512,0.01966080069541931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,512,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,256,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,256,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,256,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,128,0.016416000326474507
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,128,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,128,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,64,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,64,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,64,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,512,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,512,32,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,512,32,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,65536,0.030646399656931562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,65536,0.4243797302246094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,65536,0.1296725352605184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,16384,0.09461759726206462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,16384,0.011332266529401143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,16384,0.034884266058603924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,12288,0.07557120323181152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,12288,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,12288,0.027613866329193115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,10240,0.06557013193766276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,10240,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,10240,0.023893332481384276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,8192,0.0556714653968811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,8192,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,8192,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,7168,0.06727039813995361
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,7168,0.00999679962793986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,7168,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,6144,0.059903999169667564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,6144,0.007371733089288075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,6144,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,5120,0.0529749313990275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,5120,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,5120,0.014232533176740012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,4096,0.045499734083811444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,4096,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,4096,0.012115200360616047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,3584,0.04198400179545085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,3584,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,3584,0.011332266529401143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,3072,0.03867306709289551
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,3072,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,3072,0.010238933563232421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,2560,0.03444053332010905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,2560,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,2560,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,2048,0.03037866751352946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,2048,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,2048,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,1536,0.026658133665720625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,1536,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,1536,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,1024,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,1024,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,1024,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,768,0.02187946637471517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,768,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,768,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,512,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,512,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,512,0.005525333185990652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,256,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,256,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,256,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,128,0.016377600034077962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,128,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,128,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,64,0.002454400062561035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,64,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,64,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,256,32,0.016140799721082053
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,256,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,256,32,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,65536,0.4118186632792155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,65536,0.02935466567675273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,65536,0.12960426807403563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,16384,0.09417386849721274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,16384,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,16384,0.034986666838328045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,12288,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,12288,0.07635626792907715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,10240,0.06556479930877686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,12288,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,10240,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,10240,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,8192,0.055978667736053464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,8192,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,8192,0.019314134120941163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,7168,0.06754986445109049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,7168,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,7168,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,6144,0.05980159838994344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,6144,0.006961066524187725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,5120,0.05273600021998087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,6144,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,5120,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,5120,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,4096,0.045533867677052815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,4096,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,4096,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,3584,0.04188160101572673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,3584,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,3584,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,3072,0.037956265608469646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,3072,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,3072,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,2560,0.034406399726867674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,2560,0.00934933324654897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,2560,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,2048,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,2048,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,2048,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,1536,0.02693120042483012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,1536,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,1536,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,1024,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,1024,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,1024,0.006346666812896728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,768,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,768,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,768,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,512,0.01952106753985087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,512,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,512,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,256,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,256,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,256,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,128,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,128,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,128,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,64,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,64,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,128,32,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,128,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,128,32,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,65536,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,65536,0.4242090543111165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,16384,0.09410453637440999
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,65536,0.1279647986094157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,16384,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,16384,0.03444053332010905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,12288,0.07512746651967367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,12288,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,10240,0.06546773513158163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,12288,0.02768213351567586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,10240,0.006687999765078227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,10240,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,8192,0.05556586583455404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,8192,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,8192,0.019316265980402626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,7168,0.0672426700592041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,7168,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,7168,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,6144,0.059556265672047935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,6144,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,6144,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,5120,0.05259946584701538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,5120,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,5120,0.014233600099881491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,4096,0.04532906611760457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,4096,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,4096,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,3584,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,3584,0.04167679945627849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,3584,0.010922666390736897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,3072,0.03816106716791789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,3072,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,3072,0.010069333513577779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,2560,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,2560,0.034167468547821045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,2560,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,2048,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,2048,0.004333866635958353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,2048,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,1536,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,1536,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,1536,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,1024,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,1024,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,1024,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,768,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,768,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,768,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,512,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,512,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,512,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,256,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,256,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,128,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,128,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,128,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,64,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,64,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,64,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,64,32,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,64,32,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,64,32,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,65536,0.013034666577974955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,65536,0.42588160832722977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,65536,0.12424320379892986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,16384,0.09417386849721274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,16384,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,16384,0.03430399894714355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,12288,0.07495253086090088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,12288,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,12288,0.02648746569951375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,10240,0.06567253271738688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,10240,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,10240,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,8192,0.0556714653968811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,8192,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,8192,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,7168,0.0667306661605835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,7168,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,7168,0.01744106610616048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,6144,0.059665067990620935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,6144,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,6144,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,5120,0.05273600021998087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,5120,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,5120,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,4096,0.04522026777267456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,4096,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,4096,0.01181013286113739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,3584,0.04143786827723185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,3584,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,3584,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,3072,0.03781973520914714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,3072,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,3072,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,2560,0.034167468547821045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,2560,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,2560,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,2048,0.03061760067939758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,2048,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,2048,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,1536,0.026446932554244997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,1536,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,1536,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,1024,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,1024,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,1024,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,768,0.021297067403793335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,768,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,768,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,512,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,512,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,512,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,256,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,256,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,128,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,256,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,128,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,128,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,64,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,64,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,64,0.004638933142026265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,128,32,32,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,128,32,32,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,128,32,32,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,16384,1.5569578806559243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,16384,1.779916763305664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,12288,1.291089121500651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,12288,1.3383338928222657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,16384,3.5197611490885414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,12288,2.9055999755859374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,10240,1.114999516805013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,10240,2.235528564453125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,10240,1.0651306788126627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,8192,0.7835989634195963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,8192,0.8878421147664388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,8192,1.7398442586263019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,7168,0.7294975916544597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,7168,0.8018592198689779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,7168,1.594436264038086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,6144,0.6905173619588216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,6144,0.5996885299682617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,6144,1.287884775797526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,5120,0.5441535949707031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,5120,1.0453333536783853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,5120,0.5805056254069011
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,4096,0.46690772374471023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,4096,0.9112202962239584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,4096,0.4359850565592448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,3584,0.38734505971272787
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,3584,0.7431498845418294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,3584,0.36113065083821616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,3072,0.33597440719604493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,3072,0.33556480407714845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,3072,0.6484607696533203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,2560,0.2982229232788086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,2560,0.593715222676595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,2560,0.2840575853983561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,2048,0.2391040007273356
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,2048,0.43905385335286456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,2048,0.23237973848978677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,1536,0.3369983990987142
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,1536,0.18855253855387372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,1536,0.16683947245279948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,1024,0.08475306828816732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,1024,0.234222936630249
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,1024,0.07454720338185629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,768,0.07140693664550782
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,768,0.1710762659708659
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,768,0.06727680365244547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,512,0.06433813174565634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,512,0.04498773415883382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,512,0.05079040129979452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,256,0.04485119978586833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,256,0.04307626485824585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,256,0.034508800506591795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,128,0.03911679983139038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,128,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,64,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,128,0.029969066381454468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,64,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,64,0.029013333717981975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,65536,32,0.03737066586812337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,65536,32,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,65536,32,0.028330665826797486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,16384,0.5401599884033204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,65536,1.7491967519124347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,16384,0.43721386591593425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,16384,0.8981162389119467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,12288,0.42072321573893234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,65536,3.5412297566731774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,65536,2.05755729675293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,12288,0.748578135172526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,12288,0.3341994603474935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,10240,0.32064854303995766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,10240,0.26975574493408205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,10240,0.594770113627116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,8192,0.2798538525899251
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,8192,0.22152427037556968
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,8192,0.49100799560546876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,7168,0.2612565358479818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,7168,0.19513707160949706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,7168,0.37847038904825847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,6144,0.22033066749572755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,6144,0.1257813294728597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,6144,0.33481388092041015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,5120,0.09082880020141601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,5120,0.26801493962605794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,5120,0.07328426837921143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,4096,0.06314666668574015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,4096,0.05700266758600871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,4096,0.21719039281209312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,3584,0.057239464918772374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,3584,0.05273600021998087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,3584,0.18971306482950848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,3072,0.05068373282750448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,3072,0.07441066900889079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,3072,0.0451583981513977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,2560,0.045431466897328694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,2560,0.045943466822306316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,2560,0.04140373468399048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,2048,0.03778560161590576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,2048,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,2048,0.03956053256988525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,1536,0.034713598092397054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,1536,0.030339199304580688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,1024,0.02874026695887248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,1536,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,1024,0.021364265680313112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,1024,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,768,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,768,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,768,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,512,0.0233130673567454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,512,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,512,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,256,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,256,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,256,0.011878400047620138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,128,0.01955733299255371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,128,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,128,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,64,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,64,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,64,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,16384,32,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,16384,32,0.018498132626215615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,16384,32,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,65536,1.6373077392578126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,16384,0.43192319869995116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,16384,0.6703445434570312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,16384,0.32877225875854493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,65536,1.1997525533040365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,12288,0.3365546544392904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,65536,2.6608980814615886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,12288,0.5666816075642903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,12288,0.2535765329996745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,10240,0.19196586608886718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,10240,0.27347625096639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,10240,0.4001450538635254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,8192,0.21831679344177246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,8192,0.3233728090922038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,8192,0.10178559621175129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,7168,0.11584853331247966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,7168,0.07850666840871176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,7168,0.29395627975463867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,6144,0.08751786549886068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,6144,0.065774933497111
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,6144,0.2836138725280762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,5120,0.07137280305226644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,5120,0.2139135996500651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,5120,0.05437440077463786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,4096,0.06007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,4096,0.09120426972707113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,4096,0.043724799156188966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,3584,0.05550080140431722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,3584,0.07069013118743897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,3584,0.040140799681345624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,3072,0.049561599890391037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,3072,0.05570559899012247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,3072,0.035293865203857425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,2560,0.04485119978586833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,2560,0.03870720068613688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,2560,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,2048,0.03870720068613688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,2048,0.025634133815765382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,2048,0.031845333178838094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,1536,0.033655468622843424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,1536,0.02945706645647685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,1536,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,1024,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,1024,0.02034133275349935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,1024,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,768,0.025630933046340943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,768,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,768,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,512,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,512,0.0120469331741333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,512,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,256,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,256,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,256,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,128,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,128,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,128,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,64,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,64,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,64,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,12288,32,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,12288,32,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,12288,32,0.008087466657161712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,65536,1.3947903951009115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,16384,0.27825279235839845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,16384,0.3754666646321615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,16384,0.5958656311035156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,12288,0.2888703982035319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,65536,1.0251605351765951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,12288,0.4438015937805176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,65536,2.2150143941243488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,12288,0.21650773684183755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,10240,0.23391572634379068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,10240,0.3420159975687663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,10240,0.14138026237487794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,8192,0.30446933110555013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,8192,0.08362346490224203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,7168,0.09236480394999186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,8192,0.08864426612854004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,7168,0.2381824016571045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,6144,0.08110079765319825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,7168,0.0764245351155599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,6144,0.21087573369344076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,6144,0.06198613246281942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,5120,0.07000746726989746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,5120,0.12021760145823161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,5120,0.05017600059509277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,4096,0.059665067990620935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,4096,0.05778773228327433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,4096,0.040755200386047366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,3584,0.05522773265838623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,3584,0.052019198735555015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,3584,0.03717120091120402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,3072,0.049322664737701416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,3072,0.045021867752075194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,2560,0.04450986782709758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,3072,0.032494932413101196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,2560,0.028125866254170732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,2560,0.038809601465861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,2048,0.03860479990641276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,2048,0.03146666685740153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,2048,0.024200532833735147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,1536,0.020172800620396933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,1536,0.02501973311106364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,1024,0.028091732660929365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,1024,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,1024,0.01986453334490458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,1536,0.033411200841267905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,768,0.02542933424313863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,768,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,768,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,512,0.011878400047620138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,512,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,512,0.011944533387819926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,256,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,256,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,128,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,256,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,128,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,128,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,64,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,64,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,64,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,10240,32,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,10240,32,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,10240,32,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,16384,0.3106815973917643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,65536,0.8359935760498047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,16384,0.4325717290242513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,16384,0.23142399787902831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,65536,1.739741897583008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,65536,1.1480747222900392
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,12288,0.23859200477600098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,12288,0.35795628229777016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,12288,0.11110400358835856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,10240,0.10635946591695149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,10240,0.07406933307647705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,10240,0.29122559229532874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,8192,0.0736255963643392
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,8192,0.21818025906880698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,8192,0.05495466788609823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,7168,0.05191680192947388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,7168,0.18653653462727865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,7168,0.0892245372136434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,6144,0.07570772965749105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,6144,0.07447893619537353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,6144,0.0457045316696167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,5120,0.05905066728591919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,5120,0.058329598108927405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,5120,0.03778560161590576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,4096,0.04860586722691854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,4096,0.030139732360839843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,4096,0.04304213523864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,3072,0.041945600509643556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,3584,0.028501333793004353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,3584,0.04358826478322347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,3584,0.045602134863535565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,3072,0.03597653309504191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,3072,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,2560,0.02215253313382467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,2560,0.029762132962544756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,2560,0.037852799892425536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,2048,0.03256319959958394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,2048,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,2048,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,1536,0.028942932685216267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,1536,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,1536,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,1024,0.024641066789627075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,1024,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,1024,0.01242453356583913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,768,0.023036799828211465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,768,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,768,0.010956799983978272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,512,0.020753065745035805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,512,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,512,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,256,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,256,0.018739199638366698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,256,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,128,0.01716266671816508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,128,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,128,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,64,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,64,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,64,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,8192,32,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,8192,32,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,8192,32,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,16384,0.20609706242879233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,16384,0.2809856096903483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,16384,0.37560319900512695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,65536,0.7415807723999024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,12288,0.18896212577819824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,65536,1.553271484375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,65536,1.0385066350301106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,12288,0.08663040002187093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,10240,0.06987093289693197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,10240,0.08386560281117758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,10240,0.2564746697743734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,8192,0.06099626620610556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,12288,0.30719998677571614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,8192,0.2112170696258545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,8192,0.054542934894561766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,7168,0.07942826747894287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,7168,0.11407360235850017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,7168,0.051336534818013514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,6144,0.06481920083363851
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,6144,0.06085973183314005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,5120,0.051370668411254886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,6144,0.044202665487925216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,5120,0.056012801329294835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,4096,0.04686506589253743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,5120,0.03669333457946777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,4096,0.040550398826599124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,4096,0.028535467386245728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,3584,0.044100264708201095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,3584,0.02754559914271037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,3584,0.037034666538238524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,3072,0.03979626496632894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,3072,0.032323199510574344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,3072,0.023755733172098795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,2560,0.03659093379974365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,2048,0.031675734122594196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,2560,0.027852799495061236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,2560,0.021127466360727945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,2048,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,2048,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,1536,0.02805759906768799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,1536,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,1536,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,1024,0.024200532833735147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,1024,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,1024,0.011707733074824016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,768,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,768,0.022216532627741495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,768,0.010443733135859171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,512,0.020444800456364952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,512,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,512,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,256,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,256,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,256,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,128,0.017063466707865398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,128,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,128,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,64,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,64,0.0038901334007581077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,7168,32,0.0032405334214369455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,64,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,7168,32,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,7168,32,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,16384,0.24524799982706705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,65536,0.6524245580037434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,65536,0.9123498916625976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,16384,0.10768746534983317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,16384,0.33030827840169275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,65536,1.3395626068115234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,12288,0.0986794630686442
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,12288,0.2714282671610514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,12288,0.08959999879201254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,10240,0.06850559711456299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,8192,0.10123946666717529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,10240,0.20735999743143715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,10240,0.06761386394500732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,8192,0.04962986707687378
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,8192,0.05570559899012247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,7168,0.0713045358657837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,7168,0.06304426590601603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,7168,0.050619733333587644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,6144,0.06178133487701416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,6144,0.056149331728617344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,6144,0.04215466578801473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,5120,0.04669440189997355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,5120,0.054237866401672365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,5120,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,4096,0.046284798781077066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,4096,0.03597653309504191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,4096,0.02693120042483012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,3584,0.0425983985265096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,3584,0.032290132840474446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,3584,0.026450133323669432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,3072,0.03890986839930217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,3072,0.02839253346125285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,3072,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,2560,0.03573760191599528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,2560,0.02501973311106364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,2560,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,2048,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,2048,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,2048,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,1536,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,1536,0.014198399583498635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,1536,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,1024,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,1024,0.01122773289680481
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,1024,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,768,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,768,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,768,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,512,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,512,0.00767680009206136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,512,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,256,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,256,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,256,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,128,0.017132800817489625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,128,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,128,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,64,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,64,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,64,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,6144,32,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,6144,32,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,6144,32,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,65536,0.790664545694987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,16384,0.1922730604807536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,65536,0.6486698786417644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,65536,1.1203243255615234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,16384,0.09792853196461995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,16384,0.28129278818766273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,12288,0.07379626433054606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,12288,0.22274880409240722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,12288,0.08284160296122232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,10240,0.06389760176340739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,10240,0.05843626658121744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,10240,0.1921024004618327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,8192,0.05427093505859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,8192,0.051950931549072266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,8192,0.046728531519571945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,7168,0.0667306661605835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,7168,0.04505600134531657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,7168,0.045124268531799315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,6144,0.059699201583862306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,6144,0.040106666088104245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,6144,0.03850239912668864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,5120,0.05273600021998087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,5120,0.03543039957682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,5120,0.03293866713841756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,4096,0.04532906611760457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,4096,0.02945706645647685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,4096,0.026862933238347368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,3584,0.04276693264643351
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,3584,0.029832533995310467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,3584,0.024541866779327393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,3072,0.03819520076115926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,3072,0.02491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,3072,0.02222080032030741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,2560,0.034918399651845296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,2560,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,2560,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,2048,0.03078826665878296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,2048,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,2048,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,1536,0.02723840077718099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,1536,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,1536,0.013516799608866373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,1024,0.023552000522613525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,1024,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,768,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,1024,0.010922666390736897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,768,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,768,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,512,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,512,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,512,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,256,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,256,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,128,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,256,0.007030400137106578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,64,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,64,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,128,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,64,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,5120,32,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,5120,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,5120,32,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,65536,0.676693344116211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,16384,0.09512960116068522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,65536,0.4819967905680339
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,16384,0.24415359497070313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,16384,0.05492053429285685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,65536,0.9069226582845052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,12288,0.07154346307118734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,12288,0.13205973307291669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,10240,0.06287360191345215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,12288,0.04375893274943034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,10240,0.05215573310852051
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,10240,0.03775146802266439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,8192,0.04341760079065959
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,8192,0.05376000006993612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,8192,0.028535467386245728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,7168,0.065774933497111
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,7168,0.029013333717981975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,6144,0.05880853335062662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,7168,0.04215466578801473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,6144,0.036761601765950516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,6144,0.02529279987017314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,5120,0.05187840064366659
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,5120,0.03065173427263896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,5120,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,4096,0.04522666533788045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,4096,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,4096,0.02501973311106364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,3584,0.0413696010907491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,3584,0.0233130673567454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,3584,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,3072,0.037717334429423016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,3072,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,2560,0.03426986535390218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,2560,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,3072,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,2560,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,2048,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,2048,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,2048,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,1536,0.027033599217732747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,1536,0.012593066692352295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,1536,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,1024,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,1024,0.007811200122038524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,768,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,1024,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,768,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,768,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,512,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,512,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,512,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,256,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,256,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,128,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,256,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,128,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,128,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,64,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,64,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,64,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,4096,32,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,4096,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,4096,32,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,16384,0.08980480035146078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,65536,0.6213631947835286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,65536,0.4182357470194499
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,65536,0.8354133605957031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,16384,0.1983829339345296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,16384,0.05021013418833414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,12288,0.07137280305226644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,12288,0.05410133202870687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,12288,0.04116479953130086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,10240,0.06287360191345215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,10240,0.04727466503779094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,10240,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,8192,0.054030934969584145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,8192,0.039833601315816244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,8192,0.025975465774536133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,7168,0.03222186764081319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,7168,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,6144,0.05946026643117269
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,6144,0.029183999697367353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,7168,0.06611626545588176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,6144,0.02222080032030741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,5120,0.05174613396326701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,5120,0.02836479942003886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,5120,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,4096,0.045021867752075194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,4096,0.02296746571858724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,4096,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,3584,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,3584,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,3584,0.04167679945627849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,3072,0.03768320083618164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,3072,0.020002132654190062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,3072,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,2560,0.03389439980189006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,2560,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,2560,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,2048,0.03031040032704671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,2048,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,2048,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,1536,0.026897066831588747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,1536,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,1536,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,1024,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,1024,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,1024,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,768,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,768,0.007197866837183635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,768,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,512,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,512,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,512,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,256,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,256,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,256,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,128,0.016923733552296958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,128,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,128,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,64,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,64,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,64,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3584,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3584,32,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3584,32,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,16384,0.08983893394470215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,65536,0.5677738825480143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,65536,0.36918614705403646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,65536,0.6782976150512695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,16384,0.15387306213378907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,16384,0.04416853189468384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,12288,0.0717141310373942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,12288,0.05051733255386352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,12288,0.03638613224029541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,10240,0.04126720031102498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,10240,0.06270293394724527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,10240,0.031163734197616578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,8192,0.053350400924682614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,8192,0.036761601765950516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,8192,0.024064000447591147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,7168,0.025320533911387128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,7168,0.0661845326423645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,7168,0.028637866179148357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,6144,0.05870933135350546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,6144,0.02635093331336975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,6144,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,5120,0.051780267556508386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,5120,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,5120,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,4096,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,4096,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,3584,0.04140373468399048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,4096,0.045187199115753175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,3584,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,3584,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,3072,0.037512532869974774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,3072,0.017438934246699015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,3072,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,2560,0.03399680058161418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,2560,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,2560,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,2048,0.030105600754419964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,2048,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,2048,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,1536,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,1536,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,1536,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,1024,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,1024,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,768,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,768,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,1024,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,768,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,512,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,512,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,512,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,256,0.017850667238235474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,256,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,128,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,128,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,128,0.005050666630268097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,64,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,64,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,3072,32,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,3072,32,0.0026591998835404714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,3072,32,0.0048426667849222815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,65536,0.533401616414388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,16384,0.0896341323852539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,65536,0.3125589370727539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,65536,0.5593429565429687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,16384,0.054237866401672365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,16384,0.03737066586812337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,12288,0.07116693655649821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,12288,0.04222293297449748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,10240,0.062463998794555664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,10240,0.03805866638819377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,10240,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,12288,0.032017066081364946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,8192,0.05314346551895142
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,8192,0.02959360082944234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,8192,0.02105600039164225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,7168,0.06580906709035238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,7168,0.027613866329193115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,6144,0.0589141329129537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,6144,0.023995733261108397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,7168,0.020172800620396933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,6144,0.01832746664683024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,5120,0.051843198140462246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,5120,0.02177706758181254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,4096,0.04437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,5120,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,4096,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,4096,0.01269760032494863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,3584,0.04130133390426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,3584,0.025770666201909383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,3584,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,3072,0.03730346759160359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,3072,0.021604265769322714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,2560,0.0339626669883728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,2560,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,2560,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,3072,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,2048,0.03007146716117859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,2048,0.014131200313568116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,2048,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,1536,0.026658133665720625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,1536,0.011980799833933513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,1536,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,1024,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,1024,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,1024,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,768,0.02173759937286377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,768,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,768,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,512,0.005904000004132589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,512,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,512,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,256,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,256,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,256,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,128,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,128,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,128,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,64,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,64,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2560,32,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2560,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,64,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2560,32,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,16384,0.0893951972325643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,65536,0.5103957176208496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,65536,0.25719466209411623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,65536,0.44950186411539717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,16384,0.04256426493326823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,16384,0.03573760191599528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,12288,0.0340992013613383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,12288,0.0713045358657837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,12288,0.03051519989967346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,10240,0.062156800429026285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,10240,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,10240,0.02938773234685262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,8192,0.05303786595662435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,8192,0.024200532833735147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,8192,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,7168,0.06560426553090414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,7168,0.022425599892934165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,7168,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,6144,0.058606934547424314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,6144,0.019797333081563315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,6144,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,5120,0.05147306521733602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,5120,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,5120,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,4096,0.0443391998608907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,4096,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,4096,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,3584,0.040994131565093996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,3584,0.012589866916338602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,3072,0.037205334504445395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,3584,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,3072,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,3072,0.010922666390736897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,2560,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,2560,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,2560,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,2048,0.030173865954081218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,2048,0.012763733665148417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,2048,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,1536,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,1536,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,1024,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,1536,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,1024,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,1024,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,768,0.021329067150751748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,768,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,768,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,512,0.019797333081563315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,512,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,256,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,512,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,256,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,256,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,128,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,128,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,128,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,64,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,64,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,64,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,2048,32,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,2048,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,2048,32,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,65536,0.19814400672912597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,65536,0.45424639383951826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,65536,0.3352917353312174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,16384,0.08936106363932292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,16384,0.0340992013613383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,16384,0.035327998797098796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,12288,0.07120213508605958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,12288,0.02768213351567586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,12288,0.027818665901819868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,10240,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,10240,0.024200532833735147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,10240,0.062156800429026285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,8192,0.01986560026804606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,8192,0.05324800014495849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,8192,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,7168,0.06488746802012126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,7168,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,7168,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,6144,0.05836799939473471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,6144,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,6144,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,5120,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,5120,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,5120,0.051609599590301515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,4096,0.04444160064061482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,4096,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,4096,0.01242453356583913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,3584,0.04068586826324463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,3584,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,3584,0.011396267016728719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,3072,0.037102933724721274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,3072,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,3072,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,2560,0.03375786542892456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,2560,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,2560,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,2048,0.02986666758855184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,2048,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,2048,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,1536,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,1536,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,1536,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,1024,0.023005867004394533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,1024,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,1024,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,768,0.021845332781473794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,768,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,512,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,512,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,768,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,512,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,256,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,256,0.0037823999921480812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,256,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,128,0.01658453345298767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,128,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,128,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,64,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,64,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,64,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1536,32,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1536,32,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1536,32,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,65536,0.4036266644795735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,65536,0.1295018672943115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,65536,0.237772798538208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,16384,0.08932693004608154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,16384,0.023893332481384276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,16384,0.03515733480453491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,12288,0.0710314671198527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,12288,0.01925119956334432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,12288,0.02775040070215861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,10240,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,10240,0.062327468395233156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,10240,0.023893332481384276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,8192,0.05311146577199301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,8192,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,8192,0.020172800620396933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,7168,0.06478506724039713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,7168,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,7168,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,6144,0.05833386580149332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,6144,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,6144,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,5120,0.05113173325856527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,5120,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,5120,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,4096,0.04430506626764934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,4096,0.012219732999801636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,4096,0.010752000411351522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,3584,0.04092586835225423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,3584,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,3584,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,3072,0.037375998497009275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,3072,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,3072,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,2560,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,2560,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,2560,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,2048,0.029934932788213093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,2048,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,2048,0.008327466746171314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,1536,0.02645333409309387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,1536,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,1536,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,1024,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,1024,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,768,0.021329067150751748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,1024,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,768,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,512,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,768,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,512,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,512,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,256,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,256,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,128,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,128,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,128,0.0028309332827727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,64,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,64,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,256,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,64,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,1024,32,0.01613866686820984
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,1024,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,1024,32,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,16384,0.08901973565419516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,65536,0.1762986660003662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,16384,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,65536,0.1293994665145874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,65536,0.3762463887532552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,16384,0.03502080043156942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,12288,0.0711680014928182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,12288,0.02757333318392436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,10240,0.062293334801991784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,12288,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,8192,0.05341866811116537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,10240,0.023718400796254476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,10240,0.014706133802731832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,8192,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,8192,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,7168,0.06485333442687988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,7168,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,7168,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,6144,0.058265598615010586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,6144,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,6144,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,5120,0.05116586685180664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,5120,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,5120,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,4096,0.04437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,4096,0.011877333124478657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,4096,0.012183466553688049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,3584,0.04061866601308187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,3584,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,3584,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,3072,0.03700053294499715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,3072,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,3072,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,2560,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,2560,0.03362133502960205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,2560,0.009282132983207703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,2048,0.029934932788213093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,2048,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,2048,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,1536,0.026692267258961993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,1536,0.00737066666285197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,1536,0.007575466732184092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,1024,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,1024,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,1024,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,768,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,768,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,768,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,512,0.019385600090026857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,512,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,512,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,256,0.017541333039601644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,256,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,256,0.0032416000962257386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,128,0.016416000326474507
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,128,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,64,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,128,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,64,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,768,32,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,768,32,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,768,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,65536,0.3643733342488607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,65536,0.12892159620920818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,65536,0.042114134629567465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,16384,0.08942933082580566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,16384,0.014607999722162882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,16384,0.03519146839777629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,12288,0.07099733352661133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,12288,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,12288,0.02754559914271037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,10240,0.0636245330174764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,10240,0.016790399948755898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,10240,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,8192,0.0530730684598287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,8192,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,8192,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,7168,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,7168,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,7168,0.06519466638565063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,6144,0.05806080102920532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,6144,0.011980799833933513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,6144,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,5120,0.05109759966532389
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,5120,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,5120,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,4096,0.04423679908116658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,4096,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,4096,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,3584,0.04089173475901286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,3584,0.008906666437784832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,3584,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,3072,0.0369322657585144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,3072,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,3072,0.010342400272687275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,2560,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,2560,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,2560,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,2048,0.02955946723620097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,2048,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,2048,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,1536,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,1536,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,1536,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,1024,0.005525333185990652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,1024,0.006347733239332835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,1024,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,768,0.021367466449737547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,768,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,768,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,512,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,512,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,512,0.005456000069777171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,256,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,256,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,256,0.0031370667119820913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,128,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,128,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,128,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,64,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,64,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,64,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,512,32,0.015871999661127727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,512,32,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,512,32,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,65536,0.36136960983276367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,65536,0.04266666571299235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,65536,0.12707839806874593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,16384,0.08919040362040201
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,16384,0.013926399747530618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,16384,0.034508800506591795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,12288,0.07092906634012858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,12288,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,12288,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,10240,0.062395731608072914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,10240,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,10240,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,8192,0.053282133738199865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,8192,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,8192,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,7168,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,7168,0.06492160161336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,7168,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,6144,0.05819733142852783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,6144,0.008221866687138875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,6144,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,5120,0.007779199878374736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,5120,0.050824534893035886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,5120,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,4096,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,4096,0.04413333336512248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,4096,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,3584,0.040550398826599124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,3584,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,3584,0.0068256000677744556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,3072,0.03693013191223145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,3072,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,3072,0.010307199756304423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,2560,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,2560,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,2560,0.00921493371327718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,2048,0.029627732435862225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,2048,0.005493333439032236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,2048,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,1536,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,1536,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,1024,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,1536,0.007367466886838277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,1024,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,1024,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,768,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,768,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,768,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,512,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,512,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,512,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,256,0.017403733730316163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,256,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,128,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,128,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,128,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,64,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,64,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,64,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,256,32,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,256,32,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,256,32,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,65536,0.3610623995463053
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,65536,0.023756800095240276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,65536,0.1257472038269043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,16384,0.08891733487447104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,16384,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,16384,0.03420159816741943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,12288,0.07147520383199056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,12288,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,12288,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,10240,0.062293334801991784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,10240,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,10240,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,8192,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,8192,0.05311146577199301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,8192,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,7168,0.06467733383178711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,7168,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,7168,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,6144,0.05809493462244669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,6144,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,6144,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,5120,0.0509610652923584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,5120,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,5120,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,4096,0.044202665487925216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,4096,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,4096,0.011878400047620138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,3584,0.040447998046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,3584,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,3072,0.037137067317962645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,3584,0.011018666625022887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,3072,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,3072,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,2560,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,2560,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,2560,0.009347200393676758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,2048,0.029798400402069092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,1536,0.02645333409309387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,2048,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,1536,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,2048,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,1536,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,1024,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,1024,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,1024,0.006241066753864289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,768,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,768,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,768,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,512,0.0034805332620938623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,512,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,512,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,256,0.01750826636950175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,256,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,128,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,128,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,128,0.004572799801826477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,64,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,64,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,128,32,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,128,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,128,32,0.0047082667549451195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,65536,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,65536,0.12526506582895916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,65536,0.36300799051920574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,16384,0.08912213643391928
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,16384,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,16384,0.03423466682434082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,12288,0.07113386789957682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,12288,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,12288,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,10240,0.006211199859778086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,10240,0.0618837316830953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,10240,0.02321066657702128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,8192,0.052799999713897705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,8192,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,8192,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,7168,0.06461439927419027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,7168,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,7168,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,6144,0.057581865787506105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,6144,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,6144,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,5120,0.050925866762797034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,5120,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,5120,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,4096,0.04392960071563721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,4096,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,4096,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,3584,0.040345601240793866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,3584,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,3584,0.010922666390736897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,3072,0.0370688001314799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,3072,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,3072,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,2560,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,2560,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,2560,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,2048,0.029388799269994097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,2048,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,2048,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,1536,0.02635093331336975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,1536,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,1536,0.00719893326361974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,1024,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,1024,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,1024,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,768,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,768,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,768,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,512,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,512,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,256,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,512,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,256,0.002932266642649968
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,256,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,128,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,128,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,128,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,64,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,64,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,64,0.015871999661127727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,64,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,64,32,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,64,32,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,65536,0.367957337697347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,65536,0.012663466731707254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,65536,0.12373119990030926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,16384,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,16384,0.08932693004608154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,16384,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,12288,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,12288,0.07086079915364583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,12288,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,10240,0.061713067690531406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,10240,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,10240,0.02259413401285807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,8192,0.052940801779429114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,8192,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,8192,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,7168,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,7168,0.06454613208770751
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,6144,0.057719465096791586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,7168,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,6144,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,6144,0.015422933300336204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,5120,0.05085866848627726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,5120,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,5120,0.013755733768145243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,4096,0.04379306634267171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,4096,0.004670933385690053
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,4096,0.01181013286113739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,3584,0.040140799681345624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,3584,0.0050794666012128195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,3584,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,3072,0.036761601765950516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,3072,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,3072,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,2560,0.03304106593132019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,2560,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,2560,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,2048,0.029627732435862225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,2048,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,2048,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,1536,0.026180267333984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,1536,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,1536,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,1024,0.022730666399002075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,1024,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,1024,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,768,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,768,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,768,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,512,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,512,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,512,0.005284266670544943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,256,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,256,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,256,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,128,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,128,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,128,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,64,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,64,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,64,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,96,32,32,0.01593493322531382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,96,32,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,96,32,32,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,16384,1.770461908976237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,12288,1.1794432322184245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,16384,1.5560704549153646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,12288,2.9038592020670575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,10240,1.1035648345947267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,16384,3.510545094807943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,12288,1.3267285664876303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,10240,2.2753280639648437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,10240,1.0618197123209634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,8192,0.8813226699829102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,8192,0.8469504038492838
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,8192,1.6593578338623047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,7168,0.6908927917480469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,7168,0.7953717549641927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,7168,1.5225855509440103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,6144,0.6850218454996745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,6144,0.6448810577392579
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,6144,1.2822869618733725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,5120,0.5753856023152669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,5120,1.0423295974731446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,5120,0.5407061258951823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,4096,0.46230185826619463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,4096,0.9278111775716147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,4096,0.43250347773234055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,3584,0.4106581370035808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,3584,0.3575104077657064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,3584,0.7393621444702149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,3072,0.3365887959798177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,3072,0.6440959930419922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,3072,0.3092138608296712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,2560,0.29409281412760413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,2560,0.5367466608683269
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,2560,0.26146133740743005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,2048,0.2356565316518148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,2048,0.2210474650065104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,1536,0.18483200073242187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,2048,0.4765013376871745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,1536,0.33249279657999675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,1536,0.15267839431762695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,1024,0.07901866436004638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,1024,0.23012693723042807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,1024,0.07007573445638021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,768,0.06884693304697673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,768,0.1829546610514323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,768,0.06403306722640992
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,512,0.04468053181966146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,512,0.05918720165888468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,512,0.04983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,256,0.043724799156188966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,256,0.040652799606323245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,256,0.03426986535390218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,128,0.038499200344085695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,128,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,128,0.029798400402069092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,64,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,64,0.037205334504445395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,64,0.02877226670583089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,65536,32,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,65536,32,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,65536,32,0.028262400627136232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,65536,1.5676074981689454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,16384,0.4248575846354167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,16384,0.5289984067281087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,16384,0.9012906392415365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,12288,0.41188694636027023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,12288,0.7383381525675456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,65536,3.5340276082356774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,65536,2.034926986694336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,12288,0.331876277923584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,10240,0.33621333440144857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,10240,0.5587626775105794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,10240,0.27374931971232097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,8192,0.25350826581319175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,8192,0.21981546084086098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,8192,0.4887210528055827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,7168,0.24630400339762368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,7168,0.19411627451578778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,6144,0.20350294113159179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,7168,0.3667957305908203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,6144,0.09721173445383707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,6144,0.33621333440144857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,5120,0.08526506423950195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,5120,0.2943658510843913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,5120,0.06894933382670085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,4096,0.0614741325378418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,4096,0.05515946547190348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,3584,0.18923519452412924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,3584,0.0549888014793396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,4096,0.21564799944559732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,3584,0.05013866821924845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,3072,0.08294400374094645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,3072,0.05044266780217489
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,3072,0.043246932824452716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,2560,0.04478293259938558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,2560,0.0392192006111145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,2560,0.04685973326365153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,2048,0.037205334504445395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,2048,0.03874133427937825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,2048,0.03204906582832336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,1536,0.03368959824244182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,1536,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,1536,0.02877440055211385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,1024,0.028398933013280232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,1024,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,1024,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,768,0.019317332903544107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,768,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,768,0.018156800667444864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,512,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,512,0.013753599921862283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,512,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,256,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,256,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,256,0.011398399869600933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,128,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,128,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,128,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,64,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,64,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,64,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,16384,32,0.018500266472498576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,16384,32,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,16384,32,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,65536,1.6151893615722657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,65536,1.1905364990234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,16384,0.3012245178222656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,16384,0.4204202651977539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,12288,0.327509339650472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,16384,0.6668970743815105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,12288,0.5429589589436848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,65536,2.6556073506673177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,10240,0.39731200536092126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,12288,0.2513919989267985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,10240,0.27224747339884436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,10240,0.2078378677368164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,8192,0.21220693588256836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,8192,0.32225281397501626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,8192,0.0853333314259847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,7168,0.10734933217366535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,7168,0.07331840197245279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,6144,0.07976746559143066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,6144,0.06348479986190796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,7168,0.2861397425333659
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,6144,0.25569279988606775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,5120,0.05123413403828939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,5120,0.06990506649017333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,5120,0.21046613057454427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,4096,0.05925546487172445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,4096,0.08632319768269857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,4096,0.04293973445892334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,3584,0.06219093402226766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,3584,0.05372373263041178
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,3584,0.03822933435440064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,3072,0.05437440077463786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,3072,0.048742401599884036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,3072,0.034508800506591795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,2560,0.04389546712239583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,2560,0.04679679870605469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,2560,0.02986666758855184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,2048,0.036556800206502274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,2048,0.03822933435440064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,2048,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,1536,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,1536,0.028706133365631104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,1536,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,1024,0.02802346746126811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,1024,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,1024,0.020309333006540933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,768,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,768,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,768,0.02518613338470459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,512,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,512,0.011980799833933513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,512,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,256,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,256,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,256,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,128,0.018601600329081217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,128,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,128,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,64,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,64,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,64,0.008088533580303193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,12288,32,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,12288,32,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,12288,32,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,65536,1.3633536020914714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,65536,1.0137237548828124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,16384,0.3611647923787435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,16384,0.25630720456441247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,16384,0.5588650385538737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,65536,2.2118059794108076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,12288,0.2793813387552897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,12288,0.44404052098592117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,12288,0.21575679779052734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,10240,0.2268160025278727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,10240,0.3408896128336588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,8192,0.0823637326558431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,10240,0.1534293333689372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,8192,0.07253333727518717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,8192,0.2827605247497559
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,7168,0.09099946816762289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,6144,0.07953066825866699
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,7168,0.25882986386617024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,7168,0.06557013193766276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,6144,0.05939199924468994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,6144,0.20838400522867837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,5120,0.06959786415100097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,5120,0.13372799555460613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,5120,0.048537600040435794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,4096,0.0589141329129537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,4096,0.05768533150355021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,3584,0.05393066803614298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,4096,0.03863893349965413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,3584,0.052019198735555015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,3584,0.036008532842000326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,3072,0.048742401599884036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,3072,0.04481706619262695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,3072,0.031778132915496825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,2560,0.043656531969706217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,2560,0.03805760145187378
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,2048,0.03839999834696452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,2560,0.028194133440653486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,2048,0.03126613299051921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,2048,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,1536,0.03279786705970764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,1536,0.02481493353843689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,1536,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,1024,0.02788693308830261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,1024,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,768,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,1024,0.01562773287296295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,768,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,768,0.013755733768145243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,512,0.022765866915384927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,512,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,512,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,256,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,256,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,256,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,128,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,128,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,128,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,64,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,64,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,64,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,10240,32,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,10240,32,0.0033418667813142145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,10240,32,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,65536,1.123532740275065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,65536,0.8261290868123373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,16384,0.2975402514139811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,16384,0.22330026626586913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,16384,0.4301482518513997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,65536,1.7406976064046222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,12288,0.22381226221720377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,12288,0.35317761103312173
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,12288,0.1172106663386027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,10240,0.09294506708780924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,10240,0.06898346741994223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,10240,0.29146451950073243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,8192,0.06881279945373535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,8192,0.23736319541931153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,8192,0.053041064739227296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,7168,0.08611839612325033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,7168,0.18599252700805663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,7168,0.04966400067011516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,6144,0.0695637305577596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,6144,0.043281066417694095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,5120,0.057241598765055336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,6144,0.07321600119272867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,5120,0.054067198435465494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,5120,0.03700053294499715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,4096,0.0423583984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,4096,0.04737706581751506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,4096,0.02921813329060872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,3584,0.04478293259938558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,3584,0.027409066756566364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,3584,0.03997013171513875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,3072,0.040209066867828366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,3072,0.03382399876912435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,3072,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,2560,0.03648853302001953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,2560,0.029113600651423138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,2560,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,2048,0.031982932488123575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,2048,0.02228906750679016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,2048,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,1536,0.028808534145355225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,1536,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,1536,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,1024,0.024166399240493776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,1024,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,1024,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,768,0.022286933660507203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,768,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,768,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,512,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,512,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,512,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,256,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,256,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,256,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,128,0.017167999347050985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,128,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,128,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,64,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,8192,32,0.017406932512919106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,64,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,64,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,8192,32,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,8192,32,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,65536,0.8111786524454752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,16384,0.19360426266988118
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,16384,0.3692885398864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,16384,0.2692437489827474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,65536,1.0101418813069662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,65536,1.5343616485595704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,12288,0.132915202776591
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,12288,0.30757548014322916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,12288,0.08055466810862223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,10240,0.23179945945739747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,10240,0.08137386639912923
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,10240,0.06625279982884726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,8192,0.057309865951538086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,8192,0.20821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,8192,0.0501418670018514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,7168,0.07959893544514975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,7168,0.04778666496276855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,7168,0.09379839897155762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,6144,0.06335146824518839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,6144,0.05816320180892944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,5120,0.054169599215189615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,6144,0.04133546749750773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,5120,0.05048319896062216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,5120,0.034542934099833174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,4096,0.045875199635823566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,4096,0.03945813179016113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,4096,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,3584,0.0353216012318929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,3584,0.04266666571299235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,3584,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,3072,0.03860479990641276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,3072,0.03051519989967346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,3072,0.0234826664129893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,2560,0.03604480028152466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,2560,0.025975465774536133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,2560,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,2048,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,2048,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,2048,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,1536,0.027613866329193115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,1536,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,1536,0.014165332913398743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,1024,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,1024,0.012117333213488261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,1024,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,768,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,768,0.021845332781473794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,768,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,512,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,512,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,512,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,256,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,256,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,256,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,128,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,128,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,128,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,64,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,64,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,64,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,7168,32,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,7168,32,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,7168,32,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,65536,0.8879103978474936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,16384,0.23794347445170083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,65536,0.7137280146280924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,16384,0.32235520680745444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,65536,1.3324970245361327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,16384,0.11567786534627278
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,12288,0.08704000314076742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,12288,0.27426134745279945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,12288,0.0859818696975708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,10240,0.06447786490122477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,10240,0.06492160161336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,10240,0.21155840555826821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,8192,0.05410133202870687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,8192,0.0843775987625122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,8192,0.04618240197499593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,7168,0.06799360116322836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,7168,0.0599669337272644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,7168,0.046250665187835695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,6144,0.051370668411254886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,6144,0.04146666526794433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,6144,0.06014293432235718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,5120,0.052019198735555015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,5120,0.044202665487925216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,5120,0.03239253362019857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,4096,0.035327998797098796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,4096,0.025873066981633504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,4096,0.04502079884211223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,3584,0.04164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,3584,0.03099306623140971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,3584,0.024780799945195518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,3072,0.03775146802266439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,3072,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,3072,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,2560,0.03444053332010905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,2560,0.022562134265899658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,2560,0.01942080060640971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,2048,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,2048,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,2048,0.01607253352801005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,1536,0.027101866404215497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,1536,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,1536,0.013755733768145243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,1024,0.023514666159947715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,1024,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,1024,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,768,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,768,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,768,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,512,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,512,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,512,0.008770133058230083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,256,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,256,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,256,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,128,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,128,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,64,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,64,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,64,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,6144,32,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,6144,32,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,6144,32,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,16384,0.16776533126831056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,65536,0.648635737101237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,16384,0.2832042694091797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,65536,1.117457071940104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,16384,0.09448106288909912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,65536,0.7661226908365886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,12288,0.07167999744415283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,12288,0.22022825876871743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,12288,0.06666239897410074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,10240,0.06174720128377279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,10240,0.18565120697021484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,10240,0.05672959884007772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,8192,0.05259946584701538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,8192,0.05741226673126221
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,8192,0.045363199710845944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,7168,0.06502399841944376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,7168,0.04355413516362508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,7168,0.04154026508331299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,6144,0.05881173213322958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,6144,0.03833173513412476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,6144,0.03679573138554891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,5120,0.051370668411254886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,5120,0.03379199902216594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,5120,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,4096,0.0448853333791097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,4096,0.0315391997496287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,4096,0.025668267409006757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,3584,0.04119786818822225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,3584,0.02775040070215861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,3584,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,3072,0.03727146784464518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,3072,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,3072,0.02228906750679016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,2560,0.03385920127232869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,2560,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,2560,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,2048,0.030306132634480794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,2048,0.017268266280492148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,2048,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,1536,0.026862933238347368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,1536,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,1536,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,1024,0.024029866854349772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,1024,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,1024,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,768,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,768,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,768,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,512,0.01986560026804606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,512,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,512,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,256,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,256,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,128,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,256,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,128,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,64,0.01641706625620524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,64,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,64,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,5120,32,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,5120,32,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,5120,32,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,65536,0.6526293436686198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,16384,0.0889514684677124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,65536,0.47428267796834306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,65536,0.8791381200154623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,16384,0.224017063776652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,16384,0.05095680157343546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,12288,0.12192426522572834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,12288,0.0695637305577596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,12288,0.04150613149007161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,10240,0.06154239972432455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,10240,0.036556800206502274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,10240,0.05150506496429443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,8192,0.05277013381322225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,8192,0.04242773453394572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,8192,0.028051199515660603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,7168,0.06454613208770751
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,7168,0.02590720057487488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,7168,0.03863893349965413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,6144,0.05809493462244669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,6144,0.034264532725016277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,6144,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,5120,0.051268267631530764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,5120,0.029013333717981975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,5120,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,4096,0.0448853333791097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,4096,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,3584,0.040482131640116374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,4096,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,3584,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,3584,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,3072,0.03689813216527303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,3072,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,3072,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,2560,0.03362133502960205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,2560,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,2560,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,2048,0.030242133140563964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,2048,0.014233600099881491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,2048,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,1536,0.026555732885996504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,1536,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,1536,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,1024,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,1024,0.008738133311271667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,1024,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,768,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,768,0.007743999858697255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,768,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,512,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,512,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,512,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,256,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,256,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,256,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,128,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,128,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,128,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,64,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,64,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,64,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,4096,32,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,4096,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,4096,32,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,16384,0.08697173595428467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,65536,0.4203520139058431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,65536,0.5986986796061198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,65536,0.767795181274414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,16384,0.045841066042582194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,16384,0.20070293744405113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,12288,0.06973439852396647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,12288,0.052940801779429114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,12288,0.038024532794952395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,10240,0.06143999894460043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,10240,0.0335530678431193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,10240,0.046284798781077066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,8192,0.05287040074666342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,8192,0.039185067017873124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,8192,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,7168,0.06430720090866089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,7168,0.03150506615638733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,7168,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,6144,0.05792426665623983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,6144,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,6144,0.031197865804036457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,5120,0.05092693169911703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,5120,0.026282666126887004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,5120,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,4096,0.04427093267440796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,4096,0.014879999558130899
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,4096,0.02208426594734192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,3584,0.040755200386047366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,3584,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,3584,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,3072,0.03679573138554891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,3072,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,3072,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,2560,0.03375786542892456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,2560,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,2560,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,2048,0.013136000434557597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,2048,0.02986666758855184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,2048,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,1536,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,1536,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,1536,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,1024,0.02321066657702128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,1024,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,1024,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,768,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,768,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,768,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,512,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,512,0.005799466868241628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,512,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,256,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,256,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,256,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,128,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,128,0.0034442665676275887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,128,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,64,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,64,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,64,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3584,32,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3584,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3584,32,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,16384,0.08744959831237793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,65536,0.36154025395711265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,65536,0.6723583857218425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,65536,0.5438453038533528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,16384,0.14701226552327473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,16384,0.04280213514963786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,12288,0.06939307053883871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,12288,0.05085866848627726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,12288,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,10240,0.06099626620610556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,10240,0.04369066556294759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,10240,0.030446932713190718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,8192,0.0527018666267395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,8192,0.036761601765950516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,8192,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,7168,0.06417066653569539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,7168,0.02867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,7168,0.02297066648801168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,6144,0.05765120188395182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,6144,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,6144,0.019657599925994872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,5120,0.050756267706553136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,5120,0.022151466210683188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,5120,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,4096,0.044544001420338944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,4096,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,4096,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,3584,0.040755200386047366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,3584,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,3584,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,3072,0.03672746817270915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,3072,0.016720000902811685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,3072,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,2560,0.03334826628367106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,2560,0.01437013347943624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,2560,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,2048,0.02986666758855184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,2048,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,2048,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,1536,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,1536,0.010033067067464192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,1536,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,1024,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,1024,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,1024,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,768,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,768,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,768,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,512,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,512,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,512,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,256,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,256,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,256,0.005285333096981049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,128,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,128,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,128,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,64,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,64,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,64,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,3072,32,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,3072,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,3072,32,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,65536,0.5065685272216797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,16384,0.08659626642862955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,65536,0.30467093785603844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,65536,0.5534037272135417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,16384,0.051268267631530764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,16384,0.0362496018409729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,12288,0.06915413538614909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,12288,0.040550398826599124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,12288,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,10240,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,10240,0.035396265983581546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,10240,0.06048426628112793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,8192,0.05256533225377401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,8192,0.028330665826797486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,8192,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,7168,0.06420480012893677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,7168,0.025732266902923583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,7168,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,6144,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,6144,0.05829973220825195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,6144,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,5120,0.050892798105875645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,5120,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,5120,0.016005333264668783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,4096,0.04375893274943034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,4096,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,4096,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,3584,0.040755200386047366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,3584,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,3584,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,3072,0.036556800206502274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,3072,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,3072,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,2560,0.03338239987691243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,2560,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,2560,0.010409599542617798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,2048,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,2048,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,2048,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,1536,0.02635093331336975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,1536,0.011571199695269267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,1536,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,1024,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,1024,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,1024,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,768,0.021192532777786256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,768,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,768,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,512,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,512,0.005460266768932342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,512,0.01948480010032654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,256,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,256,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,256,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,128,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,128,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,128,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,64,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,64,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,64,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2560,32,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2560,32,0.0027978666126728057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2560,32,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,16384,0.08645973205566407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,65536,0.44588371912638347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,65536,0.2483882745107015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,65536,0.48626346588134767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,16384,0.04109653234481812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,16384,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,12288,0.06918826897939047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,12288,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,12288,0.028535467386245728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,10240,0.060347731908162436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,10240,0.02853226661682129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,8192,0.052121599515279136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,10240,0.024130133787790935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,8192,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,7168,0.06386346817016601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,8192,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,7168,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,7168,0.01900906761487325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,6144,0.057207465171813965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,6144,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,6144,0.016755199432373045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,5120,0.050551466147104894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,5120,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,4096,0.04382719993591309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,4096,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,5120,0.014847999811172486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,3584,0.040378665924072264
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,4096,0.012321066856384278
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,3584,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,3584,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,3072,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,3072,0.03645439942677815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,3072,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,2560,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,2560,0.010033067067464192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,2560,0.0335530678431193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,2048,0.029491200049718218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,2048,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,2048,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,1536,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,1536,0.026281599203745527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,1536,0.007916800181070964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,1024,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,1024,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,1024,0.0066890666882197065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,768,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,768,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,768,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,512,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,512,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,512,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,256,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,256,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,256,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,128,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,128,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,128,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,64,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,64,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,64,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,2048,32,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,2048,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,2048,32,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,65536,0.18957653045654296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,65536,0.337339719136556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,16384,0.08642559846242269
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,65536,0.4496383984883626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,12288,0.06932480335235595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,16384,0.03549866676330567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,12288,0.02696533401807149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,16384,0.03345066706339518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,12288,0.02768213351567586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,10240,0.0605183998743693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,10240,0.023961599667867026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,10240,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,8192,0.05215573310852051
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,8192,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,8192,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,7168,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,7168,0.06393173138300577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,7168,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,6144,0.05717333157857259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,6144,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,6144,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,5120,0.050312534968058265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,5120,0.014643200238545737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,5120,0.014472533265749613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,4096,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,4096,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,3584,0.039867734909057616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,4096,0.04355413516362508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,3584,0.011844266454378765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,3584,0.011400533715883891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,3072,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,3072,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,3072,0.03700053294499715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,2560,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,2560,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,2048,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,2560,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,2048,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,2048,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,1536,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,1536,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,1536,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,1024,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,1024,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,768,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,1024,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,768,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,768,0.005835733314355215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,512,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,512,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,256,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,512,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,256,0.0036490666369597114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,128,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,256,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,128,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,128,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,64,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,64,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,64,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1536,32,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1536,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1536,32,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,65536,0.37007039388020835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,65536,0.22978347142537436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,65536,0.12922879854838054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,16384,0.08622079690297445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,16384,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,16384,0.035327998797098796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,12288,0.06918826897939047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,12288,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,12288,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,10240,0.060279464721679686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,10240,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,10240,0.023720532655715942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,8192,0.05174613396326701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,8192,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,8192,0.01993173360824585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,7168,0.0636245330174764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,6144,0.057207465171813965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,7168,0.014711466431617738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,7168,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,6144,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,6144,0.013409066200256347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,5120,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,5120,0.050892798105875645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,5120,0.014165332913398743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,4096,0.043724799156188966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,4096,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,4096,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,3584,0.03973120053609212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,3584,0.014711466431617738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,3584,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,3072,0.036556800206502274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,3072,0.01269760032494863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,3072,0.010342400272687275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,2560,0.03297280073165894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,2560,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,2560,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,2048,0.029419734080632524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,2048,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,1536,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,2048,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,1536,0.02611200014750163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,1536,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,1024,0.022801067431767783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,1024,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,1024,0.006342400113741558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,768,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,768,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,512,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,768,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,512,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,512,0.005625600119431814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,256,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,256,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,256,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,128,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,128,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,64,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,64,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,128,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,64,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,1024,32,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,1024,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,1024,32,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,65536,0.16905813217163085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,65536,0.34863786697387694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,65536,0.1291925350824992
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,16384,0.08710827032725016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,16384,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,16384,0.03512320121129354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,12288,0.06925653616587321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,12288,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,12288,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,10240,0.06031359831492106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,10240,0.014165332913398743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,10240,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,8192,0.051677866776784265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,8192,0.012458667159080505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,8192,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,7168,0.06430720090866089
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,7168,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,7168,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,6144,0.057070934772491456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,6144,0.01603626708189646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,5120,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,5120,0.05017600059509277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,6144,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,5120,0.014335999886194864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,4096,0.043383467197418216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,4096,0.01204373339811961
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,4096,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,3584,0.039662933349609374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,3584,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,3584,0.010752000411351522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,3072,0.03679573138554891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,3072,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,3072,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,2560,0.03293760021527608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,2048,0.02955946723620097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,2560,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,2560,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,2048,0.007812266548474629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,1536,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,2048,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,1536,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,1536,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,1024,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,1024,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,1024,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,768,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,768,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,768,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,512,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,512,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,512,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,256,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,128,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,256,0.004981333514054617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,256,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,128,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,128,0.004675200084845225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,64,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,64,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,768,32,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,768,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,768,32,0.004502399762471517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,65536,0.04106239875157674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,65536,0.33945598602294924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,65536,0.12916053136189778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,16384,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,16384,0.03549866676330567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,16384,0.08635733127593995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,12288,0.06915413538614909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,12288,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,12288,0.02775040070215861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,10240,0.0609279990196228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,10240,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,10240,0.023756800095240276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,8192,0.052121599515279136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,8192,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,8192,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,7168,0.06365866661071777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,7168,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,6144,0.05710506836573283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,6144,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,6144,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,7168,0.017987199624379478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,5120,0.05010666847229004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,5120,0.010513066252072652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,5120,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,4096,0.043315200010935466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,4096,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,4096,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,3584,0.03949226538340251
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,3584,0.011400533715883891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,3584,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,3072,0.03618133465449015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,3072,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,3072,0.010133333007494609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,2560,0.03338239987691243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,2560,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,2560,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,2048,0.02935466567675273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,2048,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,2048,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,1536,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,1536,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,1536,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,1024,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,1024,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,1024,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,768,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,768,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,768,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,512,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,512,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,512,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,256,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,256,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,128,0.016619732975959776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,128,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,256,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,128,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,64,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,64,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,512,32,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,512,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,512,32,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,65536,0.3293866793314616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,65536,0.12690773010253906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,65536,0.040652799606323245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,16384,0.08628693421681723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,16384,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,16384,0.034577067693074545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,12288,0.06898346741994223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,12288,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,10240,0.06068906784057617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,10240,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,12288,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,10240,0.023756800095240276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,8192,0.05191680192947388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,8192,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,8192,0.019348265727361043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,7168,0.0636245330174764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,7168,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,7168,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,6144,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,6144,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,6144,0.05751466751098633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,5120,0.05010773340861002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,5120,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,5120,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,4096,0.043656531969706217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,4096,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,4096,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,3584,0.040072532494862874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,3584,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,3584,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,3072,0.036246399084726974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,3072,0.006073600053787232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,3072,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,2560,0.03273386756579082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,2560,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,2560,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,2048,0.029934932788213093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,2048,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,2048,0.008050133287906647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,1536,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,1536,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,1536,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,1024,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,1024,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,1024,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,768,0.02105493346850077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,768,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,768,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,512,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,512,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,512,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,256,0.01761173407236735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,256,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,256,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,128,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,128,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,128,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,64,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,64,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,256,32,0.0024533333877722425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,256,32,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,256,32,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,65536,0.023586134115854897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,65536,0.12902186711629232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,65536,0.3225557327270508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,16384,0.08649386564890543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,16384,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,12288,0.06949546337127685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,16384,0.034065067768096924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,12288,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,10240,0.06004053354263306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,12288,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,10240,0.0077802668015162155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,10240,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,8192,0.0522218664487203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,8192,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,8192,0.01925119956334432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,7168,0.06413653294245401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,7168,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,6144,0.056866133213043214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,7168,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,6144,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,6144,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,5120,0.050551466147104894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,5120,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,4096,0.04317440191904704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,4096,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,4096,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,5120,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,3584,0.03969706694285075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,3584,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,3584,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,3072,0.03635199864705403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,3072,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,3072,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,2560,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,2560,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,2560,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,2048,0.029047467311223346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,2048,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,2048,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,1536,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,1536,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,1536,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,1024,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,1024,0.006309333443641663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,1024,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,768,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,768,0.004026666780312856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,768,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,512,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,512,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,512,0.0058335999647776285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,256,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,256,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,128,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,128,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,128,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,64,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,64,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,128,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,128,32,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,128,32,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,65536,0.3175082524617513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,65536,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,65536,0.12912639776865642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,16384,0.08618666330973307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,16384,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,16384,0.03399680058161418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,12288,0.06935893694559733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,12288,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,12288,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,10240,0.06031359831492106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,10240,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,10240,0.023040000597635904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,8192,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,8192,0.0515008012453715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,8192,0.01925119956334432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,7168,0.06331733465194703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,7168,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,7168,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,6144,0.05676373243331909
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,6144,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,6144,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,5120,0.0497322678565979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,5120,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,5120,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,4096,0.04300800164540609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,4096,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,4096,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,3584,0.03979946772257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,3584,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,3584,0.01081706682840983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,3072,0.03652266661326091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,3072,0.004534400006135305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,3072,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,2560,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,2560,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,2560,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,2048,0.0289792001247406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,2048,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,2048,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,1536,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,1536,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,1536,0.025872000058492023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,1024,0.022562134265899658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,1024,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,1024,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,768,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,768,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,512,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,512,0.0034111998975276947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,768,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,256,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,512,0.005454933146635691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,256,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,256,0.004845866560935974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,128,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,128,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,128,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,64,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,64,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,64,32,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,64,32,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,64,32,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,65536,0.3231744130452474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,65536,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,16384,0.08656213283538819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,65536,0.12397226492563884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,16384,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,16384,0.033618132273356124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,12288,0.06877760092417398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,12288,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,12288,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,10240,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,10240,0.06004053354263306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,8192,0.05147306521733602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,10240,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,8192,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,8192,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,7168,0.06331733465194703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,7168,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,7168,0.017781333128611247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,6144,0.056523732344309484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,6144,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,6144,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,5120,0.04966400067011516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,5120,0.004739200075467428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,5120,0.013682132959365845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,4096,0.042973868052164715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,4096,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,4096,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,3584,0.03942399819691976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,3584,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,3584,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,3072,0.03566933472951253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,3072,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,3072,0.010204799969991048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,2560,0.03290453354517619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,2560,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,2560,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,2048,0.029491200049718218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,2048,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,2048,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,1536,0.025838933388392132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,1536,0.007302399973074596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,1536,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,1024,0.022425599892934165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,1024,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,1024,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,768,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,768,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,768,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,512,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,512,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,512,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,256,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,256,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,128,0.016108799974123636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,128,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,128,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,64,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,64,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,64,0.004466133316357931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,80,32,32,0.002418133368094762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,80,32,32,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,80,32,32,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,16384,1.5368874867757163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,16384,3.5444394429524744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,16384,1.870370101928711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,12288,1.43189328511556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,12288,1.1635028839111328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,12288,2.9520212809244795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,10240,1.166540781656901
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,10240,2.2467882792154947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,10240,1.0534912109375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,8192,0.8427520116170248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,8192,0.9349802652994791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,8192,1.7655125935872396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,7168,0.8302602767944336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,7168,0.742843755086263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,7168,1.588428751627604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,6144,0.7176533381144206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,6144,1.2779861450195313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,6144,0.588151486714681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,5120,0.49373865127563477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,5120,0.5989717483520508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,5120,1.0379605611165366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,4096,0.4891968091328939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,4096,0.43134291966756183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,4096,0.9042261123657227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,3584,0.4230826695760091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,3584,0.3533482551574707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,3584,0.7352319717407226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,3072,0.36031147638956706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,3072,0.6397951761881511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,3072,0.3048096021016439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,2560,0.3019434611002604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,2560,0.5329237302144368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,2048,0.24060586293538413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,2560,0.26361172993977866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,2048,0.209169069925944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,2048,0.4295680046081543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,1536,0.18244266510009766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,1536,0.3295520146687826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,1536,0.14390613238016764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,1024,0.04700160026550293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,1024,0.09277439912160237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,768,0.04225493272145589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,1024,0.22572372754414877
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,768,0.07799466451009116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,768,0.14332586924235027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,512,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,512,0.026348799467086792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,512,0.05877759853998819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,256,0.03051519989967346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,256,0.04184746742248535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,256,0.016207999984423318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,128,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,128,0.012422399719556172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,128,0.03638506730397542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,64,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,64,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,64,0.03573760191599528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,65536,32,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,65536,32,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,65536,32,0.03502080043156942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,16384,0.5172512054443359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,16384,0.4311381340026855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,16384,0.891258684794108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,65536,1.5638528188069662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,12288,0.400930118560791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,12288,0.7140010833740235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,65536,1.9693227132161457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,65536,3.5198633829752604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,12288,0.30098772048950195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,10240,0.3258709271748861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,10240,0.27555840810139975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,8192,0.2167125384012858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,10240,0.5703338623046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,8192,0.262826665242513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,8192,0.46076587041219075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,7168,0.24951465924580893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,7168,0.36532481511433923
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,7168,0.19408213297526042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,6144,0.2046293258666992
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,6144,0.08857066631317138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,6144,0.32518825531005857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,5120,0.06079146862030029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,5120,0.26548906962076824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,5120,0.05765120188395182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,4096,0.049422931671142575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,4096,0.2156544049580892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,4096,0.04710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,3584,0.04782079855600993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,3584,0.20596052805582682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,3584,0.042837333679199216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,3072,0.04205226500829061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,3072,0.06410133441289266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,3072,0.036556800206502274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,2560,0.03915093342463176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,2560,0.04457813501358032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,2560,0.032767999172210696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,2048,0.03317546645800273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,2048,0.02693120042483012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,2048,0.036556800206502274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,1536,0.029764266808827718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,1536,0.028702932596206664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,1024,0.025598933299382527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,1536,0.022459733486175536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,1024,0.020411733786265054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,1024,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,768,0.02358293334643046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,768,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,768,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,512,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,512,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,512,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,256,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,256,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,256,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,128,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,128,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,128,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,64,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,64,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,64,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,16384,32,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,16384,32,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,16384,32,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,65536,1.1826517740885416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,16384,0.4052309354146321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,65536,1.5323135375976562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,16384,0.649625587463379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,16384,0.3253898620605469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,12288,0.314572811126709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,65536,2.6305877685546877
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,12288,0.5356885274251302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,12288,0.2523135979970296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,10240,0.23770453135172526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,10240,0.3906559944152832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,10240,0.20896426836649576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,8192,0.1738752047220866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,8192,0.35075413386027016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,8192,0.10417493184407552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,7168,0.08663040002187093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,7168,0.061269334952036535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,7168,0.273305606842041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,6144,0.06331733465194703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,6144,0.24534719785054526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,6144,0.053657599290212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,5120,0.0553984006245931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,5120,0.20002026557922364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,5120,0.04440746704737346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,4096,0.0361130674680074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,4096,0.0457045316696167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,4096,0.05160853465398153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,3584,0.06417066653569539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,3584,0.043383467197418216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,3584,0.03314346671104431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,3072,0.03932160139083862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,3072,0.05437440077463786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,3072,0.028535467386245728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,2560,0.04686506589253743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,2560,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,2560,0.025600000222524004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,2048,0.03133440017700195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,2048,0.03757866621017456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,2048,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,1536,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,1536,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,1536,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,1024,0.024234666426976522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,1024,0.011844266454378765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,1024,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,768,0.022425599892934165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,768,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,768,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,512,0.020889600118001304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,512,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,256,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,512,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,256,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,256,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,128,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,128,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,128,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,64,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,64,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,12288,32,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,12288,32,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,12288,32,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,16384,0.3493546803792318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,65536,1.2997632344563803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,65536,0.9935189565022787
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,16384,0.5711189270019531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,16384,0.26968746185302733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,65536,2.1744298299153644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,12288,0.2663424015045166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,12288,0.43997866312662764
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,12288,0.19302399953206378
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,10240,0.12090026537577311
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,10240,0.36085761388142906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,10240,0.21777067184448243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,8192,0.054440534114837645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,8192,0.0609279990196228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,8192,0.2695850690205892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,7168,0.06714026927947998
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,7168,0.054271999994913736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,7168,0.22729387283325195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,6144,0.06055253346761068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,5120,0.0527018666267395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,6144,0.048128000895182294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,6144,0.22193493843078613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,5120,0.11861013571421306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,5120,0.040445868174235025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,4096,0.032085333267847696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,4096,0.03399680058161418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,4096,0.04498773415883382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,3584,0.04184746742248535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,3584,0.030173865954081218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,3584,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,3072,0.038024532794952395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,3072,0.05355519851048788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,3072,0.026552534103393553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,2560,0.03549866676330567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,2560,0.023893332481384276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,2560,0.04505600134531657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,2048,0.03696639935175578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,2048,0.03078826665878296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,2048,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,1536,0.02754559914271037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,1536,0.029285333553949994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,1024,0.024200532833735147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,1536,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,1024,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,1024,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,768,0.02225493391354879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,768,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,768,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,512,0.020411733786265054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,512,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,512,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,256,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,256,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,256,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,128,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,128,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,128,0.006994133194287618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,64,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,64,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,64,0.0068245331446329755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,10240,32,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,10240,32,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,10240,32,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,65536,0.7999146779378254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,16384,0.4640768051147461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,65536,1.0240341186523438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,16384,0.28074668248494467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,65536,1.7319936116536456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,16384,0.2125482718149821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,12288,0.21101226806640624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,12288,0.35563519795735676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,12288,0.07420586744944255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,10240,0.06816426912943521
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,10240,0.26299734115600587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,8192,0.05010773340861002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,10240,0.05836799939473471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,8192,0.2140501340230306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,8192,0.04491946697235107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,7168,0.06376106739044189
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,7168,0.1869482676188151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,7168,0.04212053219477336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,6144,0.05785599946975708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,6144,0.03648853302001953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,6144,0.0741376002629598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,5120,0.05058559974034628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,5120,0.051677866776784265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,5120,0.03099306623140971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,4096,0.04181333382924397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,4096,0.0417087992032369
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,4096,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,3584,0.04031146764755249
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,3584,0.03594239950180054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,3072,0.036659200986226395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,3584,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,3072,0.030719999472300214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,3072,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,2560,0.032358400026957196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,2560,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,2560,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,2048,0.029183999697367353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,2048,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,2048,0.02225493391354879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,1536,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,1536,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,1536,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,1024,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,1024,0.012457600235939026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,1024,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,768,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,768,0.010513066252072652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,768,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,512,0.01986560026804606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,512,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,512,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,256,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,256,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,256,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,128,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,128,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,128,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,64,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,64,0.0033429334561030067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,64,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,8192,32,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,8192,32,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,8192,32,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,65536,0.9184597651163736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,16384,0.24968533515930175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,65536,0.709836769104004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,65536,1.518353017171224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,16384,0.38328320185343423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,16384,0.18725546201070148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,12288,0.1257813294728597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,12288,0.30539093017578123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,12288,0.06458026568094889
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,10240,0.056866133213043214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,10240,0.05877759853998819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,10240,0.2541909376780192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,8192,0.04706986745198567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,8192,0.18885973294576008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,8192,0.04369066556294759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,7168,0.06113280057907104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,7168,0.09601706663767497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,7168,0.040516265233357746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,6144,0.05215573310852051
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,6144,0.05495466788609823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,6144,0.03560106754302979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,5120,0.0457045316696167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,5120,0.048742401599884036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,5120,0.030105600754419964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,4096,0.040516265233357746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,4096,0.03628373146057129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,4096,0.02491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,3584,0.03833173513412476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,3584,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,3584,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,3072,0.034645334879557295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,3072,0.022459733486175536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,3072,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,2560,0.031674667199452715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,2560,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,2560,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,2048,0.02836479942003886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,2048,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,2048,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,1536,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,1536,0.025326933463414508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,1536,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,1024,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,1024,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,1024,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,768,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,768,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,768,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,512,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,512,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,512,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,256,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,256,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,256,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,128,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,128,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,64,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,64,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,128,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,64,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,7168,32,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,7168,32,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,7168,32,0.005696000158786773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,16384,0.21678080558776855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,65536,0.8085845311482748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,65536,0.6080810546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,65536,1.3209259033203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,16384,0.3273728052775065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,16384,0.10738346576690674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,12288,0.06539946794509888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,12288,0.26715841293334963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,12288,0.06717332998911539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,10240,0.05748053391774496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,10240,0.21927253405253092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,10240,0.055569068590799964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,8192,0.04382719993591309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,8192,0.07768747011820475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,8192,0.04092586835225423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,7168,0.06085973183314005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,7168,0.045873065789540604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,7168,0.04133546749750773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,6144,0.05273600021998087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,6144,0.03809279998143514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,6144,0.032290132840474446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,5120,0.04645546674728394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,5120,0.03304106593132019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,5120,0.02802346746126811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,4096,0.024541866779327393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,4096,0.03915093342463176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,4096,0.023040000597635904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,3584,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,3584,0.03686399857203166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,3584,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,3072,0.0338261326154073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,3072,0.021466666460037233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,3072,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,2560,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,2560,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,2560,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,2048,0.027374933163324993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,2048,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,1536,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,2048,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,1536,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,1536,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,1024,0.022357332706451415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,1024,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,1024,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,768,0.020308266083399452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,768,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,768,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,512,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,512,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,512,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,256,0.017438934246699015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,256,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,256,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,128,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,128,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,128,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,64,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,64,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,6144,32,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,6144,32,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,6144,32,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,65536,0.6971733093261718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,16384,0.1206613302230835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,65536,0.6177450815836589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,65536,1.100561014811198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,16384,0.2711210568745931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,16384,0.08656213283538819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,12288,0.055910400549570714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,12288,0.20333226521809897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,12288,0.0609279990196228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,10240,0.049561599890391037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,10240,0.12083199818929036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,10240,0.04918613433837891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,8192,0.04041386842727661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,8192,0.057036801179250085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,8192,0.040755200386047366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,7168,0.05550080140431722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,7168,0.0362496018409729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,7168,0.037375998497009275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,6144,0.05051733255386352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,6144,0.031197865804036457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,6144,0.02986666758855184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,5120,0.04399786790211995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,5120,0.02696320017178853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,5120,0.02792106668154399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,4096,0.03778560161590576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,4096,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,4096,0.023040000597635904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,3584,0.03546453317006429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,3584,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,3584,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,3072,0.03211946686108907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,3072,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,3072,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,2560,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,2560,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,2560,0.01474240024884542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,2048,0.026927999655405682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,2048,0.012761599818865457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,2048,0.011571199695269267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,1536,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,1536,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,1536,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,1024,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,1024,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,1024,0.009245866537094116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,768,0.020138667027155558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,768,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,768,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,512,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,512,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,512,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,256,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,256,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,256,0.005830400188763936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,128,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,128,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,64,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,64,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,64,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,5120,32,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,5120,32,0.0028266665836175283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,5120,32,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,16384,0.06075733502705892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,65536,0.5882879892985027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,65536,0.4529834747314453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,65536,0.876475715637207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,16384,0.19684693018595378
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,16384,0.042803200085957845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,12288,0.0507861336072286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,12288,0.09311573505401612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,12288,0.036761601765950516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,10240,0.07010986804962158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,10240,0.04556800127029419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,10240,0.030139732360839843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,8192,0.040755200386047366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,8192,0.03884373505910237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,8192,0.02320746580759684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,7168,0.05420373280843099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,7168,0.05058559974034628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,7168,0.02215253313382467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,6144,0.04778666496276855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,6144,0.04181333382924397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,6144,0.01924906571706136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,5120,0.042632532119750974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,5120,0.03679573138554891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,5120,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,4096,0.03679573138554891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,4096,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,4096,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,3584,0.034815998872121175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,3584,0.028330665826797486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,3584,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,3072,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,3072,0.02225493391354879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,2560,0.02908160090446472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,3072,0.012458667159080505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,2560,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,2560,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,2048,0.026316799720128375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,2048,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,2048,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,1536,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,1536,0.013585066795349121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,1536,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,1024,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,1024,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,1024,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,768,0.020002132654190062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,768,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,768,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,512,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,512,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,512,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,256,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,256,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,256,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,128,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,128,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,128,0.0046069333950678505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,64,0.015836800138155617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,64,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,64,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,4096,32,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,4096,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,4096,32,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,65536,0.39830185572306315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,65536,0.7642112096150716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,16384,0.060108800729115806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,65536,0.5319967905680338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,16384,0.19025920232137045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,16384,0.03942399819691976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,12288,0.04990293184916179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,12288,0.034508800506591795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,12288,0.032015999158223465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,10240,0.04437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,10240,0.027613866329193115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,10240,0.027643734216690065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,8192,0.03891199827194214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,8192,0.024199465910593666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,8192,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,7168,0.0527018666267395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,7168,0.04771840174992879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,7168,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,6144,0.04768426815668742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,6144,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,6144,0.03894613186518352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,5120,0.0423253337542216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,5120,0.03382399876912435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,5120,0.015868799885114034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,4096,0.036761601765950516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,4096,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,4096,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,3584,0.03420159816741943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,3584,0.012151466806729634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,3584,0.025804799795150758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,3072,0.03160746693611145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,3072,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,3072,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,2560,0.028842665751775104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,2560,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,2560,0.010035199920336406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,2048,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,2048,0.013818666338920593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,2048,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,1536,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,1536,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,1536,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,1024,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,1024,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,1024,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,768,0.019797333081563315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,768,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,768,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,512,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,512,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,512,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,256,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,256,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,256,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,128,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,128,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,128,0.004570666452248891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,64,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,64,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,64,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3584,32,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3584,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3584,32,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,65536,0.4797098795572917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,65536,0.3410943984985352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,65536,0.6615338643391927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,16384,0.059699201583862306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,16384,0.08908800284067789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,16384,0.03614720106124878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,12288,0.02955946723620097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,12288,0.04911786715189616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,12288,0.03027626673380534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,10240,0.04427093267440796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,10240,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,10240,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,8192,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,8192,0.03853653271993001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,8192,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,7168,0.05236053466796875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,7168,0.04198400179545085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,7168,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,6144,0.04710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,6144,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,6144,0.035870933532714845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,5120,0.04225706656773885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,5120,0.030958932638168336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,5120,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,4096,0.03672746817270915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,4096,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,4096,0.012183466553688049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,3584,0.03420053323109944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,3584,0.024098134040832518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,3584,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,3072,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,3072,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,3072,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,2560,0.029047467311223346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,2560,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,2560,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,2048,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,2048,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,2048,0.008050133287906647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,1536,0.023756800095240276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,1536,0.011400533715883891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,1536,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,1024,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,1024,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,1024,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,768,0.01966080069541931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,768,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,768,0.006002133091290792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,512,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,512,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,512,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,256,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,256,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,256,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,128,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,128,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,128,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,64,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,64,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,64,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,3072,32,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,3072,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,3072,32,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,65536,0.4211711883544922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,16384,0.05928959846496582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,65536,0.5527210871378581
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,65536,0.2874368031819662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,16384,0.05079040129979452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,16384,0.031027199824651082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,12288,0.048844798405965166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,12288,0.03894613186518352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,10240,0.02211839954058329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,10240,0.043622398376464845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,10240,0.03375786542892456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,12288,0.02481493353843689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,8192,0.03850239912668864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,8192,0.027374933163324993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,8192,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,7168,0.05249706506729126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,7168,0.03935573498408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,7168,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,6144,0.047172268231709794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,6144,0.03372373183568318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,6144,0.014711466431617738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,5120,0.04235733350118001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,5120,0.02935466567675273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,5120,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,4096,0.03648853302001953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,4096,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,4096,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,3584,0.033928533395131424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,3584,0.020889600118001304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,3584,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,3072,0.03133440017700195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,3072,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,2560,0.02874026695887248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,3072,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,2560,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,2560,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,2048,0.026180267333984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,2048,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,2048,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,1536,0.023654399315516154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,1536,0.010513066252072652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,1536,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,1024,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,1024,0.00621013343334198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,1024,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,768,0.019692800442377725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,768,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,768,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,512,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,512,0.018293333053588868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,512,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,256,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,256,0.003886933376391729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,256,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,128,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,128,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,64,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,128,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,64,0.0027978666126728057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,64,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2560,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2560,32,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2560,32,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,65536,0.22958079973856607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,16384,0.05928959846496582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,65536,0.43782825469970704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,65536,0.36140267054239905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,16384,0.04061866601308187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,16384,0.026999467611312868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,12288,0.048401065667470294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,12288,0.03160746693611145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,10240,0.043246932824452716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,12288,0.022937599817911783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,10240,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,10240,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,8192,0.03816106716791789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,8192,0.022459733486175536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,8192,0.01570026675860087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,7168,0.05205333232879639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,7168,0.03259733319282532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,7168,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,6144,0.04700160026550293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,6144,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,6144,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,5120,0.04201813141504924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,5120,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,5120,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,4096,0.03635199864705403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,4096,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,4096,0.009898666540781658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,3584,0.03399680058161418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,3584,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,3584,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,3072,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,3072,0.0310591995716095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,3072,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,2560,0.028461867570877077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,2560,0.013174399733543396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,2560,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,2048,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,2048,0.009796266754468281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,2048,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,1536,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,1536,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,1536,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,1024,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,1024,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,1024,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,768,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,768,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,512,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,768,0.005492266515890757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,512,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,256,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,256,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,128,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,256,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,128,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,128,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,64,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,64,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,64,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,2048,32,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,2048,32,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,2048,32,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,65536,0.33013760248819984
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,65536,0.29518505732218425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,65536,0.17083733876546223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,16384,0.058982400099436436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,16384,0.031845333178838094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,16384,0.025975465774536133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,12288,0.025326933463414508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,12288,0.048674134413401286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,12288,0.020753065745035805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,10240,0.04352000157038371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,10240,0.02239146629969279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,10240,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,8192,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,8192,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,8192,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,7168,0.05191680192947388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,7168,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,7168,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,6144,0.04676266511281331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,6144,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,6144,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,5120,0.04157439867655437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,5120,0.020002132654190062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,5120,0.011398399869600933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,4096,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,4096,0.03601066668828328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,4096,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,3584,0.03379199902216594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,3584,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,3584,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,3072,0.031194667021433514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,3072,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,3072,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,2560,0.028398933013280232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,2560,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,2560,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,2048,0.025941334168116253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,2048,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,2048,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,1536,0.023512534300486245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,1536,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,1536,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,1024,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,1024,0.02068480054537455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,1024,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,768,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,768,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,512,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,768,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,512,0.004229333500067393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,512,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,256,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,256,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,256,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,128,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,64,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,128,0.004468266665935516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,128,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,64,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,64,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1536,32,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1536,32,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1536,32,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,65536,0.20800533294677734
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,65536,0.24023040135701498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,65536,0.09099946816762289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,16384,0.05908480087916056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,16384,0.030105600754419964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,16384,0.02539199988047282
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,12288,0.048810664812723795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,12288,0.024678399165471397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,12288,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,10240,0.04352000157038371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,10240,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,10240,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,8192,0.03799039920171102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,8192,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,8192,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,7168,0.051780267556508386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,7168,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,7168,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,6144,0.046728531519571945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,6144,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,6144,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,5120,0.04143786827723185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,5120,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,5120,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,4096,0.03597653309504191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,4096,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,3584,0.03362133502960205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,4096,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,3584,0.01037440001964569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,3072,0.03092479904492696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,3584,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,3072,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,3072,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,2560,0.028501333793004353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,2560,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,2560,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,2048,0.025975465774536133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,2048,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,2048,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,1536,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,1536,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,1536,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,1024,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,1024,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,1024,0.005594666798909505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,768,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,768,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,768,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,512,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,512,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,512,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,256,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,256,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,256,0.004364799956480662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,128,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,128,0.004333866635958353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,128,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,64,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,64,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,1024,32,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,1024,32,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,1024,32,0.0022528000175952912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,65536,0.22568960189819337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,16384,0.058879999319712314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,65536,0.13677226702372233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,65536,0.09079466660817465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,16384,0.018602667252222697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,16384,0.02542613347371419
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,12288,0.048469332853953044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,12288,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,12288,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,10240,0.043246932824452716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,10240,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,10240,0.01781546672185262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,8192,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,8192,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,8192,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,7168,0.05184853474299113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,7168,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,7168,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,6144,0.04652373393376668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,6144,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,6144,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,5120,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,5120,0.041501867771148684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,5120,0.011361066500345867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,4096,0.009488000472386678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,4096,0.03594239950180054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,4096,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,3584,0.03372373183568318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,3584,0.009004799524943034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,3584,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,3072,0.03068586587905884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,3072,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,3072,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,2560,0.028465066353480024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,2560,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,2560,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,2048,0.025702399015426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,2048,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,1536,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,2048,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,1536,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,1536,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,1024,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,1024,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,1024,0.005525333185990652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,768,0.01917866667111715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,768,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,512,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,768,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,512,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,512,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,256,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,256,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,256,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,128,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,128,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,128,0.004227200150489807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,64,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,768,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,768,32,0.015559466679890952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,768,32,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,65536,0.2109781265258789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,65536,0.05485226710637411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,65536,0.09082667032877603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,16384,0.05911893447240194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,16384,0.016723199685414632
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,16384,0.025361067056655882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,12288,0.048469332853953044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,12288,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,12288,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,10240,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,10240,0.04352000157038371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,10240,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,8192,0.03758080005645752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,8192,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,7168,0.051712000370025636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,8192,0.014711466431617738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,7168,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,7168,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,6144,0.04689919948577881
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,6144,0.009485866626103718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,6144,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,5120,0.04119893312454224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,5120,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,5120,0.010988799730936687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,4096,0.03594239950180054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,4096,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,4096,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,3584,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,3584,0.008905599514643352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,3584,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,3072,0.03089066743850708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,3072,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,3072,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,2560,0.028433066606521607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,2560,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,2560,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,2048,0.02553173303604126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,2048,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,2048,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,1536,0.02321066657702128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,1536,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,1536,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,1024,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,1024,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,1024,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,768,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,768,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,768,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,512,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,512,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,512,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,256,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,256,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,128,0.01570026675860087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,128,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,128,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,64,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,64,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,512,32,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,64,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,512,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,512,32,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,65536,0.19565226236979166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,65536,0.029491200049718218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,16384,0.05925546487172445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,65536,0.09062399864196777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,16384,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,12288,0.048640000820159915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,16384,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,12288,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,12288,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,10240,0.04341760079065959
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,10240,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,10240,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,8192,0.03758080005645752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,8192,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,8192,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,7168,0.051540267467498777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,7168,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,7168,0.013585066795349121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,6144,0.04676266511281331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,6144,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,6144,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,5120,0.04150613149007161
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,5120,0.006175999840100607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,5120,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,4096,0.03597653309504191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,4096,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,4096,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,3584,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,3584,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,3584,0.033480532964070636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,3072,0.03065173427263896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,3072,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,3072,0.008156799773375193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,2560,0.028398933013280232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,2560,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,2560,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,2048,0.025497599442799883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,2048,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,2048,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,1536,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,1536,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,1536,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,1024,0.020753065745035805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,1024,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,1024,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,768,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,768,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,768,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,512,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,512,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,512,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,256,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,256,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,256,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,128,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,128,0.01576533317565918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,64,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,64,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,64,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,256,32,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,256,32,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,256,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,65536,0.18865493138631184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,65536,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,65536,0.09076053301493327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,16384,0.058982400099436436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,16384,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,16384,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,12288,0.048674134413401286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,12288,0.007233066856861115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,10240,0.04345173438390096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,12288,0.020205867290496827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,10240,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,10240,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,8192,0.037512532869974774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,8192,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,8192,0.014643200238545737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,7168,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,7168,0.05140479803085327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,7168,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,6144,0.04618240197499593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,6144,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,6144,0.012351999680201214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,5120,0.04160853226979573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,5120,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,5120,0.010990933577219645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,4096,0.03570346832275391
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,4096,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,4096,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,3584,0.03321173389752706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,3584,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,3584,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,3072,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,3072,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,3072,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,2560,0.027955200274785357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,2560,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,2560,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,2048,0.02552853425343831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,2048,0.004329599936803182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,2048,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,1536,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,1536,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,1536,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,1024,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,1024,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,1024,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,768,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,768,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,768,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,512,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,512,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,256,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,512,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,256,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,256,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,128,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,128,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,64,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,64,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,64,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,128,32,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,128,32,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,128,32,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,65536,0.18865493138631184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,65536,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,16384,0.05881173213322958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,16384,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,16384,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,65536,0.09068693319956461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,12288,0.048640000820159915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,12288,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,12288,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,10240,0.04304213523864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,10240,0.017541333039601644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,10240,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,8192,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,8192,0.03747733434041341
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,8192,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,7168,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,7168,0.051677866776784265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,7168,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,6144,0.046250665187835695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,6144,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,6144,0.012180266777674358
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,5120,0.040994131565093996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,5120,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,5120,0.010990933577219645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,4096,0.03597546815872192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,4096,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,4096,0.009313066800435383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,3584,0.0047082667549451195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,3584,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,3584,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,3072,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,3072,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,3072,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,2560,0.028125866254170732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,2560,0.0049450665712356566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,2560,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,2048,0.025326933463414508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,2048,0.006723199784755707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,2048,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,1536,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,1536,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,1536,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,1024,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,1024,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,1024,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,768,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,768,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,512,0.01802026629447937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,768,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,512,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,512,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,256,0.00296426663796107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,256,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,256,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,128,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,128,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,64,0.01566506624221802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,128,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,64,0.002587733417749405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,64,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,64,32,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,64,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,64,32,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,65536,0.18878506024678549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,65536,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,16384,0.05874346494674683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,65536,0.08628906408945719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,12288,0.048401065667470294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,16384,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,16384,0.023859200874964397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,12288,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,12288,0.018874667088190713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,10240,0.04293973445892334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,10240,0.005764266848564148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,10240,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,8192,0.037544532616933184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,8192,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,8192,0.01365226705869039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,7168,0.051336534818013514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,7168,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,7168,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,6144,0.046011734008789065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,6144,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,5120,0.010342400272687275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,5120,0.04078933397928874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,5120,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,6144,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,4096,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,4096,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,4096,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,3584,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,3584,0.03310933311780294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,3584,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,3072,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,3072,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,3072,0.004706133405367533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,2560,0.027955200274785357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,2560,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,2560,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,2048,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,2048,0.004364799956480662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,2048,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,1536,0.02290133237838745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,1536,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,1536,0.004025600105524063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,1024,0.020411733786265054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,1024,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,1024,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,768,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,768,0.018937599658966065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,768,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,512,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,512,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,512,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,256,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,256,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,128,0.015561599532763162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,128,0.002587733417749405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,64,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,64,0.0024853333830833434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,64,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,64,32,32,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,64,32,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,64,32,32,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,16384,1.6772383371988933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,16384,1.85709228515625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,12288,1.1508052825927735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,16384,3.5334485371907554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,12288,2.9042689005533857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,12288,1.4161237080891929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,10240,1.156231435139974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,10240,2.2431060791015627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,10240,1.0498730977376303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,8192,0.925047492980957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,8192,0.8387925465901693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,8192,1.7996458689371746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,7168,0.8227498372395834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,7168,1.5429290771484374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,7168,0.7387466430664062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,6144,0.7091541290283203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,6144,1.411583964029948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,6144,0.5826559702555338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,5120,0.5956949234008789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,5120,0.5311146736145019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,5120,1.033181889851888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,4096,0.47871999740600585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,4096,0.8327509562174479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,4096,0.3940352121988932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,3584,0.3483306566874186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,3584,0.422877852121989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,3584,0.7302143732706706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,3072,0.36399787267049155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,3072,0.6353578567504883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,3072,0.3247786521911621
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,2560,0.3079850514729818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,2560,0.5287562688191731
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,2560,0.25610240300496423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,2048,0.23712426821390786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,2048,0.4674218813578288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,2048,0.20695039431254067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,1536,0.16342825889587403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,1536,0.32655359903971354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,1536,0.13550933202107746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,1024,0.04498773415883382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,1024,0.2414250691731771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,1024,0.07048532962799073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,768,0.06096213261286417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,768,0.039628799756368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,768,0.16288426717122395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,512,0.034918399651845296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,512,0.03979946772257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,512,0.04505600134531657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,256,0.02945706645647685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,256,0.025156267484029132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,256,0.032767999172210696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,128,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,128,0.028194133440653486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,128,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,64,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,64,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,64,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,65536,32,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,65536,32,0.008738133311271667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,65536,32,0.027135999997456868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,16384,0.8715605417887369
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,16384,0.4322250684102376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,16384,0.5037397384643555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,65536,1.7219924926757812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,65536,1.943005879720052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,12288,0.39082667032877605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,12288,0.709666124979655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,65536,3.5108182271321615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,12288,0.3283967971801758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,10240,0.5122389475504557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,10240,0.31866881052652996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,10240,0.27333974838256836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,8192,0.259003734588623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,8192,0.21998933156331382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,8192,0.45076481501261395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,7168,0.24316587448120117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,7168,0.3783679962158203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,6144,0.09130667050679525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,6144,0.3210240046183268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,7168,0.19148799578348796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,6144,0.16151787439982096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,5120,0.059426132837931314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,5120,0.2651072025299072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,5120,0.054340267181396486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,4096,0.04737706581751506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,4096,0.21664427121480306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,3584,0.18616320292154948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,4096,0.043656531969706217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,3584,0.0443391998608907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,3584,0.03979946772257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,3072,0.04174506664276123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,3072,0.06181546847025553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,3072,0.034713598092397054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,2560,0.037205334504445395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,2560,0.03092479904492696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,2048,0.03593920071919759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,2560,0.04450986782709758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,2048,0.032153600454330446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,2048,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,1536,0.028534400463104247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,1536,0.028467200199762982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,1536,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,1024,0.02491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,768,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,1024,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,1024,0.020138667027155558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,768,0.016450132926305136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,768,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,512,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,512,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,512,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,256,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,256,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,256,0.009244799613952637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,128,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,128,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,128,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,64,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,64,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,64,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,16384,32,0.017778132359186807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,16384,32,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,16384,32,0.008152533570925396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,65536,1.4978047688802083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,16384,0.39516159693400066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,16384,0.29453652699788413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,16384,0.64672425587972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,65536,1.169271469116211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,12288,0.30880425771077474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,12288,0.53394775390625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,65536,2.6188458760579425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,10240,0.23043413162231446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,10240,0.38795948028564453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,12288,0.24937814076741538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,8192,0.17489919662475586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,10240,0.20770133336385094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,8192,0.34597546259562173
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,8192,0.06546773513158163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,7168,0.07082666556040446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,6144,0.06007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,7168,0.27303253809611
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,7168,0.05580799976984659
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,6144,0.04829866488774617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,6144,0.24333653450012208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,5120,0.051609599590301515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,5120,0.19616427421569824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,5120,0.04160853226979573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,4096,0.04413439830144246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,4096,0.03345066706339518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,4096,0.059835731983184814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,3584,0.04119893312454224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,3584,0.06266453266143798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,3584,0.03068586587905884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,3072,0.03856426477432251
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,3072,0.05386240084966024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,3072,0.02754559914271037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,2560,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,2560,0.02379093368848165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,2560,0.04573866526285807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,2048,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,2048,0.037307735284169516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,1536,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,2048,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,1536,0.029832533995310467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,1536,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,1024,0.023618133862813313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,1024,0.021367466449737547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,1024,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,768,0.022425599892934165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,768,0.012389333049456278
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,768,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,512,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,512,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,512,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,256,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,128,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,256,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,256,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,128,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,128,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,64,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,64,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,64,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,12288,32,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,12288,32,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,12288,32,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,65536,1.2806143442789713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,16384,0.34157225290934246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,16384,0.2729301452636719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,16384,0.5332991917928059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,65536,0.9759061177571615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,12288,0.2606762727101644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,12288,0.4379989306131999
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,65536,2.162995147705078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,12288,0.20920319557189943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,10240,0.3225599924723307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,10240,0.1421994686126709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,10240,0.21210452715555825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,8192,0.05304319858551025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,8192,0.057137068112691244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,8192,0.2868543942769369
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,7168,0.06498986482620239
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,7168,0.05147306521733602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,7168,0.22760106722513834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,6144,0.0575488011042277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,6144,0.20193279584248863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,6144,0.04369066556294759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,5120,0.1118890682856242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,5120,0.03816106716791789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,5120,0.0501418670018514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,4096,0.03089066743850708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,4096,0.04375893274943034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,4096,0.03126613299051921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,3584,0.04071466525395711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,3584,0.06072320143381754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,3584,0.028330665826797486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,3072,0.037478399276733396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,3072,0.05280426740646362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,3072,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,2560,0.033928533395131424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,2560,0.022459733486175536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,2560,0.04519253174463908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,2048,0.03020799954732259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,2048,0.03638613224029541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,2048,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,1536,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,1536,0.029320534070332843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,1536,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,1024,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,1024,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,1024,0.01300266683101654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,768,0.022016000747680665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,768,0.01241919994354248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,768,0.011707733074824016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,512,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,512,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,512,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,256,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,256,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,256,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,128,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,128,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,128,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,64,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,64,0.003479466587305069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,64,0.006791466474533081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,10240,32,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,10240,32,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,10240,32,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,65536,0.9979904174804688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,16384,0.2701984087626139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,16384,0.2168490727742513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,16384,0.42758827209472655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,12288,0.20244480768839518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,65536,0.7888213475545247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,65536,1.7348256429036457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,12288,0.354747740427653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,12288,0.07219200134277344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,10240,0.25972053209940593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,10240,0.05379199981689453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,8192,0.04526079893112182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,10240,0.0605183998743693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,8192,0.04215466578801473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,8192,0.23719253540039062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,7168,0.06068906784057617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,7168,0.03894613186518352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,7168,0.1826133410135905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,6144,0.035225598017374675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,6144,0.05478399991989136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,6144,0.062156800429026285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,5120,0.04788906574249267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,5120,0.050995198885599766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,5120,0.029146667321523028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,4096,0.03973120053609212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,4096,0.04092586835225423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,4096,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,3584,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,3584,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,3584,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,3072,0.03423466682434082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,3072,0.02986666758855184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,3072,0.01986560026804606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,2560,0.031197865804036457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,2560,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,2560,0.025462400913238526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,2048,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,2048,0.020475733280181884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,2048,0.027818665901819868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,1536,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,1536,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,1536,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,1024,0.02225493391354879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,1024,0.01030613382657369
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,1024,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,768,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,768,0.010410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,768,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,512,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,512,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,512,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,256,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,256,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,256,0.006450133522351582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,128,0.01641386648019155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,128,0.004368000229199727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,128,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,64,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,64,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,64,0.0057322666049003605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,8192,32,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,8192,32,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,8192,32,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,65536,0.7041706720987956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,16384,0.24033279418945314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,16384,0.3791189193725586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,16384,0.17544533411661783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,65536,1.5093077341715495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,12288,0.07874560356140137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,65536,0.8844970703125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,12288,0.30286506017049153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,12288,0.062498132387797035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,10240,0.05881173213322958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,10240,0.22869226137797036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,10240,0.051950931549072266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,8192,0.042871467272440594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,8192,0.189030392964681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,8192,0.04102826515833537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,7168,0.05847040017445883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,7168,0.08444586594899496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,7168,0.03768320083618164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,6144,0.05259946584701538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,6144,0.05147306521733602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,6144,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,5120,0.04618240197499593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,5120,0.04317333300908406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,5120,0.02757546703020732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,4096,0.03853653271993001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,4096,0.034713598092397054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,4096,0.022937599817911783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,3584,0.03696639935175578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,3584,0.03133440017700195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,3584,0.02081706722577413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,3072,0.03324586749076843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,3072,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,3072,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,2560,0.030513066053390502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,2560,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,2048,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,2560,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,2048,0.013717333475748697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,2048,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,1536,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,1536,0.011980799833933513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,1536,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,1024,0.021845332781473794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,1024,0.010069333513577779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,1024,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,768,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,768,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,768,0.007402666906515758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,512,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,512,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,512,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,256,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,256,0.0048096001148223875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,256,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,128,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,128,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,128,0.005835733314355215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,64,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,64,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,64,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,7168,32,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,7168,32,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,7168,32,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,65536,0.7867701212565105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,16384,0.20688212712605797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,16384,0.1047210693359375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,65536,0.6643029530843099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,16384,0.3204437255859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,65536,1.2926976521809896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,12288,0.057275732358296715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,12288,0.25978879928588866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,10240,0.050722134113311765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,10240,0.19667627016703287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,12288,0.05939199924468994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,10240,0.04949333270390828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,8192,0.06918826897939047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,8192,0.04089173475901286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,8192,0.0421887993812561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,7168,0.05631999969482422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,7168,0.05256533225377401
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,6144,0.04526079893112182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,6144,0.048742401599884036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,6144,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,7168,0.03645439942677815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,5120,0.04358506600062052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,5120,0.03822933435440064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,5120,0.02672213315963745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,4096,0.0373418649037679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,4096,0.031402667363484696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,4096,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,3584,0.03587413231531779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,3584,0.02788693308830261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,3584,0.019797333081563315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,3072,0.03283626635869344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,3072,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,3072,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,2560,0.029491200049718218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,2560,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,2560,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,2048,0.012457600235939026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,2048,0.012862933675448099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,2048,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,1536,0.024064000447591147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,1536,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,1536,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,1024,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,1024,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,1024,0.0095551997423172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,768,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,768,0.009010133147239686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,512,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,768,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,512,0.0056639999151229855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,512,0.007333333293596904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,256,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,256,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,256,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,128,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,128,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,64,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,64,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,128,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,6144,32,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,6144,32,0.0029290666182835894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,6144,32,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,16384,0.0774826685587565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,65536,0.6033407847086589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,16384,0.2930346806844076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,65536,1.0925397237141927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,65536,0.6696618398030598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,16384,0.07014400164286295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,12288,0.051780267556508386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,10240,0.0451530655225118
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,12288,0.05765120188395182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,10240,0.045636268456776936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,10240,0.12513279914855957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,8192,0.03781973520914714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,8192,0.05683199961980184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,8192,0.037546666463216145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,12288,0.2164031982421875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,7168,0.03505493402481079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,7168,0.052940801779429114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,6144,0.04741119941075643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,7168,0.03307519952456157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,6144,0.02921813329060872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,6144,0.029149866104125975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,5120,0.041843199729919435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,5120,0.02450773318608602
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,5120,0.025634133815765382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,4096,0.03628373146057129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,4096,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,3584,0.034406399726867674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,4096,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,3584,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,3584,0.018602667252222697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,3072,0.03143253326416016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,3072,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,3072,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,2560,0.029149866104125975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,2560,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,2560,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,2048,0.02590720057487488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,2048,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,2048,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,1536,0.024166399240493776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,1536,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,1536,0.010750933488210043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,1024,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,1024,0.010648533701896667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,768,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,1024,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,768,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,512,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,512,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,512,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,768,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,256,0.01736746629079183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,256,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,256,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,128,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,128,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,128,0.0060032000144322716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,64,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,64,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,5120,32,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,5120,64,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,5120,32,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,65536,0.8679082870483399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,65536,0.4456106821695964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,65536,0.5623125076293946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,16384,0.0392192006111145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,16384,0.2150335947672526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,16384,0.05683199961980184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,12288,0.04723946650822957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,12288,0.06997333367665609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,12288,0.03307519952456157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,10240,0.04263039827346802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,10240,0.048503466447194415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,10240,0.027374933163324993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,8192,0.037239468097686766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,8192,0.04078720013300578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,8192,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,7168,0.05208746592203776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,7168,0.04608000119527181
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,7168,0.020476800203323365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,6144,0.04655786752700806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,6144,0.03829760154088338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,6144,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,5120,0.04096000194549561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,5120,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,5120,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,4096,0.03587413231531779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,4096,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,4096,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,3584,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,3584,0.024098134040832518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,3584,0.012219732999801636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,3072,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,3072,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,3072,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,2560,0.028427733977635698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,2560,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,2560,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,2048,0.025600000222524004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,2048,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,2048,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,1536,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,1536,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,1024,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,1536,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,1024,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,1024,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,768,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,768,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,768,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,512,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,512,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,256,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,512,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,256,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,256,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,128,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,128,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,128,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,64,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,64,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,4096,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,4096,32,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,32,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,4096,64,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,65536,0.5088255882263184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,65536,0.7630165100097657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,65536,0.3915775934855143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,16384,0.05717333157857259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,16384,0.19131733576456705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,16384,0.03635199864705403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,12288,0.04833279848098755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,12288,0.031675734122594196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,12288,0.03109546701113383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,10240,0.04205226500829061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,10240,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,10240,0.026897066831588747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,8192,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,8192,0.03700053294499715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,8192,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,7168,0.05130240122477213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,7168,0.04403200149536133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,7168,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,6144,0.04638719956080119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,6144,0.03648853302001953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,6144,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,5120,0.04040640195210775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,5120,0.03089066743850708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,4096,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,5120,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,4096,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,4096,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,3584,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,3584,0.0233130673567454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,3584,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,3072,0.03089066743850708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,3072,0.010683733224868774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,2560,0.028194133440653486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,2560,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,3072,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,2560,0.009522133072217305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,2048,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,2048,0.013242666920026144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,2048,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,1536,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,1536,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,1536,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,1024,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,1024,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,1024,0.020753065745035805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,768,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,768,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,768,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,512,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,512,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,512,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,256,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,256,0.004914133250713349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,256,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,128,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,128,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,128,0.004502399762471517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,64,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,64,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,64,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3584,32,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3584,32,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3584,32,0.0026911998788515727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,65536,0.4507306734720866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,65536,0.35638612111409507
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,65536,0.6488405227661133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,16384,0.057036801179250085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,16384,0.08468480110168457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,16384,0.03296746611595154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,12288,0.04703573385874431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,12288,0.028569600979487103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,12288,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,10240,0.04181333382924397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,10240,0.023005867004394533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,10240,0.024473599592844644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,8192,0.0370688001314799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,8192,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,8192,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,7168,0.05120000044504801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,7168,0.04041386842727661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,7168,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,6144,0.04614826838175456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,6144,0.032631466786066696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,6144,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,5120,0.040345601240793866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,5120,0.02863466739654541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,4096,0.03570346832275391
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,4096,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,4096,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,3584,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,5120,0.013926399747530618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,3584,0.02183893322944641
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,3584,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,3072,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,3072,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,3072,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,2560,0.028501333793004353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,2560,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,2560,0.015729066729545594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,2048,0.025565866629282636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,2048,0.012385066350301106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,2048,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,1536,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,1536,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,1536,0.007233066856861115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,1024,0.02218666672706604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,1024,0.006141866743564606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,1024,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,768,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,768,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,768,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,512,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,512,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,512,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,256,0.016724266608556113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,256,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,256,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,128,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,128,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,128,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,64,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,64,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,64,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,3072,32,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,3072,32,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,3072,32,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,65536,0.39458131790161133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,65536,0.283682123819987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,65536,0.5435391743977864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,16384,0.056695465246836343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,16384,0.049491198857625325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,16384,0.028637866179148357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,12288,0.04669440189997355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,12288,0.037853864828745525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,12288,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,10240,0.04205226500829061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,10240,0.03273386756579082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,10240,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,8192,0.037063467502594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,8192,0.02696533401807149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,8192,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,7168,0.03631786505381267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,7168,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,6144,0.046011734008789065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,6144,0.03041279911994934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,7168,0.050995198885599766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,6144,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,5120,0.04020586808522542
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,5120,0.026587732632954914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,5120,0.012898133198420206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,4096,0.03549866676330567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,4096,0.020172800620396933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,4096,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,3584,0.03324586749076843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,3584,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,3584,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,3072,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,3072,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,3072,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,2560,0.028194133440653486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,2560,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,2048,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,2560,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,2048,0.011805867155392964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,2048,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,1536,0.0233130673567454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,1536,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,1536,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,1024,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,1024,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,1024,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,768,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,768,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,768,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,512,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,512,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,512,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,256,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,256,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,256,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,128,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,128,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,128,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,64,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,64,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,64,0.004264533519744873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2560,32,0.01590079963207245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2560,32,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2560,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,16384,0.0566271980603536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,65536,0.3300352096557617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,65536,0.2249728043874105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,65536,0.42932694753011064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,16384,0.039287467797597245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,16384,0.026589866479237872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,12288,0.046487466494242353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,12288,0.021845332781473794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,10240,0.04160853226979573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,10240,0.026999467611312868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,12288,0.03057919939359029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,10240,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,8192,0.03669333457946777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,8192,0.02228906750679016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,8192,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,7168,0.05113173325856527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,7168,0.0315391997496287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,6144,0.0457045316696167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,6144,0.024951465924580894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,7168,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,6144,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,5120,0.03997013171513875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,5120,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,5120,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,4096,0.03549866676330567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,4096,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,4096,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,3584,0.03297280073165894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,3584,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,3584,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,3072,0.030446932713190718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,3072,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,3072,0.008805333574612936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,2560,0.028091732660929365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,2560,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,2048,0.025597866376241046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,2560,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,1536,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,2048,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,2048,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,1536,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,1536,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,1024,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,1024,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,1024,0.005727999905745188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,768,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,768,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,768,0.005494399865468343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,512,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,512,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,512,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,256,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,256,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,128,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,128,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,64,0.01563093364238739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,128,0.004226133227348328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,256,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,64,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,64,0.004195199906826019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,2048,32,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,2048,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,2048,32,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,65536,0.1575935999552409
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,65536,0.25889813105265297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,16384,0.056081068515777585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,65536,0.32109225591023766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,16384,0.03123093247413635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,16384,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,12288,0.04642133315404256
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,12288,0.02529279987017314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,12288,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,10240,0.04154026508331299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,10240,0.021600000063578286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,10240,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,8192,0.03641706705093384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,8192,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,8192,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,7168,0.05044906536738077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,7168,0.027272532383600872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,7168,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,6144,0.04522666533788045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,6144,0.022835199038187662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,6144,0.012663466731707254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,5120,0.040447998046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,5120,0.01962560017903646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,5120,0.011400533715883891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,4096,0.035396265983581546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,4096,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,4096,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,3584,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,3584,0.009282132983207703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,3584,0.03297066688537598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,3072,0.03068586587905884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,3072,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,2560,0.02815999984741211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,3072,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,2560,0.01102186640103658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,2560,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,2048,0.025497599442799883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,2048,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,2048,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,1536,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,1536,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,1536,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,1024,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,1024,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,1024,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,768,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,768,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,512,0.01819093426068624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,768,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,512,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,512,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,256,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,256,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,256,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,128,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,128,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,64,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,128,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,64,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1536,32,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1536,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1536,32,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,65536,0.20698453585306803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,65536,0.22391467094421386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,16384,0.056012801329294835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,65536,0.0912384033203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,16384,0.02263039946556091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,16384,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,12288,0.04648960034052531
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,12288,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,12288,0.020411733786265054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,10240,0.04126720031102498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,10240,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,8192,0.03645439942677815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,10240,0.017778132359186807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,8192,0.014847999811172486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,8192,0.015733333428700765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,7168,0.05058559974034628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,7168,0.015320533514022827
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,7168,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,6144,0.045090134938557944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,6144,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,5120,0.040345601240793866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,5120,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,6144,0.012387200196584066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,5120,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,4096,0.035293865203857425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,3584,0.03297280073165894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,4096,0.010067199667294819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,4096,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,3584,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,3584,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,3072,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,3072,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,3072,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,2560,0.028398933013280232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,2560,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,2560,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,2048,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,2048,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,1536,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,2048,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,1536,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,1536,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,1024,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,1024,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,1024,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,768,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,768,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,512,0.018153599898020425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,768,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,512,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,256,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,512,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,256,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,128,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,128,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,64,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,128,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,64,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,1024,32,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,1024,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,1024,32,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,65536,0.19647146860758463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,65536,0.09076053301493327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,16384,0.05604693492253622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,65536,0.10584746996561686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,16384,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,16384,0.02529279987017314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,12288,0.046250665187835695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,12288,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,12288,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,10240,0.04126720031102498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,10240,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,8192,0.03601066668828328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,10240,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,8192,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,8192,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,7168,0.05038079818089804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,7168,0.012727466225624085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,7168,0.013755733768145243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,6144,0.04526079893112182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,6144,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,5120,0.039628799756368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,6144,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,5120,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,5120,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,4096,0.03512320121129354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,4096,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,4096,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,3584,0.03266026576360066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,3584,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,3072,0.030138667424519854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,3584,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,3072,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,3072,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,2560,0.028262400627136232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,2560,0.00737066666285197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,2560,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,2048,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,2048,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,1536,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,2048,0.0067552000284194945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,1536,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,1536,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,1024,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,1024,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,1024,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,768,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,768,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,768,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,512,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,512,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,512,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,256,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,256,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,256,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,128,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,128,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,64,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,128,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,64,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,768,32,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,768,32,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,768,32,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,65536,0.040140799681345624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,65536,0.17612800598144532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,16384,0.05604693492253622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,65536,0.09076053301493327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,16384,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,16384,0.025321600834528606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,12288,0.04618240197499593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,12288,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,12288,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,10240,0.04102826515833537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,10240,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,10240,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,8192,0.03597653309504191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,8192,0.00757120003302892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,8192,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,7168,0.05034666856129964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,7168,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,7168,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,6144,0.04526079893112182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,6144,0.009074133634567261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,6144,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,5120,0.03997013171513875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,5120,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,5120,0.011878400047620138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,4096,0.03505493402481079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,4096,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,4096,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,3584,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,3584,0.008701866865158081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,3072,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,3584,0.03259733319282532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,3072,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,3072,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,2560,0.02768213351567586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,2560,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,2560,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,2048,0.0056991999348004665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,2048,0.02501973311106364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,1536,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,2048,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,1536,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,1536,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,1024,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,1024,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,1024,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,768,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,768,0.004265599946180979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,512,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,768,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,512,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,512,0.004877866804599762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,256,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,256,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,256,0.004333866635958353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,128,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,128,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,64,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,64,0.002593066543340683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,128,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,64,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,512,32,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,512,32,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,512,32,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,65536,0.17640105883280438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,65536,0.027409066756566364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,65536,0.09062399864196777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,16384,0.05611519813537598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,16384,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,16384,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,12288,0.046284798781077066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,12288,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,12288,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,10240,0.04126720031102498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,10240,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,10240,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,8192,0.03618133465449015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,8192,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,8192,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,7168,0.050619733333587644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,7168,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,7168,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,6144,0.045021867752075194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,6144,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,6144,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,5120,0.040140799681345624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,5120,0.010956799983978272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,5120,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,4096,0.03495253324508667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,4096,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,4096,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,3584,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,3584,0.03256319959958394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,3584,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,3072,0.03061760067939758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,3072,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,3072,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,2560,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,2560,0.0049781332413355505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,2560,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,2048,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,2048,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,2048,0.024985599517822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,1536,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,1536,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,1536,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,1024,0.020272000630696615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,1024,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,1024,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,768,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,768,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,512,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,768,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,512,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,512,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,256,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,128,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,256,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,64,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,128,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,64,0.003066666672627131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,64,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,256,32,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,256,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,256,32,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,65536,0.175820795694987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,65536,0.01621119976043701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,65536,0.09072639942169189
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,16384,0.055978667736053464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,16384,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,16384,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,12288,0.04631893237431844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,12288,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,12288,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,10240,0.04102826515833537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,10240,0.0067221333583196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,10240,0.017505067586898803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,8192,0.03614720106124878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,8192,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,8192,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,7168,0.05003946622212728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,7168,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,7168,0.013477333386739097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,6144,0.0447104016939799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,6144,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,6144,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,5120,0.040277334054311116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,5120,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,5120,0.010813867052396137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,4096,0.03484799861907959
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,4096,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,4096,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,3584,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,3584,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,3072,0.03007146716117859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,3584,0.032529066006342575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,3072,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,3072,0.008156799773375193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,2560,0.02768213351567586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,2560,0.004845866560935974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,2560,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,2048,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,2048,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,2048,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,1536,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,1536,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,1536,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,1024,0.020103466510772706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,1024,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,1024,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,768,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,768,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,512,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,768,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,512,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,512,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,256,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,256,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,256,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,128,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,128,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,128,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,64,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,64,0.0023829333484172823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,64,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,128,32,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,128,32,0.00262719988822937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,128,32,0.004129066566626231
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,65536,0.17592320442199708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,65536,0.012900267044703165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,65536,0.09048746426900228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,16384,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,16384,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,16384,0.05587626695632934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,12288,0.046353065967559816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,12288,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,12288,0.020067199071248373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,10240,0.04109653234481812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,10240,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,10240,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,8192,0.03642026583353679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,8192,0.005420800050099691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,8192,0.014643200238545737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,7168,0.04993173281351725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,7168,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,6144,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,6144,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,6144,0.0448853333791097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,7168,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,5120,0.03932160139083862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,5120,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,5120,0.01071679989496867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,4096,0.0347818652788798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,4096,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,4096,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,3584,0.03239253362019857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,3584,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,3072,0.030139732360839843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,3584,0.005047466854254405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,3072,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,3072,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,2560,0.028125866254170732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,2560,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,2560,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,2048,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,2048,0.024707200129826863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,2048,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,1536,0.022562134265899658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,1536,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,1536,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,1024,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,1024,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,1024,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,768,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,768,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,512,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,768,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,512,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,512,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,256,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,256,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,256,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,128,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,128,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,64,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,64,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,128,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,64,32,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,64,32,0.002418133368094762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,64,32,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,65536,0.1755136013031006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,65536,0.01242453356583913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,65536,0.08488960266113281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,16384,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,16384,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,12288,0.046011734008789065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,16384,0.05604693492253622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,12288,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,10240,0.04143786827723185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,10240,0.00621013343334198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,12288,0.018666666746139527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,10240,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,8192,0.03587413231531779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,8192,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,7168,0.04969813426335652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,8192,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,7168,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,7168,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,6144,0.04491946697235107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,6144,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,6144,0.011400533715883891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,5120,0.03911679983139038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,5120,0.004840533435344696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,4096,0.03512320121129354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,4096,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,5120,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,4096,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,3584,0.03242666721343994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,3584,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,3584,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,3072,0.030003199974695843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,3072,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,3072,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,2560,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,2560,0.004466133316357931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,2560,0.027409066756566364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,2048,0.004229333500067393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,2048,0.02491413354873657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,2048,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,1536,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,1536,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,1536,0.012458667159080505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,1024,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,768,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,1024,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,768,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,1024,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,768,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,512,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,512,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,256,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,512,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,256,0.004299733539422353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,256,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,128,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,128,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,128,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,64,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,32,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,48,32,32,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,48,32,64,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,48,32,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,16384,1.8176682790120442
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,16384,1.6791893005371095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,16384,3.5343360900878906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,12288,1.1455829620361329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,12288,1.3801130930582681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,10240,1.1316224416097005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,12288,2.8966570536295575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,10240,2.281847381591797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,10240,1.053218142191569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,8192,0.7663957595825195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,8192,1.6738304138183593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,8192,0.9079423904418945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,7168,0.8140799840291342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,7168,1.5103317260742188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,7168,0.7392255783081054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,6144,0.7015082677205403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,6144,1.265561548868815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,6144,0.580300776163737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,5120,0.531387742360433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,5120,1.0268330891927084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,5120,0.5857621510823567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,4096,0.4253354708353679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,4096,0.8301567713419595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,4096,0.47209812800089523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,3584,0.4181333223978679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,3584,0.7239679972330729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,3584,0.3448831876118978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,3072,0.36017494201660155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,3072,0.6343338648478191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,3072,0.32245759963989257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,2560,0.30235306421915686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,2560,0.5239466667175293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,2560,0.25135787328084314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,2048,0.24664746920267738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,2048,0.421614933013916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,1536,0.13015039761861164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,2048,0.2029184023539225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,1536,0.32242345809936523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,1536,0.09157973130544027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,1024,0.2190336068471273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,1024,0.04017173449198405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,1024,0.048093867301940915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,768,0.034167468547821045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,768,0.04164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,768,0.09888426462809244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,512,0.0315391997496287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,512,0.031300266583760575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,512,0.03969706694285075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,256,0.023552000522613525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,256,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,128,0.02208426594734192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,256,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,128,0.01136959989865621
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,128,0.02003413240114848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,64,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,64,0.006687999765078227
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,64,0.019796266158421835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,65536,32,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,65536,32,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,65536,32,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,16384,0.48421548207600906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,16384,0.9046357472737631
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,65536,1.5902015686035156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,12288,0.37621758778889974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,65536,1.890884272257487
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,12288,0.7053653081258138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,16384,0.44189014434814455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,65536,3.454498036702474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,12288,0.3354624112447103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,10240,0.2866175969441732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,10240,0.5559637069702148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,8192,0.2237781365712484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,10240,0.2789375940958659
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,7168,0.22070612907409667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,8192,0.24750080108642578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,8192,0.415231990814209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,7168,0.374613348642985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,7168,0.18152106602986653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,6144,0.04344853162765503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,6144,0.1282047986984253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,6144,0.3156309445699056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,5120,0.0506880005200704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,5120,0.031778132915496825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,5120,0.26036906242370605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,4096,0.042973868052164715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,4096,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,3584,0.04167679945627849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,4096,0.20974933306376137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,3584,0.1784490744272868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,3584,0.02430293361345927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,3072,0.03682986497879028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,3072,0.06748159726460776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,3072,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,2560,0.034577067693074545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,2560,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,2560,0.04925440152486165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,2048,0.029491200049718218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,2048,0.04160853226979573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,2048,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,1536,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,1536,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,1536,0.03099306623140971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,1024,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,1024,0.02255786657333374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,1024,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,768,0.02105600039164225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,768,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,768,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,512,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,512,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,512,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,256,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,256,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,256,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,128,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,128,0.004911999901135763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,128,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,64,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,64,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,64,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,16384,32,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,16384,32,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,16384,32,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,16384,0.38024425506591797
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,65536,1.2014933268229167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,16384,0.3342677434285482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,16384,0.6861130396525066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,12288,0.2938197453816732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,65536,2.5996971130371094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,65536,1.4547285715738931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,12288,0.5368490854899088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,12288,0.25190186500549316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,10240,0.23886507352193198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,10240,0.3827370643615723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,10240,0.20991999308268228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,8192,0.1422335942586263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,8192,0.04580693244934082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,8192,0.32563199996948244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,7168,0.06253226598103842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,7168,0.04085760116577149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,7168,0.27026774088541666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,6144,0.05451093514760336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,6144,0.23883093198140465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,6144,0.03246080080668132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,5120,0.04768213431040446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,5120,0.19394559860229493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,5120,0.02792106668154399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,4096,0.040004265308380124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,4096,0.07092906634012858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,4096,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,3584,0.037512532869974774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,3584,0.06072320143381754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,3584,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,3072,0.034474666913350424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,3072,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,3072,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,2560,0.031846400101979574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,2560,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,2048,0.028228267033894854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,2560,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,2048,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,1536,0.024678399165471397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,2048,0.01395840048789978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,1536,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,1536,0.012151466806729634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,1024,0.02177706758181254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,1024,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,1024,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,768,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,768,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,768,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,512,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,512,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,512,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,256,0.017646932601928712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,256,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,256,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,128,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,128,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,64,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,128,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,64,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,64,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,12288,32,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,12288,32,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,12288,32,0.00573333352804184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,65536,1.013756815592448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,16384,0.32307198842366536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,16384,0.5306367874145508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,65536,1.2202303568522135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,16384,0.28293120066324867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,12288,0.24818347295125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,65536,2.3204180399576826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,12288,0.44315306345621747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,10240,0.1737386703491211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,12288,0.21357226371765137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,10240,0.131822935740153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,10240,0.3325610796610514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,8192,0.29399038950602213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,8192,0.040447998046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,8192,0.04119893312454224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,7168,0.055668266614278164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,7168,0.037102933724721274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,7168,0.266376527150472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,6144,0.05259946584701538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,6144,0.21674666404724122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,6144,0.03194560011227925
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,5120,0.04427093267440796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,5120,0.02788693308830261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,5120,0.11202560265858967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,4096,0.037956265608469646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,4096,0.06847146352132162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,4096,0.023005867004394533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,3584,0.03607893387476603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,3584,0.06072320143381754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,3584,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,3072,0.03262613415718078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,3072,0.024200532833735147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,3072,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,2560,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,2560,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,2560,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,2048,0.026897066831588747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,2048,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,2048,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,1536,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,1536,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,1024,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,1024,0.010342400272687275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,1536,0.023995733261108397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,1024,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,768,0.0081535999973615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,768,0.01986560026804606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,768,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,512,0.01853440006573995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,512,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,512,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,256,0.01699413259824117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,256,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,128,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,128,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,128,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,64,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,64,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,64,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,10240,32,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,10240,32,0.0029994666576385496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,10240,32,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,16384,0.2584575970967611
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,16384,0.4583413441975911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,65536,0.8038058598836264
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,65536,1.725610605875651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,65536,0.9556650797526041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,12288,0.1544874668121338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,16384,0.22493866284688316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,12288,0.35440425872802733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,12288,0.05908480087916056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,10240,0.05034666856129964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,10240,0.03031040032704671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,10240,0.2870272000630697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,8192,0.039901868502298994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,8192,0.21387947400410973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,7168,0.05410133202870687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,8192,0.023411200443903605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,7168,0.1670453389485677
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,7168,0.022835199038187662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,6144,0.0506880005200704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,6144,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,6144,0.04174506664276123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,5120,0.04440746704737346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,5120,0.01658453345298767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,5120,0.03816106716791789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,4096,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,4096,0.03659093379974365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,4096,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,3584,0.03594239950180054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,3584,0.026180267333984375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,3584,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,3072,0.03194453318913777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,3072,0.01242453356583913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,3072,0.024678399165471397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,2560,0.029320534070332843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,2560,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,2560,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,2048,0.02635093331336975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,2048,0.013926399747530618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,2048,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,1536,0.023859200874964397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,1536,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,1024,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,1024,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,1024,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,1536,0.01269760032494863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,768,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,768,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,768,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,512,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,512,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,512,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,256,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,256,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,256,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,128,0.01590293347835541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,128,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,128,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,64,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,64,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,64,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,8192,32,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,8192,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,8192,32,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,16384,0.23043413162231446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,16384,0.4028042793273926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,65536,0.7693311691284179
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,16384,0.19206825892130536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,65536,0.8517973581949869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,65536,1.5020373026529947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,12288,0.05403306484222412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,12288,0.3044682820638021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,12288,0.03351893424987793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,10240,0.04532906611760457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,10240,0.028808534145355225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,10240,0.2511189301808675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,8192,0.03781973520914714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,8192,0.20517546335856118
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,8192,0.02177706758181254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,7168,0.05290453433990479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,7168,0.020411733786265054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,7168,0.06546773513158163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,6144,0.04652373393376668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,6144,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,6144,0.038362665971120195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,5120,0.04229120016098022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,5120,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,5120,0.03556693394978841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,4096,0.03543039957682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,4096,0.02461013396581014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,4096,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,3584,0.03505493402481079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,3584,0.024780799945195518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,3584,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,3072,0.03099306623140971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,3072,0.01102186640103658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,2560,0.02955946723620097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,3072,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,2560,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,2560,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,2048,0.025668267409006757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,2048,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,2048,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,1536,0.02321066657702128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,1536,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,1024,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,1024,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,1024,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,768,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,1536,0.0077461332082748415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,768,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,768,0.006312533219655354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,512,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,512,0.005423999826113383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,512,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,256,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,256,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,256,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,128,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,128,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,128,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,64,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,64,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,64,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,7168,32,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,7168,32,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,7168,32,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,16384,0.0556714653968811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,16384,0.34515625635782876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,65536,0.5939882914225261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,16384,0.1964373270670573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,65536,1.2782250722249349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,65536,0.7418197631835938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,12288,0.04669440189997355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,12288,0.26309866905212403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,12288,0.030958932638168336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,10240,0.0444757342338562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,10240,0.026316799720128375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,10240,0.2045952002207438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,8192,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,8192,0.07393279870351156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,8192,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,7168,0.0506880005200704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,7168,0.04700160026550293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,7168,0.020102399587631225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,6144,0.04693333307902019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,6144,0.03638293345769246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,6144,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,5120,0.04089173475901286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,5120,0.014572800199190775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,5120,0.03324586749076843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,4096,0.03433813254038493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,4096,0.023040000597635904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,4096,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,3584,0.03348480065663655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,3584,0.023859200874964397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,3584,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,3072,0.029934932788213093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,3072,0.010376532872517902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,2560,0.02815893292427063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,3072,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,2560,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,2048,0.025156267484029132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,2560,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,2048,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,2048,0.008088533580303193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,1536,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,1536,0.007439999779065449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,1024,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,1024,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,1024,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,1536,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,768,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,768,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,768,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,512,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,512,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,512,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,256,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,256,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,256,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,128,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,128,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,128,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,64,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,64,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,64,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,6144,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,6144,32,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,6144,32,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,16384,0.05184853474299113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,65536,0.5026133219401042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,16384,0.25579519271850587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,65536,0.6340949376424153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,65536,1.0863957722981772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,16384,0.03031040032704671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,12288,0.21483519872029624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,12288,0.04379306634267171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,12288,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,10240,0.03932160139083862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,10240,0.10888533592224121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,10240,0.02286613384882609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,8192,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,8192,0.05666026671727499
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,8192,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,7168,0.04761600097020467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,7168,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,7168,0.042973868052164715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,6144,0.04416853189468384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,6144,0.03389439980189006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,6144,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,5120,0.03863893349965413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,5120,0.031191466252009074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,5120,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,4096,0.03321173389752706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,4096,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,4096,0.01102186640103658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,3584,0.032358400026957196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,3584,0.02263039946556091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,3584,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,3072,0.02911146680514018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,3072,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,3072,0.021538132429122926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,2560,0.027409066756566364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,2560,0.016344533363978068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,2560,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,2048,0.024434133370717367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,2048,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,2048,0.007713066538174947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,1536,0.02225493391354879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,1536,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,1024,0.020172800620396933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,1024,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,1024,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,1536,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,768,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,768,0.007539199789365132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,768,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,512,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,512,0.00573333352804184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,512,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,256,0.01641279955705007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,128,0.01570026675860087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,256,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,128,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,128,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,64,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,64,0.015287466843922935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,64,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,5120,32,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,5120,32,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,5120,32,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,16384,0.045533867677052815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,65536,0.5285888036092122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,65536,0.44100265502929686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,16384,0.2213205337524414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,65536,0.8050346374511719
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,16384,0.026862933238347368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,12288,0.03942399819691976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,12288,0.05287253459294637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,12288,0.02327679991722107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,10240,0.04352000157038371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,10240,0.03621546824773152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,10240,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,8192,0.03092479904492696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,8192,0.03160746693611145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,8192,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,7168,0.046011734008789065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,7168,0.03553280035654704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,7168,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,6144,0.042393600940704344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,6144,0.028637866179148357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,6144,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,5120,0.037273601690928145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,5120,0.026589866479237872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,5120,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,4096,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,4096,0.032085333267847696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,4096,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,3584,0.03126613299051921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,3584,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,3584,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,3072,0.028330665826797486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,3072,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,2560,0.026521599292755126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,2560,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,2560,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,2048,0.024029866854349772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,3072,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,2048,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,2048,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,1536,0.02187946637471517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,1536,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,1536,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,1024,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,1024,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,768,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,1024,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,768,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,768,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,512,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,512,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,512,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,256,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,256,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,256,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,128,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,128,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,128,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,64,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,64,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,4096,32,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,4096,32,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,4096,32,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,16384,0.04437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,16384,0.026009599367777508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,16384,0.19309226671854657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,65536,0.3878229459126791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,65536,0.695364252726237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,65536,0.47745707829793294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,12288,0.03805546760559082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,12288,0.0457045316696167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,10240,0.03853653271993001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,12288,0.021808000405629475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,10240,0.034747731685638425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,10240,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,8192,0.03037866751352946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,8192,0.02877440055211385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,8192,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,7168,0.04468053181966146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,7168,0.034850132465362546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,7168,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,6144,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,5120,0.03662506739298503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,6144,0.025838933388392132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,6144,0.04061866601308187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,5120,0.02430293361345927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,5120,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,4096,0.031948800881703696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,4096,0.009796266754468281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,4096,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,3584,0.03054933349291484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,3584,0.009213866790135701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,3584,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,3072,0.027784534295399982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,3072,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,3072,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,2560,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,2560,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,2560,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,2048,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,2048,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,2048,0.007095466554164887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,1536,0.021603200833002725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,1536,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,1536,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,1024,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,1024,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,1024,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,768,0.018461867173512777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,768,0.005493333439032236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,768,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,512,0.017437867323557534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,512,0.005493333439032236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,256,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,256,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,256,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,128,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,128,0.003068800022204717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,128,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,64,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,64,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,64,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3584,32,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3584,32,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3584,32,0.004363733530044556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,65536,0.4198741277058919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,16384,0.04437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,65536,0.33297065099080403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,16384,0.0605183998743693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,65536,0.594158935546875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,16384,0.025804799795150758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,12288,0.03768320083618164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,12288,0.04089173475901286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,12288,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,10240,0.034065067768096924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,10240,0.03444053332010905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,10240,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,8192,0.03020799954732259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,8192,0.025804799795150758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,8192,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,7168,0.04413333336512248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,7168,0.029934932788213093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,6144,0.040482131640116374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,7168,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,6144,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,6144,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,5120,0.03628373146057129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,5120,0.02177706758181254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,5120,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,4096,0.031982932488123575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,4096,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,3584,0.03051519989967346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,3584,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,4096,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,3584,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,3072,0.02805759906768799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,3072,0.015594666202863058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,3072,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,2560,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,2560,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,2560,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,2048,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,2048,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,2048,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,1536,0.021538132429122926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,1536,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,1536,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,1024,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,1024,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,1024,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,768,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,768,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,768,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,512,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,512,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,512,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,256,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,256,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,256,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,128,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,128,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,128,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,64,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,64,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,64,0.004262400170167288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,3072,32,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,3072,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,3072,32,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,65536,0.3666261355082194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,65536,0.27381760279337564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,16384,0.04450986782709758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,16384,0.039009066422780354
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,65536,0.5039445241292317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,16384,0.025733333826065064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,12288,0.037511467933654785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,12288,0.03065173427263896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,12288,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,10240,0.0339626669883728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,10240,0.0289792001247406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,10240,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,8192,0.029969066381454468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,8192,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,8192,0.014916266997655234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,7168,0.04427093267440796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,7168,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,7168,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,6144,0.040277334054311116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,6144,0.022801067431767783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,6144,0.01269760032494863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,5120,0.03618133465449015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,5120,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,5120,0.011194666226704914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,4096,0.031573333342870075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,4096,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,3584,0.030481066306432086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,4096,0.010001066327095031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,3584,0.014711466431617738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,3584,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,3072,0.02792106668154399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,3072,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,3072,0.008361599842707316
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,2560,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,2560,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,2560,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,2048,0.023654399315516154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,2048,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,2048,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,1536,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,1536,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,1536,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,1024,0.02013333241144816
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,1024,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,1024,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,768,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,768,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,768,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,512,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,512,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,256,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,512,0.005013333261013031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,256,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,256,0.004465066889921824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,128,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,128,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,128,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,64,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,64,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,64,0.004262400170167288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2560,32,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2560,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2560,32,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,65536,0.3092138608296712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,65536,0.39980373382568357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,65536,0.220467201868693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,16384,0.04392960071563721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,16384,0.031402667363484696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,16384,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,12288,0.03700053294499715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,12288,0.025600000222524004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,12288,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,10240,0.03341653347015381
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,10240,0.02239146629969279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,10240,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,8192,0.030003199974695843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,8192,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,8192,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,7168,0.04444160064061482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,7168,0.020477867126464842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,7168,0.013753599921862283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,6144,0.040174933274586995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,6144,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,6144,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,5120,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,5120,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,5120,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,4096,0.03150506615638733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,4096,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,4096,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,3584,0.03027626673380534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,3584,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,3584,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,3072,0.027812266349792482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,3072,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,3072,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,2560,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,2560,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,2560,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,2048,0.023483733336130776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,2048,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,2048,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,1536,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,1536,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,1536,0.006206933160622915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,1024,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,1024,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,1024,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,768,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,768,0.005182933310667674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,768,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,512,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,512,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,512,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,256,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,256,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,256,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,128,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,128,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,64,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,64,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,32,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,2048,64,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,2048,32,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,2048,32,0.004435199995835622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,16384,0.043724799156188966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,65536,0.24326826731363932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,65536,0.1060863971710205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,65536,0.30887254079182946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,16384,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,12288,0.03682986497879028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,16384,0.025361067056655882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,12288,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,12288,0.02068480054537455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,10240,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,10240,0.017643733819325765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,8192,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,10240,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,8192,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,8192,0.01437013347943624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,7168,0.044202665487925216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,7168,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,7168,0.013755733768145243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,6144,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,6144,0.03952639897664388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,5120,0.03607893387476603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,5120,0.010679466525713603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,6144,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,5120,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,4096,0.031573333342870075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,4096,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,4096,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,3584,0.029422932863235475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,3584,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,3584,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,3072,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,3072,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,3072,0.008292266726493835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,2560,0.025634133815765382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,2560,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,2560,0.00757120003302892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,2048,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,2048,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,1536,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,2048,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,1536,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,1536,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,1024,0.019387733936309815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,1024,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,1024,0.005525333185990652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,768,0.018500266472498576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,512,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,512,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,768,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,512,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,768,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,256,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,256,0.0032042667269706728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,256,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,128,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,128,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,64,0.01532373329003652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,128,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,64,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,64,0.0025578667720158894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1536,32,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1536,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1536,32,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,65536,0.17087146441141765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,65536,0.1998165289560954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,16384,0.043144532044728595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,65536,0.0905898650487264
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,16384,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,12288,0.03696639935175578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,16384,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,12288,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,12288,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,10240,0.03348480065663655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,10240,0.012795733412106833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,8192,0.029491200049718218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,10240,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,8192,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,8192,0.014674133062362671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,7168,0.04389546712239583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,7168,0.011092266440391541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,7168,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,6144,0.039389868577321366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,6144,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,5120,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,6144,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,5120,0.011025066177050274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,5120,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,4096,0.031402667363484696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,4096,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,4096,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,3584,0.02959360082944234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,3584,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,3584,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,3072,0.027613866329193115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,3072,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,2560,0.025668267409006757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,3072,0.008255999783674877
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,2560,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,2560,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,2048,0.02362026572227478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,2048,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,2048,0.005801600217819214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,1536,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,1536,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,1024,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,1536,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,1024,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,1024,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,768,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,768,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,768,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,512,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,512,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,256,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,256,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,512,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,256,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,128,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,128,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,128,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,64,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,64,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,1024,32,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,1024,32,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,1024,32,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,65536,0.09492479960123698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,65536,0.1259178638458252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,65536,0.0904533306757609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,16384,0.04317440191904704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,16384,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,16384,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,12288,0.012117333213488261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,12288,0.03665706713994344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,12288,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,10240,0.010683733224868774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,10240,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,10240,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,8192,0.029388799269994097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,8192,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,8192,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,7168,0.043622398376464845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,7168,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,7168,0.01344533363978068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,6144,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,6144,0.03932160139083862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,5120,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,6144,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,5120,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,4096,0.03136853377024333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,5120,0.010853333274523418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,4096,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,4096,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,3584,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,3584,0.029764266808827718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,3072,0.02754559914271037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,3584,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,3072,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,3072,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,2560,0.025600000222524004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,2560,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,2560,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,2048,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,2048,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,2048,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,1536,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,1536,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,1024,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,1024,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,1536,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,1024,0.005389866729577383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,768,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,768,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,768,0.005220266679922739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,512,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,512,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,256,0.01576533317565918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,256,0.003340800106525421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,256,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,512,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,128,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,128,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,128,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,64,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,768,32,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,768,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,65536,0.12595199743906657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,768,32,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,65536,0.030003199974695843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,65536,0.09048746426900228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,16384,0.04334933360417684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,16384,0.02505386670430501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,16384,0.010752000411351522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,12288,0.03652266661326091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,12288,0.020138667027155558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,10240,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,12288,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,10240,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,10240,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,8192,0.029283199707667035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,8192,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,7168,0.04389546712239583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,8192,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,7168,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,7168,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,6144,0.03969706694285075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,6144,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,6144,0.012117333213488261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,5120,0.03515733480453491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,5120,0.010990933577219645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,4096,0.03133440017700195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,5120,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,4096,0.009282132983207703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,4096,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,3584,0.029730133215586346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,3584,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,3584,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,3072,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,3072,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,2560,0.025497599442799883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,3072,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,2560,0.0077802668015162155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,2560,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,2048,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,2048,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,2048,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,1536,0.021501866976420085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,1536,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,1024,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,1024,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,1536,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,1024,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,768,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,768,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,768,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,512,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,512,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,512,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,256,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,256,0.015801599621772765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,256,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,128,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,128,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,128,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,64,0.015324800213177999
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,64,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,512,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,512,32,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,512,32,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,65536,0.12581546306610109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,65536,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,65536,0.09048746426900228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,16384,0.04355413516362508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,16384,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,12288,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,16384,0.025156267484029132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,12288,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,12288,0.036556800206502274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,10240,0.03321173389752706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,10240,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,10240,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,8192,0.02921813329060872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,8192,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,8192,0.006446933249632518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,7168,0.04375893274943034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,7168,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,7168,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,6144,0.039901868502298994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,6144,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,6144,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,5120,0.03546453317006429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,5120,0.010922666390736897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,5120,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,4096,0.031229867537816362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,4096,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,4096,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,3584,0.029661866029103597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,3584,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,3584,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,3072,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,3072,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,3072,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,2560,0.025497599442799883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,2560,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,2560,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,2048,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,2048,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,2048,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,1536,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,1536,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,1024,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,1024,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,1024,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,1536,0.006140799820423126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,768,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,768,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,768,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,512,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,512,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,512,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,256,0.015867732961972556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,256,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,256,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,128,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,128,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,128,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,64,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,64,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,256,32,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,256,32,0.0024864000578721365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,256,32,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,65536,0.12567893664042157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,65536,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,65536,0.09051946798960367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,16384,0.04358826478322347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,16384,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,16384,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,12288,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,12288,0.03648746808369954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,12288,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,10240,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,10240,0.032867199182510375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,10240,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,8192,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,8192,0.02921813329060872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,8192,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,7168,0.04345173438390096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,7168,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,7168,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,6144,0.03959466616312663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,6144,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,6144,0.0049781332413355505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,5120,0.03556693394978841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,5120,0.005187200009822845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,5120,0.010683733224868774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,4096,0.03139946659406026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,4096,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,4096,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,3584,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,3584,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,3584,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,3072,0.027374933163324993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,3072,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,3072,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,2560,0.025565866629282636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,2560,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,2048,0.02327893376350403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,2048,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,2560,0.004948266843954722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,2048,0.0067552000284194945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,1536,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,1536,0.004126933217048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,1536,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,1024,0.01914773384730021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,1024,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,1024,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,768,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,768,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,768,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,512,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,512,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,512,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,256,0.015904000401496886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,128,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,256,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,128,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,64,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,128,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,128,32,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,128,32,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,65536,0.1253376007080078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,65536,0.012458667159080505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,65536,0.09089706738789877
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,16384,0.04345173438390096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,16384,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,16384,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,12288,0.036761601765950516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,12288,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,12288,0.019797333081563315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,10240,0.03310933311780294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,10240,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,10240,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,8192,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,8192,0.0291157325108846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,8192,0.01437013347943624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,7168,0.04352000157038371
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,7168,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,7168,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,6144,0.039662933349609374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,6144,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,6144,0.012117333213488261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,5120,0.03543039957682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,5120,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,4096,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,4096,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,4096,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,5120,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,3584,0.029491200049718218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,3584,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,3584,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,3072,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,2560,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,3072,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,3072,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,2560,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,2560,0.007473066449165344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,2048,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,2048,0.004535466432571411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,2048,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,1536,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,1536,0.004021333406368891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,1536,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,1024,0.019006933768590292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,1024,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,1024,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,768,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,768,0.00446720023949941
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,768,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,512,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,512,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,512,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,256,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,256,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,256,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,128,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,128,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,64,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,64,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,64,32,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,64,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,64,32,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,65536,0.12417706648508708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,65536,0.012219732999801636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,65536,0.08468053340911866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,16384,0.04304213523864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,16384,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,16384,0.006172800064086914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,12288,0.0361130674680074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,12288,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,12288,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,10240,0.03259733319282532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,10240,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,10240,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,8192,0.0289792001247406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,8192,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,8192,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,7168,0.04345173438390096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,7168,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,6144,0.039082666238149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,7168,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,6144,0.0055285334587097164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,6144,0.011332266529401143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,5120,0.03519146839777629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,5120,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,4096,0.031163734197616578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,5120,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,3584,0.029320534070332843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,4096,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,4096,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,3584,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,3584,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,3072,0.02693120042483012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,3072,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,3072,0.007334400216738384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,2560,0.025121066967646283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,2048,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,2560,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,2560,0.006854400038719177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,2048,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,2048,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,1536,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,1536,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,1536,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,1024,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,1024,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,1024,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,768,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,768,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,768,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,512,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,512,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,512,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,256,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,256,0.015732266505559287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,128,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,128,0.002624000112215678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,64,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,32,32,32,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,32,32,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,32,32,32,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,16384,1.5661738077799479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,16384,1.7886549631754558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,12288,1.1780095418294272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,16384,3.256866200764974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,12288,1.3630804697672525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,12288,2.5970687866210938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,10240,1.1197099049886068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,10240,2.0661589304606123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,10240,1.0708309173583985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,8192,0.857702382405599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,8192,1.6481280008951824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,8192,0.8950784047444662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,7168,0.804692268371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,7168,1.3101739247639974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,7168,0.7444138844807943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,6144,0.6911317189534505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,6144,1.138857650756836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,6144,0.6423893610636393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,5120,0.5781845092773438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,5120,0.5369173049926758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,5120,1.0260138829549155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,4096,0.4633599917093913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,4096,0.43062613805135086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,4096,0.7545173645019532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,3584,0.4107946713765462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,3584,0.3530069351196289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,3584,0.6617088317871094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,3072,0.3531434694925944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,3072,0.3225941340128581
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,3072,0.6186325073242187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,2560,0.29699414571126304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,2560,0.520089594523112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,2560,0.25442986488342284
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,2048,0.22923946380615234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,2048,0.3829749425252279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,2048,0.20278612772623697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,1536,0.1013760010401408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,1536,0.2889024098714193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,1536,0.06017706791559855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,1024,0.0362496018409729
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,1024,0.19459412892659506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,1024,0.02815999984741211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,768,0.031778132915496825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,768,0.06584320068359376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,768,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,512,0.024641066789627075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,512,0.02805759906768799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,512,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,256,0.022425599892934165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,256,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,256,0.013755733768145243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,128,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,128,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,128,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,64,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,64,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,64,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,65536,32,0.01925119956334432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,65536,32,0.004811733464399974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,65536,32,0.011942399541536967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,16384,0.4689578692118327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,16384,0.4378965377807617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,16384,0.8045226414998373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,65536,1.5272618611653646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,12288,0.36601174672444664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,65536,3.230754089355469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,12288,0.6559402465820312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,65536,1.8427562713623047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,12288,0.2982912063598633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,10240,0.2731349309285482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,10240,0.29805227915445964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,8192,0.24067413012186684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,10240,0.5200554529825847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,8192,0.2175999959309896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,8192,0.42775894800821945
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,7168,0.22797652880350747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,7168,0.3292501449584961
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,7168,0.17322667439778644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,6144,0.08639146486918131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,6144,0.03590826590855916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,6144,0.28917760848999025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,5120,0.04741119941075643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,5120,0.23709012667338053
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,5120,0.028228267033894854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,4096,0.03969706694285075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,4096,0.1911466598510742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,4096,0.022323199113210044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,3584,0.03758080005645752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,3584,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,3584,0.16694614092508953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,3072,0.033928533395131424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,3072,0.04423679908116658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,3072,0.01853440006573995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,2560,0.03126399914423625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,2560,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,2560,0.03577173153559367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,2048,0.028194133440653486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,2048,0.024098134040832518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,2048,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,1536,0.024676267306009927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,1536,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,1536,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,1024,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,1024,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,1024,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,768,0.020307199160257975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,768,0.010410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,768,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,512,0.018396800756454466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,512,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,512,0.008254933357238769
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,256,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,256,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,128,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,256,0.006413866579532623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,128,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,128,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,64,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,64,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,64,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,16384,32,0.015871999661127727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,16384,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,16384,32,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,16384,0.3684351921081543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,65536,1.1526143391927084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,16384,0.6449493408203125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,16384,0.33129812876383463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,65536,1.40141232808431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,65536,2.4142506917317705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,12288,0.28334080378214516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,12288,0.5068117459615071
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,12288,0.25238186518351235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,10240,0.2304298718770345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,8192,0.07243093649546305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,10240,0.3703808148701986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,10240,0.20722346305847167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,8192,0.32580267588297523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,8192,0.055528533458709714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,7168,0.0553984006245931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,7168,0.26978985468546546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,7168,0.03290453354517619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,6144,0.04901546637217204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,6144,0.029388799269994097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,6144,0.22773760159810386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,5120,0.04379306634267171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,5120,0.1866751988728841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,5120,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,4096,0.04198400179545085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,4096,0.037307735284169516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,4096,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,3584,0.03846826553344727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,3584,0.03477973143259684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,3584,0.020171733697255452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,3072,0.03218773404757182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,3072,0.03259733319282532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,3072,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,2560,0.029183999697367353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,2560,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,2560,0.029934932788213093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,2048,0.026282666126887004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,2048,0.020029866695404054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,2048,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,1536,0.023411200443903605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,1536,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,1536,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,1024,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,1024,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,1024,0.009180800120035807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,768,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,768,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,768,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,512,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,512,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,512,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,256,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,256,0.004402133325735727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,256,0.005936000247796377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,128,0.01562879979610443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,128,0.003719466676314672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,128,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,64,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,64,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,64,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,12288,32,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,12288,32,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,12288,32,0.005320533116658529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,16384,0.27607040405273436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,16384,0.31225172678629554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,65536,0.976964251200358
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,16384,0.5023743947347005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,65536,1.1722751617431642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,65536,2.015505091349284
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,12288,0.23770453135172526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,12288,0.4096682548522949
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,10240,0.15564799308776855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,12288,0.21364053090413412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,10240,0.1276245355606079
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,10240,0.29593601226806643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,8192,0.03345066706339518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,8192,0.24388052622477213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,7168,0.050995198885599766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,8192,0.037034666538238524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,7168,0.031982932488123575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,7168,0.2083434740702311
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,6144,0.04614826838175456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,6144,0.028091732660929365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,6144,0.1826133410135905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,5120,0.04177920023600261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,5120,0.023991467555363973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,5120,0.09806506633758545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,4096,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,4096,0.03563520113627116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,3584,0.033587201436360674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,4096,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,3584,0.02624853253364563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,3584,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,3072,0.03065173427263896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,3072,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,3072,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,2560,0.028228267033894854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,2560,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,2560,0.02146880030632019
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,2048,0.025463465849558515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,2048,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,2048,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,1536,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,1536,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,1536,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,1024,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,1024,0.008873599767684936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,1024,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,768,0.019010132551193236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,768,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,768,0.0081535999973615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,512,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,512,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,512,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,256,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,256,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,256,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,128,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,128,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,128,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,64,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,64,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,64,0.005153066913286845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,10240,32,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,10240,32,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,10240,32,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,65536,0.9174357096354167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,16384,0.2510165373484293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,65536,0.8013482411702475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,16384,0.39574187596639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,65536,1.609762191772461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,16384,0.2233344078063965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,12288,0.10123946666717529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,12288,0.3245727856953939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,12288,0.03502080043156942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,10240,0.04116479953130086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,10240,0.2358954588572184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,8192,0.03433813254038493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,10240,0.028091732660929365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,8192,0.1958229382832845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,8192,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,7168,0.05051733255386352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,7168,0.16609279314676922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,7168,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,6144,0.03775146802266439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,6144,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,6144,0.04519253174463908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,5120,0.03997013171513875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,5120,0.030139732360839843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,5120,0.015733333428700765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,4096,0.034030934174855546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,4096,0.012663466731707254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,4096,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,3584,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,3584,0.023108265797297158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,3584,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,3072,0.02955946723620097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,3072,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,3072,0.011025066177050274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,2560,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,2560,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,2560,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,2048,0.02505386670430501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,2048,0.013209600249926248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,2048,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,1536,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,1536,0.01181013286113739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,1536,0.007474133372306823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,1024,0.020411733786265054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,1024,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,1024,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,768,0.019381332397460937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,768,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,768,0.006071466704209646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,512,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,512,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,512,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,256,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,256,0.0039893334110577905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,256,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,128,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,128,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,128,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,64,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,64,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,64,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,8192,32,0.003031466652949651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,8192,32,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,8192,32,0.004328533510367076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,65536,0.8097792307535807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,16384,0.2217642625172933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,65536,0.7845888137817383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,16384,0.18721812566121418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,65536,1.406054433186849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,16384,0.3457706769307455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,12288,0.04430506626764934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,12288,0.28020054499308267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,12288,0.02877440055211385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,10240,0.03949226538340251
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,10240,0.2229248046875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,10240,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,8192,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,8192,0.1710026741027832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,8192,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,7168,0.04785279830296834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,7168,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,7168,0.05522666772206625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,6144,0.034745601812998454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,6144,0.04293973445892334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,6144,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,5120,0.03915093342463176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,5120,0.029315199454625445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,5120,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,4096,0.03334826628367106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,4096,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,3584,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,4096,0.01948480010032654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,3584,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,3584,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,3072,0.02959360082944234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,3072,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,3072,0.010342400272687275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,2560,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,2560,0.00921493371327718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,2560,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,2048,0.024366933107376098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,2048,0.012151466806729634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,2048,0.008087466657161712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,1536,0.022016000747680665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,1536,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,1536,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,1024,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,1024,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,1024,0.006413866579532623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,768,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,768,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,768,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,512,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,512,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,512,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,256,0.00382080003619194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,256,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,256,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,128,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,128,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,64,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,64,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,128,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,64,0.004329599936803182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,7168,32,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,7168,32,0.0027978666126728057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,7168,32,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,16384,0.15138133366902667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,65536,0.6784341176350911
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,65536,1.2009471893310546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,65536,0.7004159927368164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,16384,0.059869865576426186
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,16384,0.29129387537638346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,12288,0.03942399819691976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,12288,0.24125439325968423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,12288,0.026555732885996504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,10240,0.022523732980092366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,10240,0.03631786505381267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,10240,0.198963197072347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,8192,0.03099306623140971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,8192,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,8192,0.042734932899475095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,7168,0.04700160026550293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,7168,0.034918399651845296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,6144,0.03160746693611145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,7168,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,6144,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,6144,0.04078933397928874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,5120,0.03741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,5120,0.025116799275080363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,5120,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,4096,0.03259733319282532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,4096,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,3584,0.03065173427263896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,4096,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,3584,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,3072,0.028501333793004353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,3584,0.01037440001964569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,3072,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,3072,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,2560,0.026692267258961993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,2560,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,2560,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,2048,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,2048,0.011534933249155681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,2048,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,1536,0.021640533208847047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,1536,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,1536,0.006723199784755707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,1024,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,1024,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,1024,0.005767466624577841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,768,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,768,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,768,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,512,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,512,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,512,0.004947199920813243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,256,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,256,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,256,0.00453653335571289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,128,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,128,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,128,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,64,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,64,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,64,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,6144,32,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,6144,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,6144,32,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,16384,0.042700799306233723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,65536,0.5620394388834635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,65536,0.9985024134318033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,65536,0.5923840204874675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,16384,0.2478421370188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,16384,0.027818665901819868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,12288,0.03758080005645752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,12288,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,12288,0.19861760139465331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,10240,0.03372373183568318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,10240,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,10240,0.10779306888580323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,8192,0.028945066531499225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,8192,0.034474666913350424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,8192,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,7168,0.045431466897328694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,7168,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,6144,0.03956053256988525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,7168,0.015666133165359496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,6144,0.026550400257110595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,6144,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,5120,0.036177066961924235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,5120,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,5120,0.012458667159080505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,4096,0.03160746693611145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,4096,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,3584,0.02955946723620097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,3584,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,4096,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,3584,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,3072,0.028228267033894854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,3072,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,3072,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,2560,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,2560,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,2560,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,2048,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,2048,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,2048,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,1536,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,1536,0.00904319981733958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,1536,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,1024,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,1024,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,1024,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,768,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,768,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,768,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,512,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,512,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,512,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,256,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,256,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,128,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,128,0.003684266656637192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,128,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,64,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,5120,32,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,5120,32,0.00279573326309522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,5120,32,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,65536,0.4842837333679199
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,65536,0.7982101440429688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,65536,0.44967254002888996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,16384,0.04143786827723185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,16384,0.19418452580769857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,16384,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,12288,0.035531731446584065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,12288,0.062225067615509035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,12288,0.020753065745035805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,10240,0.031880533695220946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,10240,0.03853653271993001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,10240,0.01747093399365743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,8192,0.03037866751352946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,8192,0.028603732585906982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,8192,0.01344319979349772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,7168,0.029969066381454468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,7168,0.043110398451487224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,7168,0.013550933202107748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,6144,0.03935573498408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,6144,0.02392639915148417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,6144,0.011980799833933513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,5120,0.035396265983581546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,5120,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,5120,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,4096,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,4096,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,4096,0.030958932638168336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,3584,0.029047467311223346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,3584,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,3584,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,3072,0.027540266513824463
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,3072,0.01437013347943624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,3072,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,2560,0.02529279987017314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,2560,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,2560,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,2048,0.02300373315811157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,2048,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,2048,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,1536,0.022016000747680665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,1536,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,1536,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,1024,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,1024,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,1024,0.005182933310667674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,768,0.018055466810862224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,768,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,768,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,512,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,512,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,512,0.004572799801826477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,256,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,256,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,256,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,128,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,128,0.015593600273132325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,128,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,64,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,64,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,4096,32,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,64,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,4096,32,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,4096,32,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,65536,0.4315135955810547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,65536,0.3962538719177246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,65536,0.7053994496663412
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,16384,0.02228906750679016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,16384,0.04133546749750773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,16384,0.17076692581176758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,12288,0.04430506626764934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,12288,0.034815998872121175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,10240,0.032085333267847696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,12288,0.019693867365519205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,10240,0.03775146802266439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,10240,0.016586666305859886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,8192,0.02959253390630086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,8192,0.028433066606521607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,7168,0.042496001720428465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,8192,0.013106133540471396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,7168,0.028842665751775104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,7168,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,6144,0.03901439905166626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,6144,0.011332266529401143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,6144,0.025939200321833295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,5120,0.03491520086924235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,5120,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,5120,0.01023466686407725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,4096,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,4096,0.031129600604375203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,4096,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,3584,0.03068586587905884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,3584,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,3584,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,3072,0.02682346701622009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,3072,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,3072,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,2560,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,2560,0.025361067056655882
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,2560,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,2048,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,2048,0.009796266754468281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,2048,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,1536,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,1536,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,1024,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,1536,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,1024,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,1024,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,768,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,768,0.005118933320045471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,768,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,512,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,512,0.004299733539422353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,512,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,256,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,256,0.0035818666219711304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,256,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,128,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,128,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,64,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,64,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3584,32,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3584,32,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3584,32,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,65536,0.37925545374552405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,65536,0.5950805028279622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,65536,0.3129685401916504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,16384,0.06614933411280313
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,16384,0.04133546749750773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,16384,0.021943465868632
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,12288,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,12288,0.042461868127187094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,12288,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,10240,0.031568000713984176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,10240,0.03607893387476603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,10240,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,8192,0.02836479942003886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,8192,0.028262400627136232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,7168,0.042905600865681966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,8192,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,7168,0.027409066756566364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,7168,0.012117333213488261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,6144,0.025804799795150758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,6144,0.03915093342463176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,5120,0.03502080043156942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,6144,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,5120,0.010069333513577779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,5120,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,4096,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,4096,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,4096,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,3584,0.02934933304786682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,3584,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,3072,0.027167999744415285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,3584,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,3072,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,3072,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,2560,0.025326933463414508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,2560,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,2560,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,2048,0.023000532388687135
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,2048,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,1536,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,2048,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,1536,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,1536,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,1024,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,1024,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,1024,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,768,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,768,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,768,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,512,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,512,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,512,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,256,0.016177067160606386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,256,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,256,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,128,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,128,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,64,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,128,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,64,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,64,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,3072,32,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,3072,32,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,3072,32,0.00402453343073527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,65536,0.32709973653157554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,65536,0.2863445281982422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,65536,0.49380693435668943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,16384,0.05024426778157552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,16384,0.04102826515833537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,16384,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,12288,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,12288,0.041262932618459064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,12288,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,10240,0.031778132915496825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,10240,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,10240,0.03375786542892456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,8192,0.028433066606521607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,8192,0.026862933238347368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,8192,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,7168,0.0425983985265096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,7168,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,7168,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,6144,0.03826346794764201
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,6144,0.024132267634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,5120,0.034918399651845296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,6144,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,5120,0.018500266472498576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,5120,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,4096,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,4096,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,4096,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,3584,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,3584,0.029217066367467244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,3072,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,3584,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,3072,0.011944533387819926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,2560,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,3072,0.007439999779065449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,2560,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,2560,0.011707733074824016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,2048,0.023108265797297158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,2048,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,2048,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,1536,0.0210207998752594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,1536,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,1536,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,1024,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,1024,0.005628799895445505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,1024,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,768,0.0180896004041036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,768,0.004942933221658071
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,768,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,512,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,512,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,512,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,256,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,256,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,256,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,128,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,128,0.0032746667663256326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,64,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,128,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,64,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,64,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2560,32,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2560,32,0.002930133293072383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,65536,0.39325014750162757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2560,32,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,65536,0.23364267349243165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,65536,0.2752511978149414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,16384,0.04089173475901286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,16384,0.04392960071563721
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,16384,0.02184106707572937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,12288,0.03453760147094727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,12288,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,10240,0.03160746693611145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,12288,0.03583999872207642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,10240,0.030003199974695843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,10240,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,8192,0.028194133440653486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,8192,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,8192,0.012727466225624085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,7168,0.042803200085957845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,7168,0.022765866915384927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,6144,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,6144,0.03867306709289551
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,7168,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,6144,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,5120,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,5120,0.034918399651845296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,5120,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,4096,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,3584,0.02887679934501648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,4096,0.01242453356583913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,4096,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,3584,0.014195199807484946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,3584,0.0076789334416389465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,3072,0.026999467611312868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,3072,0.011400533715883891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,3072,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,2560,0.024951465924580894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,2560,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,2560,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,2048,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,2048,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,2048,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,1536,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,1536,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,1536,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,1024,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,1024,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,1024,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,768,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,768,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,768,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,512,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,512,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,512,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,256,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,256,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,256,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,128,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,128,0.0031744000812371576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,128,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,64,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,64,0.00310506671667099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,64,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,2048,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,2048,32,0.014847999811172486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,2048,32,0.004090666770935059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,65536,0.21971626281738282
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,65536,0.13499733606974285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,65536,0.28818772633870443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,16384,0.04082346757253011
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,16384,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,16384,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,12288,0.0341973344484965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,12288,0.018500266472498576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,12288,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,10240,0.031231999397277832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,10240,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,10240,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,8192,0.02805759906768799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,8192,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,8192,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,7168,0.04229120016098022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,7168,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,7168,0.01181013286113739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,6144,0.03857066631317139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,6144,0.01744106610616048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,6144,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,5120,0.03423573176066081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,5120,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,5120,0.009588266412417095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,4096,0.030719999472300214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,4096,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,4096,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,3584,0.02874026695887248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,3584,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,3584,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,3072,0.026658133665720625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,3072,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,3072,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,2560,0.024985599517822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,2560,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,2560,0.006650666892528534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,2048,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,2048,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,2048,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,1536,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,1536,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,1536,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,1024,0.01884160041809082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,1024,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,1024,0.004844800134499868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,768,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,768,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,768,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,512,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,512,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,512,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,256,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,256,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,128,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,256,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,64,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,128,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,128,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,64,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,64,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1536,32,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1536,32,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1536,32,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,65536,0.11666773160298664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,65536,0.19292160669962566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,65536,0.07741440137227376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,16384,0.040652799606323245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,16384,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,12288,0.034474666913350424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,16384,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,12288,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,12288,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,10240,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,10240,0.031128533681233722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,10240,0.015084800124168397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,8192,0.027955200274785357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,8192,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,8192,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,7168,0.04194986820220947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,7168,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,7168,0.01181013286113739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,6144,0.03850239912668864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,6144,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,6144,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,5120,0.034679468472798666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,5120,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,5120,0.00972266693909963
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,4096,0.03054933349291484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,4096,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,4096,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,3584,0.02877440055211385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,3584,0.006820266445477803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,3584,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,3072,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,3072,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,3072,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,2560,0.02481493353843689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,2560,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,2560,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,2048,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,2048,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,2048,0.005904000004132589
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,1536,0.02068480054537455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,1536,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,1536,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,1024,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,1024,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,1024,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,768,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,768,0.004231466849644979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,768,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,512,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,512,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,512,0.004328533510367076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,256,0.015733333428700765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,256,0.003239466746648153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,256,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,128,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,128,0.015118933717409768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,64,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,128,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,64,0.002690133452415466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,1024,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,1024,32,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,1024,32,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,65536,0.11649706363677978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,65536,0.07898453076680502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,65536,0.07731200059254964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,16384,0.04051200151443481
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,16384,0.013585066795349121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,16384,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,12288,0.03420159816741943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,12288,0.011025066177050274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,12288,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,10240,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,10240,0.030958932638168336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,10240,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,8192,0.027846399943033857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,8192,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,8192,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,7168,0.04215466578801473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,7168,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,7168,0.011707733074824016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,6144,0.03799039920171102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,6144,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,6144,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,5120,0.034577067693074545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,5120,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,5120,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,4096,0.03027626673380534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,4096,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,4096,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,3584,0.028563199440638225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,3584,0.006142933170000712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,3584,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,3072,0.027135999997456868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,3072,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,3072,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,2560,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,2560,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,2560,0.0067893331249554946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,2048,0.022528000672658286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,2048,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,2048,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,1536,0.02068480054537455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,1536,0.005898666878541311
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,1536,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,1024,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,1024,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,768,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,1024,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,768,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,768,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,512,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,512,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,512,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,256,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,256,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,256,0.004091733445723852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,128,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,128,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,128,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,64,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,768,32,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,64,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,768,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,768,32,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,65536,0.11663359800974529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,65536,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,65536,0.0771071990331014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,16384,0.040482131640116374
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,16384,0.021640533208847047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,16384,0.011332266529401143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,12288,0.03413333495457967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,12288,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,12288,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,10240,0.031129600604375203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,10240,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,8192,0.02805759906768799
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,10240,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,8192,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,8192,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,7168,0.04225706656773885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,7168,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,7168,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,6144,0.03819093306859334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,6144,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,6144,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,5120,0.03430399894714355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,5120,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,5120,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,4096,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,4096,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,4096,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,3584,0.02877440055211385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,3584,0.007572266459465027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,3584,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,3072,0.026555732885996504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,3072,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,3072,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,2560,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,2560,0.005218133330345154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,2560,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,2048,0.022459733486175536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,2048,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,2048,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,1536,0.02068480054537455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,1536,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,1536,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,1024,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,1024,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,1024,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,768,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,768,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,768,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,512,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,512,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,256,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,512,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,256,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,256,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,128,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,128,0.0025216000775496166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,128,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,64,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,64,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,64,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,512,32,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,512,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,512,32,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,65536,0.1162922700246175
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,65536,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,65536,0.07731200059254964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,16384,0.040686933199564616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,16384,0.008904533584912618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,16384,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,12288,0.0339626669883728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,12288,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,12288,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,10240,0.031027199824651082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,10240,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,10240,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,8192,0.02775040070215861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,8192,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,8192,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,7168,0.041674665609995526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,7168,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,7168,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,6144,0.03839999834696452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,6144,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,6144,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,5120,0.03430399894714355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,5120,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,5120,0.00941439966360728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,4096,0.03027626673380534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,4096,0.005153066913286845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,4096,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,3584,0.02874026695887248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,3584,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,3584,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,3072,0.02641493280728658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,3072,0.004743466774622599
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,3072,0.00709440012772878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,2560,0.02501973311106364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,2560,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,2560,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,2048,0.0225600004196167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,2048,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,2048,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,1536,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,1536,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,1536,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,1024,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,1024,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,1024,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,768,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,768,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,768,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,512,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,512,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,512,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,256,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,256,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,128,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,128,0.0024213333924611407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,128,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,64,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,64,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,256,32,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,256,32,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,256,32,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,65536,0.11622400283813476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,65536,0.010748799641927083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,65536,0.07720959981282552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,16384,0.04089173475901286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,16384,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,16384,0.021640533208847047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,12288,0.0340992013613383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,12288,0.0058005332946777345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,12288,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,10240,0.03092479904492696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,10240,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,10240,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,8192,0.02788693308830261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,8192,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,8192,0.012593066692352295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,7168,0.04205226500829061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,7168,0.005865600208441416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,6144,0.038365864753723146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,7168,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,6144,0.010580266515413921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,6144,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,5120,0.03433813254038493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,5120,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,5120,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,4096,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,4096,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,4096,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,3584,0.028637866179148357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,3584,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,3584,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,3072,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,3072,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,2560,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,3072,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,2560,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,2560,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,2048,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,2048,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,2048,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,1536,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,1536,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,1536,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,1024,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,1024,0.003956266740957896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,1024,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,768,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,768,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,768,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,512,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,512,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,512,0.0036170666416486105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,256,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,256,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,256,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,128,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,128,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,64,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,64,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,128,32,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,128,64,0.0024565334121386213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,128,32,0.004127999891837438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,65536,0.11663359800974529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,65536,0.07707306543986002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,65536,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,16384,0.04037973483403524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,16384,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,16384,0.021604265769322714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,12288,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,12288,0.03402560154596965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,12288,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,10240,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,10240,0.03092479904492696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,10240,0.014983466267585755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,8192,0.02788693308830261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,8192,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,8192,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,7168,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,6144,0.03822933435440064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,7168,0.04181119998296102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,7168,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,6144,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,6144,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,5120,0.03460799853006999
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,5120,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,5120,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,4096,0.03041279911994934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,4096,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,3584,0.028566400210062664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,3584,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,4096,0.004640000065167745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,3584,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,3072,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,3072,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,3072,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,2560,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,2560,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,2560,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,2048,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,2048,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,2048,0.022323199113210044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,1536,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,1536,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,1536,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,1024,0.018602667252222697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,1024,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,1024,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,768,0.017746132612228394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,768,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,512,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,768,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,512,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,512,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,256,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,256,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,128,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,128,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,64,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,64,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,64,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,64,32,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,64,32,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,64,32,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,65536,0.11636053721110026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,65536,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,16384,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,16384,0.040277334054311116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,16384,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,65536,0.07471786340077718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,12288,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,12288,0.034030934174855546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,12288,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,10240,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,10240,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,10240,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,8192,0.02815999984741211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,8192,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,8192,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,7168,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,7168,0.04205226500829061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,7168,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,6144,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,6144,0.038365864753723146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,6144,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,5120,0.03444053332010905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,5120,0.00928106705347697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,5120,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,4096,0.004192000130812327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,4096,0.03031040032704671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,3584,0.028603732585906982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,4096,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,3584,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,3584,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,3072,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,3072,0.0068245331446329755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,2560,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,2560,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,3072,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,2560,0.024541866779327393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,2048,0.022801067431767783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,2048,0.008703999718030294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,2048,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,1536,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,1536,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,1536,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,1024,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,1024,0.005320533116658529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,768,0.017850667238235474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,1024,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,768,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,768,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,512,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,512,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,256,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,128,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,256,0.003988266736268997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,128,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,64,0.014916266997655234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,64,0.0028949332733949023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,64,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,16,32,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,16,32,32,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,16,32,32,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,16384,1.5377066294352213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,16384,3.257275644938151
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,16384,1.773841094970703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,12288,1.1505663553873697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,12288,1.355464553833008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,12288,2.5732096354166667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,10240,1.1124053955078126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,10240,2.038545099894206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,10240,1.0564245223999023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,8192,0.8480085372924805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,8192,0.8891050974527994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,8192,1.5143253326416015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,7168,0.7376213073730469
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,7168,0.7886847813924154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,7168,1.4250666300455728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,6144,0.6329343795776368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,6144,0.6779904047648112
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,6144,1.1264959971110025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,5120,0.567364247639974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,5120,0.4913834571838379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,5120,1.0191189448038735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,4096,0.74738343556722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,4096,0.4545877456665039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,4096,0.3968682607014974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,3584,0.3457365353902181
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,3584,0.6555989583333333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,3072,0.34618028004964196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,3072,0.5623807907104492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,3072,0.31883945465087893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,3584,0.40256853103637696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,2560,0.290611203511556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,2560,0.24958292643229166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,2560,0.47018667856852214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,2048,0.22278827031453452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,2048,0.1988266626993815
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,2048,0.378163210550944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,1536,0.06331733465194703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,1536,0.30774081548055016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,1536,0.06017706791559855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,1024,0.0335530678431193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,1024,0.19196160634358722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,1024,0.025497599442799883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,768,0.029764266808827718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,768,0.031436800956726074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,768,0.02136639952659607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,512,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,512,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,512,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,256,0.021708800395329794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,256,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,128,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,128,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,256,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,64,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,128,0.011844266454378765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,64,0.011635200182596842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,32,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,65536,64,0.018768000602722167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,65536,32,0.012117333213488261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,65536,32,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,16384,0.4360191980997722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,16384,0.8106666564941406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,16384,0.4635648091634114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,65536,1.5206741333007812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,12288,0.6761098861694336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,65536,1.81572265625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,12288,0.3618816057840983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,65536,3.231914774576823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,12288,0.3293866793314616
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,10240,0.2954239845275879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,10240,0.48472747802734373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,10240,0.27132587432861327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,8192,0.223197873433431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,8192,0.21633599599202474
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,8192,0.42615467707316085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,7168,0.21804374059041343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,7168,0.3444735844930013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,7168,0.18961067199707032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,6144,0.05495466788609823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,6144,0.06048426628112793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,6144,0.28739840189615884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,5120,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,5120,0.03945706685384114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,5120,0.2357247988382975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,4096,0.033928533395131424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,4096,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,4096,0.18916692733764648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,3584,0.03170986572901408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,3584,0.16541013717651368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,3584,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,3072,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,2560,0.027374933163324993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,2560,0.02826026678085327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,3072,0.03635199864705403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,3072,0.029286400477091475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,2560,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,2048,0.024780799945195518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,2048,0.013687466581662497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,1536,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,1536,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,2048,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,1024,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,1536,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,1024,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,1024,0.020273067553838096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,768,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,768,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,768,0.00870293378829956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,512,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,512,0.007469866673151653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,256,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,512,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,256,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,256,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,128,0.01604159971078237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,128,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,64,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,64,0.015495466192563376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,128,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,64,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,16384,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,16384,32,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,16384,32,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,16384,0.35754667917887367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,65536,1.3749930063883462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,65536,1.1474549611409506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,65536,2.4239178975423177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,16384,0.32989867528279626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,12288,0.2787328084309896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,16384,0.6290431976318359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,12288,0.4986879984537761
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,12288,0.22773760159810386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,10240,0.2268160025278727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,10240,0.38942718505859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,8192,0.05481813351313273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,10240,0.2063360055287679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,8192,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,8192,0.3073024113972982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,7168,0.04471466541290283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,7168,0.24763733545939126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,7168,0.03256319959958394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,6144,0.03959466616312663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,6144,0.2179072062174479
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,5120,0.03543039957682292
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,6144,0.02699306607246399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,5120,0.17834666570027669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,4096,0.038126933574676516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,4096,0.030958932638168336
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,5120,0.02225493391354879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,4096,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,3584,0.029798400402069092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,3584,0.031914667288462324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,3584,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,3072,0.027373866240183516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,3072,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,3072,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,2560,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,2560,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,2048,0.023483733336130776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,2560,0.025702399015426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,2048,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,1536,0.021640533208847047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,2048,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,1536,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,1536,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,1024,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,1024,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,1024,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,768,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,768,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,512,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,512,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,768,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,256,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,512,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,256,0.004295466840267182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,256,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,128,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,128,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,128,0.005459199845790863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,64,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,64,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,64,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,12288,32,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,12288,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,12288,32,0.014233600099881491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,65536,0.97259521484375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,16384,0.3036842664082845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,65536,2.0180309295654295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,65536,1.1520000457763673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,16384,0.5005653381347657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,16384,0.2797226587931315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,12288,0.23377920786539713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,12288,0.4092565218607585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,12288,0.21473066012064615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,10240,0.1421994686126709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,10240,0.11663359800974529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,10240,0.3219114621480306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,8192,0.24169813791910805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,8192,0.034850132465362546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,8192,0.03136853377024333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,7168,0.029422932863235475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,7168,0.20643839836120606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,7168,0.040853333473205564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,6144,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,6144,0.1820672035217285
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,6144,0.025634133815765382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,5120,0.033655468622843424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,5120,0.08426986535390218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,5120,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,4096,0.029422932863235475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,4096,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,3584,0.027818665901819868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,4096,0.022765866915384927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,3584,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,3584,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,3072,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,3072,0.02600533366203308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,2560,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,3072,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,2048,0.022801067431767783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,2560,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,2560,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,2048,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,2048,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,1536,0.010786133011182149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,1536,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,1536,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,1024,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,1024,0.008532266815503438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,1024,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,768,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,768,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,768,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,512,0.005285333096981049
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,256,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,512,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,512,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,256,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,256,0.004060799876848856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,128,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,128,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,64,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,64,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,64,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,10240,32,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,10240,32,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,32,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,65536,0.8981162389119467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,65536,1.6116736094156903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,10240,128,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,65536,0.8979114532470703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,16384,0.39516159693400066
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,16384,0.24388267199198405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,16384,0.22667946815490722
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,12288,0.10659840106964111
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,12288,0.35280211766560876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,12288,0.04751360019048055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,10240,0.2632021268208822
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,10240,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,8192,0.19500373204549154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,8192,0.032255999247233075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,10240,0.037922132015228274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,8192,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,7168,0.03894613186518352
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,7168,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,7168,0.16482987403869628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,6144,0.03669333457946777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,6144,0.053657599290212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,6144,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,5120,0.027883734305699664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,4096,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,4096,0.028262400627136232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,5120,0.013989333311716715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,4096,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,3584,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,5120,0.031914667288462324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,3584,0.020241065820058187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,3072,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,3072,0.024985599517822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,3584,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,2560,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,3072,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,2560,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,2048,0.02211839954058329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,2048,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,2048,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,1536,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,2560,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,1536,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,1024,0.018634666999181114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,1536,0.007264000177383423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,1024,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,1024,0.006175999840100607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,768,0.017812265952428182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,768,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,512,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,512,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,512,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,256,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,256,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,768,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,256,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,128,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,128,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,64,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,64,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,128,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,64,0.015150933464368185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,8192,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,8192,32,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,8192,32,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,65536,0.7910741170247395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,16384,0.34143571853637694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,16384,0.2022709369659424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,65536,1.408238983154297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,12288,0.04126720031102498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,16384,0.21445867220560708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,65536,0.6988117218017578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,12288,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,10240,0.03549866676330567
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,10240,0.20551679929097494
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,10240,0.022937599817911783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,8192,0.03068586587905884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,12288,0.2789375940958659
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,8192,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,8192,0.19244373639424642
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,7168,0.0373418649037679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,7168,0.05058559974034628
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,7168,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,6144,0.03426986535390218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,6144,0.03041279911994934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,5120,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,5120,0.022562134265899658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,5120,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,4096,0.027101866404215497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,4096,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,6144,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,4096,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,3584,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,3584,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,3072,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,3072,0.016178133090337117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,2560,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,3584,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,2560,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,2560,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,3072,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,2048,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,2048,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,1536,0.02023893396059672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,1536,0.009177600344022114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,1024,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,2048,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,1536,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,1024,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,768,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,768,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,512,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,768,0.005491200089454651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,512,0.004640000065167745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,1024,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,512,0.005221333106358846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,256,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,256,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,256,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,128,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,128,0.004329599936803182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,128,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,64,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,64,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,32,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,7168,32,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,7168,32,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,7168,64,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,16384,0.03324586749076843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,16384,0.290065066019694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,16384,0.07884799639383952
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,12288,0.037034666538238524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,65536,0.6712661107381185
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,65536,0.6823935826619466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,65536,1.223953119913737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,12288,0.022801067431767783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,12288,0.23831893603006998
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,10240,0.18763093948364257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,10240,0.01952000061670939
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,10240,0.03399680058161418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,8192,0.05287253459294637
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,8192,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,7168,0.03566933472951253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,8192,0.029013333717981975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,7168,0.015868799885114034
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,6144,0.032767999172210696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,6144,0.027613866329193115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,6144,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,7168,0.03334826628367106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,4096,0.026245333751042682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,5120,0.012117333213488261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,4096,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,4096,0.009898666540781658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,5120,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,3584,0.02505386670430501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,5120,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,3584,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,3072,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,3072,0.02392746607462565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,3072,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,3584,0.017573332786560057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,2560,0.025088000297546386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,2560,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,2048,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,2048,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,1536,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,2048,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,1536,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,2560,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,1536,0.00890773336092631
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,1024,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,1024,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,768,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,768,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,1024,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,512,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,768,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,256,0.015901866555213928
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,256,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,256,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,512,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,128,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,512,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,128,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,64,0.003035733352104823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,64,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,64,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,128,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,6144,32,0.0030005333324273427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,6144,32,0.014262400070826211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,65536,0.5713919957478841
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,16384,0.040345601240793866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,65536,1.0029397328694662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,65536,0.5600255966186524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,6144,32,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,16384,0.24528212547302247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,16384,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,12288,0.19742719332377118
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,12288,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,10240,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,10240,0.09209173520406087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,12288,0.03444053332010905
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,8192,0.028535467386245728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,8192,0.02836479942003886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,8192,0.01409066617488861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,7168,0.03406293392181396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,6144,0.032051199674606325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,10240,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,7168,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,7168,0.014131200313568116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,6144,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,5120,0.028806400299072266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,6144,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,5120,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,5120,0.010682666301727295
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,4096,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,3584,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,3072,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,4096,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,3584,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,3072,0.01395626664161682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,4096,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,3072,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,3584,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,2560,0.021977599461873373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,2560,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,2048,0.00689279983441035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,1536,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,2048,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,1536,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,2048,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,2560,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,1536,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,1024,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,1024,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,1024,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,768,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,768,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,512,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,512,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,768,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,256,0.015836800138155617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,256,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,256,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,512,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,128,0.0032074667513370516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,128,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,128,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,64,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,64,0.015222400426864624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,5120,32,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,5120,32,0.0040949332217375435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,65536,0.4656810760498047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,16384,0.04037973483403524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,65536,0.8004255930582682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,5120,32,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,65536,0.44837547938028977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,12288,0.03399680058161418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,16384,0.02197973330815633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,16384,0.19708587328592936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,12288,0.05781759818394979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,12288,0.01884053349494934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,10240,0.03089066743850708
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,10240,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,8192,0.027818665901819868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,8192,0.025459200143814087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,7168,0.033655468622843424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,10240,0.03239253362019857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,7168,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,7168,0.027716267108917236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,8192,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,6144,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,6144,0.03136853377024333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,6144,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,5120,0.028501333793004353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,5120,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,4096,0.025668267409006757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,4096,0.008292266726493835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,4096,0.013892267147699991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,5120,0.010342400272687275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,3584,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,3584,0.024780799945195518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,3072,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,3072,0.012558933099110922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,3584,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,3072,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,2560,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,2048,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,2048,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,2048,0.020753065745035805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,2560,0.021875200668970744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,1536,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,1536,0.007812266548474629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,2560,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,1536,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,1024,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,1024,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,768,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,1024,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,768,0.005115733544031779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,768,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,512,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,512,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,256,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,256,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,128,0.003138133386770884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,512,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,256,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,128,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,128,0.004057600100835165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,64,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,4096,32,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,4096,32,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,65536,0.4118517239888509
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,65536,0.39491732915242517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,16384,0.040140799681345624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,65536,0.707583999633789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,16384,0.17001813252766926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,4096,32,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,16384,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,12288,0.03420159816741943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,12288,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,12288,0.037375998497009275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,10240,0.031402667363484696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,10240,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,10240,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,8192,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,7168,0.03389439980189006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,7168,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,7168,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,8192,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,8192,0.027818665901819868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,6144,0.010717866818110149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,6144,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,6144,0.030716800689697267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,5120,0.028501333793004353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,5120,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,5120,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,4096,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,3584,0.024268800020217897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,4096,0.025941334168116253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,3584,0.013926399747530618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,3072,0.023005867004394533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,3584,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,3072,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,4096,0.008428800106048583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,3072,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,2560,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,2560,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,2560,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,2048,0.020371200640996297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,2048,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,2048,0.006039466460545858
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,1536,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,1536,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,1024,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,1024,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,768,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,1024,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,1536,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,768,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,768,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,512,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,256,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,512,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,256,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,128,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,512,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,128,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,128,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,256,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,64,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,64,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3584,32,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,64,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3584,32,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3584,32,0.004124800115823746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,65536,0.3591509183247884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,16384,0.039765334129333495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,65536,0.33628158569335936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,16384,0.0711680014928182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,12288,0.036556800206502274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,16384,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,12288,0.03368959824244182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,12288,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,65536,0.6078122456868489
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,10240,0.030481066306432086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,10240,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,10240,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,8192,0.02754559914271037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,7168,0.03351893424987793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,8192,0.012558933099110922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,8192,0.023995733261108397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,7168,0.01181013286113739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,6144,0.030853333075841267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,6144,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,6144,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,5120,0.028398933013280232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,5120,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,5120,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,7168,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,4096,0.025565866629282636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,4096,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,4096,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,3584,0.024234666426976522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,3584,0.013375999530156455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,3584,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,3072,0.023654399315516154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,2560,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,3072,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,2560,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,2560,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,2048,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,3072,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,1536,0.019010132551193236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,1536,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,2048,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,2048,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,1536,0.008663466572761536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,1024,0.01771413286526998
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,1024,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,768,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,768,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,512,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,512,0.004602666695912679
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,768,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,512,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,1024,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,256,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,256,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,256,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,128,0.015083733201026916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,128,0.0032757334411144257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,128,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,64,0.003852800031503042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,64,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,3072,32,0.014129066467285156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,32,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,3072,32,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,3072,64,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,65536,0.3084287961324056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,65536,0.4964351971944173
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,65536,0.2878080050150553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,16384,0.0400383989016215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,16384,0.04471466541290283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,16384,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,12288,0.03368959824244182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,12288,0.01699733336766561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,10240,0.030719999472300214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,10240,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,8192,0.02754559914271037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,12288,0.035293865203857425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,10240,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,8192,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,8192,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,7168,0.03338239987691243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,7168,0.024098134040832518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,6144,0.03078826665878296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,5120,0.02836479942003886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,5120,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,7168,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,6144,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,5120,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,4096,0.02542933424313863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,6144,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,4096,0.01306773324807485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,4096,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,3584,0.024267733097076416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,3584,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,3072,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,3072,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,3584,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,2560,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,2560,0.009898666540781658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,2560,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,2048,0.007946666578451793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,2048,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,2048,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,3072,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,1536,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,1536,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,1024,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,1536,0.01884160041809082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,1024,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,1024,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,768,0.017474132776260375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,768,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,512,0.016309332847595216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,512,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,512,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,256,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,768,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,256,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,128,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,256,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,128,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,64,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,32,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2560,64,0.014980266491572062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2560,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,32,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,65536,0.2574336051940918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2560,64,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,65536,0.3954005241394043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,65536,0.23125119209289552
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,16384,0.039389868577321366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,16384,0.021538132429122926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,16384,0.0400383989016215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,12288,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,12288,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,10240,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,10240,0.02693120042483012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,10240,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,8192,0.02744213342666626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,12288,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,8192,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,7168,0.02218666672706604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,7168,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,8192,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,7168,0.03362133502960205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,6144,0.03054933349291484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,6144,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,5120,0.028091732660929365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,5120,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,5120,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,4096,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,6144,0.010513066252072652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,4096,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,3584,0.024129066864649454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,4096,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,3584,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,3584,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,3072,0.023040000597635904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,3072,0.010956799983978272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,2560,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,2560,0.02198186715443929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,2048,0.020172800620396933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,2560,0.006481066842873891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,2048,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,2048,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,3072,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,1536,0.019011199474334717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,1536,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,1024,0.017473065853118898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,1536,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,1024,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,768,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,1024,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,768,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,512,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,768,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,512,0.0037162666519482933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,512,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,256,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,256,0.0035135999321937563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,128,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,128,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,256,0.004091733445723852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,64,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,64,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,2048,32,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,64,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,2048,32,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,32,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,2048,128,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,65536,0.16424959500630695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,65536,0.292631467183431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,16384,0.03942399819691976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,65536,0.11741866270701092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,16384,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,16384,0.020480000972747804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,12288,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,10240,0.03064853350321452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,10240,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,12288,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,12288,0.017100799083709716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,8192,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,10240,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,8192,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,8192,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,7168,0.03351893424987793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,7168,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,6144,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,5120,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,6144,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,7168,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,5120,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,5120,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,6144,0.03061760067939758
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,4096,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,4096,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,4096,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,3584,0.023995733261108397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,3584,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,3584,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,3072,0.02300373315811157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,3072,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,2560,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,2560,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,3072,0.010035199920336406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,2560,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,2048,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,2048,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,1536,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,2048,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,1536,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,1536,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,1024,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,1024,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,768,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,768,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,1024,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,512,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,512,0.0037503999968369803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,768,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,512,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,256,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,256,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,128,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,256,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,128,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,64,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,64,0.014949333667755128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,128,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,64,0.003990400085846583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1536,32,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1536,32,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,65536,0.11117226282755535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1536,32,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,65536,0.07426986694335938
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,65536,0.19135146141052245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,16384,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,12288,0.033243733644485476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,16384,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,12288,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,16384,0.03925333420435588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,12288,0.010990933577219645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,10240,0.030481066306432086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,10240,0.010035199920336406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,8192,0.027135999997456868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,8192,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,10240,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,8192,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,7168,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,7168,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,6144,0.030616533756256104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,6144,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,7168,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,5120,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,6144,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,5120,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,5120,0.02792106668154399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,4096,0.02542933424313863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,4096,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,4096,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,3584,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,3584,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,3072,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,3584,0.024098134040832518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,3072,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,3072,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,2560,0.021396267414093017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,2560,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,2560,0.006481066842873891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,2048,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,2048,0.020002132654190062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,1536,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,1024,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,1536,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,1536,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,2048,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,1024,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,1024,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,768,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,768,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,768,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,512,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,256,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,256,0.003032533327738444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,512,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,256,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,128,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,128,0.00262719988822937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,512,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,64,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,64,0.0025557334224383037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,64,0.0038506666819254553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,1024,32,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,1024,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,1024,32,0.013926399747530618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,65536,0.07475199699401855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,65536,0.06389760176340739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,16384,0.03935573498408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,65536,0.1112063964207967
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,16384,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,12288,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,16384,0.011566932996114094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,12288,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,12288,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,10240,0.030173865954081218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,10240,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,8192,0.02723840077718099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,8192,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,10240,0.014711466431617738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,7168,0.03307200074195862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,8192,0.012219732999801636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,7168,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,7168,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,6144,0.030718932549158733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,6144,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,6144,0.010410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,4096,0.025357866287231447
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,4096,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,5120,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,5120,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,3584,0.02395840088526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,5120,0.027955200274785357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,4096,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,3584,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,3584,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,3072,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,3072,0.005356800059477488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,2560,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,3072,0.007202133536338806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,2048,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,2560,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,2048,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,2048,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,2560,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,1536,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,1536,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,1024,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,1536,0.005321600039800008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,1024,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,1024,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,768,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,512,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,768,0.004535466432571411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,512,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,768,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,256,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,256,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,512,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,128,0.015117866794268289
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,128,0.0025216000775496166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,128,0.003992533435424169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,64,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,32,0.013892267147699991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,768,64,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,768,32,0.002455466737349828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,768,32,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,65536,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,65536,0.11113813718159993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,65536,0.07505813439687094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,16384,0.039077333609263104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,12288,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,16384,0.009898666540781658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,12288,0.03307519952456157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,12288,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,10240,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,16384,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,10240,0.030339199304580688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,10240,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,8192,0.027170133590698243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,8192,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,7168,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,7168,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,8192,0.012151466806729634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,6144,0.03031040032704671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,6144,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,6144,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,7168,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,5120,0.028091732660929365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,5120,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,5120,0.009211732943852743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,4096,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,4096,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,4096,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,3584,0.023990400632222495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,3072,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,3584,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,3072,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,3584,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,3072,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,2560,0.021329067150751748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,2560,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,2560,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,2048,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,2048,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,2048,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,1536,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,1024,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,1536,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,1536,0.01884160041809082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,1024,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,768,0.01679253379503886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,1024,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,768,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,768,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,512,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,512,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,256,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,256,0.002863999952872594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,256,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,512,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,128,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,128,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,64,0.015289599696795145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,65536,0.11086506843566894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,65536,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,512,32,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,512,32,0.013923199971516928
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,512,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,16384,0.0392192006111145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,16384,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,65536,0.0740010658899943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,16384,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,12288,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,12288,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,12288,0.016619732975959776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,10240,0.006073600053787232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,10240,0.014709333578745524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,8192,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,8192,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,8192,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,10240,0.030242133140563964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,7168,0.03321173389752706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,7168,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,6144,0.03054933349291484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,7168,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,6144,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,6144,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,5120,0.02792106668154399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,4096,0.02529279987017314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,5120,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,5120,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,4096,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,4096,0.007912533481915791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,3584,0.004875733455022176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,3584,0.023995733261108397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,3584,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,3072,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,3072,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,2560,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,2560,0.004567466676235199
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,2048,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,2560,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,3072,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,2048,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,2048,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,1536,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,1536,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,1536,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,1024,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,1024,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,768,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,768,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,1024,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,768,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,512,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,512,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,512,0.004368000229199727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,256,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,256,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,128,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,256,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,64,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,128,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,128,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,256,32,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,64,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,256,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,65536,0.11100160280863444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,256,32,0.0038549333810806276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,65536,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,65536,0.07406933307647705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,16384,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,16384,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,16384,0.03911679983139038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,12288,0.03317760030428569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,12288,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,10240,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,10240,0.03037866751352946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,12288,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,10240,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,8192,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,8192,0.027272532383600872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,7168,0.033070933818817136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,7168,0.005118933320045471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,8192,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,7168,0.011332266529401143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,6144,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,6144,0.03007146716117859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,6144,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,5120,0.027852799495061236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,5120,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,5120,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,4096,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,4096,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,3584,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,4096,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,3584,0.02368853290875753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,3072,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,3584,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,3072,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,3072,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,2560,0.021538132429122926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,2560,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,2048,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,2048,0.005835733314355215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,2048,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,1536,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,2560,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,1536,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,1536,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,1024,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,1024,0.005082666873931885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,1024,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,768,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,512,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,768,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,768,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,512,0.003684266656637192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,512,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,256,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,256,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,128,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,256,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,128,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,64,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,128,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,128,32,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,128,32,0.013926399747530618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,128,32,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,65536,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,65536,0.07393279870351156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,65536,0.1107968012491862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,16384,0.039287467797597245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,16384,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,12288,0.03310933311780294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,12288,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,16384,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,12288,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,10240,0.03020799954732259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,10240,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,10240,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,8192,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,8192,0.0049792001644770306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,8192,0.012151466806729634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,7168,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,7168,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,6144,0.010069333513577779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,7168,0.03300693432490031
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,5120,0.027850667635599773
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,6144,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,6144,0.030003199974695843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,5120,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,4096,0.025156267484029132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,5120,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,4096,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,3584,0.023893332481384276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,3584,0.01269760032494863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,3584,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,4096,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,3072,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,3072,0.011191466450691223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,3072,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,2560,0.021499733130137123
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,2560,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,2560,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,2048,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,1536,0.01869973341623942
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,2048,0.019760000705718993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,1536,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,1536,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,2048,0.008224000036716462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,1024,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,1024,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,1024,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,768,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,512,0.016178133090337117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,768,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,768,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,512,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,512,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,256,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,256,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,256,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,128,0.015153066317240397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,128,0.004125866790612539
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,64,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,128,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,64,32,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,64,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,64,32,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,64,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,65536,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,65536,0.11185493469238281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,65536,0.07471786340077718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,16384,0.03911679983139038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,16384,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,16384,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,12288,0.03321173389752706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,12288,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,10240,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,12288,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,10240,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,10240,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,8192,0.027204267183939618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,8192,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,7168,0.03293866713841756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,7168,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,8192,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,6144,0.03020799954732259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,6144,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,6144,0.01972800095876058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,5120,0.02802346746126811
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,7168,0.011365333199501037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,5120,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,5120,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,4096,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,4096,0.025224532683690386
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,4096,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,3584,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,3584,0.023654399315516154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,3072,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,3072,0.011534933249155681
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,3584,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,3072,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,2560,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,2560,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,2560,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,2048,0.007981866598129272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,2048,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,2048,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,1536,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,1536,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,1536,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,1024,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,1024,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,768,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,1024,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,768,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,768,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,512,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,256,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,256,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,512,0.004227200150489807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,256,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,512,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,128,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,128,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,128,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,64,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,32,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,8,32,32,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,8,32,64,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,8,32,32,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,16384,1.7684139251708983
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,16384,1.6713045756022136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,16384,3.2295913696289062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,12288,1.3486058553059896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,12288,1.1429205576578776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,12288,2.594713592529297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,10240,1.1088554382324218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,10240,2.0198741912841798
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,10240,1.046562131245931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,8192,0.8414549509684244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,8192,1.6255317687988282
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,8192,0.8864426930745444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,7168,0.7333205540974934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,7168,1.4166698455810547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,7168,0.7824725468953451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,6144,1.1232597351074218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,6144,0.5773994445800781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,5120,0.562278429667155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,5120,0.5258240063985189
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,6144,0.6719488143920899
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,5120,0.9275391896565756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,4096,0.4501503944396973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,4096,0.8078336079915365
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,4096,0.3920554796854655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,3584,0.39867734909057617
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,3584,0.6510592142740885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,3584,0.37000532150268556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,3072,0.5592405319213867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,3072,0.3193514823913574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,3072,0.3162453333536784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,2560,0.24582826296488441
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,2560,0.28777812321980795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,2560,0.46677331924438475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,2048,0.23367679913838707
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,2048,0.37495466868082683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,2048,0.21186559995015464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,1536,0.06932480335235595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,1536,0.2822826703389486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,1024,0.19002025922139484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,1536,0.048981332778930665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,1024,0.03269973397254944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,1024,0.024985599517822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,768,0.0291157325108846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,768,0.026589866479237872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,768,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,512,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,512,0.026897066831588747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,256,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,256,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,256,0.02115839918454488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,512,0.01685973405838013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,128,0.01986560026804606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,128,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,128,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,64,0.005354666709899902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,64,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,64,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,65536,32,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,65536,32,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,65536,32,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,16384,0.8175957361857096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,16384,0.4621312141418457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,65536,1.8045951843261718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,65536,3.2349525451660157
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,16384,0.4332544008890788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,12288,0.6692864100138347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,12288,0.2945023854573568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,12288,0.36000426610310876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,10240,0.2705749193827311
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,10240,0.5114879926045736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,10240,0.2938538551330566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,8192,0.23695359230041504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,65536,1.51637331644694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,8192,0.42298027674357097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,8192,0.19568640391031902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,7168,0.3268266677856445
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,7168,0.17025386492411296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,6144,0.04379306634267171
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,6144,0.03740799824396769
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,7168,0.21248000462849936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,5120,0.03546453317006429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,5120,0.024064000447591147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,4096,0.031231999397277832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,6144,0.3132416089375814
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,4096,0.2058922608693441
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,5120,0.23418879508972168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,4096,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,3584,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,3072,0.027818665901819868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,3584,0.029832533995310467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,3072,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,3072,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,2560,0.015051733454068503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,2560,0.025900799036026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,2048,0.02317439913749695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,2560,0.025873066981633504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,3584,0.16892587343851725
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,2048,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,1536,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,1536,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,1024,0.01945599913597107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,1536,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,2048,0.01269760032494863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,1024,0.011912533640861511
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,768,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,512,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,768,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,1024,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,512,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,768,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,512,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,256,0.0047423998514811196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,256,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,128,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,256,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,128,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,64,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,64,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,32,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,16384,64,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,128,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,16384,32,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,16384,32,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,16384,0.3560106595357259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,65536,1.1466709136962892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,65536,2.4233983357747393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,65536,1.3666987101236978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,12288,0.2775381406148275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,16384,0.6245717366536458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,16384,0.29484373728434243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,12288,0.2260650634765625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,10240,0.2268160025278727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,10240,0.38543361028035483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,12288,0.49670826594034834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,10240,0.20558506647745767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,8192,0.07847253481547037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,7168,0.03860479990641276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,7168,0.2623487949371338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,8192,0.312661329905192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,7168,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,8192,0.034508800506591795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,6144,0.232584540049235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,6144,0.026862933238347368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,5120,0.031880533695220946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,5120,0.18725546201070148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,6144,0.03519146839777629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,4096,0.02836479942003886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,5120,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,4096,0.01764586567878723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,3584,0.028842665751775104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,3584,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,3584,0.026589866479237872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,3072,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,4096,0.031948800881703696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,3072,0.025873066981633504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,2560,0.022528000672658286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,2560,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,2048,0.017779199282328288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,2560,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,3072,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,2048,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,1536,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,1536,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,2048,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,1024,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,1536,0.020411733786265054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,1024,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,768,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,1024,0.008533333738644917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,768,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,512,0.005592533449331919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,768,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,256,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,256,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,512,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,512,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,128,0.015734400351842245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,128,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,128,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,256,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,64,0.014131200313568116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,64,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,12288,32,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,32,0.00508480022350947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,12288,64,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,12288,32,0.013346133629480996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,65536,1.1412479400634765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,16384,0.2523135979970296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,16384,0.2982229232788086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,16384,0.49991680781046555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,65536,0.9683285395304362
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,65536,2.0529525756835936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,12288,0.23391572634379068
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,12288,0.21483519872029624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,10240,0.2919765472412109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,10240,0.13547520637512206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,10240,0.12038826942443848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,8192,0.034167468547821045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,12288,0.4044458707173665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,7168,0.03519146839777629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,7168,0.027306665976842243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,7168,0.20520960489908852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,8192,0.26306559244791666
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,8192,0.02986666758855184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,6144,0.03266560037930806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,6144,0.024268800020217897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,5120,0.02959360082944234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,5120,0.09755307038625081
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,5120,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,6144,0.18240853945414226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,4096,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,4096,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,3584,0.025565866629282636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,4096,0.022357332706451415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,3584,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,3072,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,3584,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,3072,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,3072,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,2048,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,2560,0.01634773313999176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,2048,0.01269760032494863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,2048,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,2560,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,2560,0.013346133629480996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,1536,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,1536,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,1024,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,1024,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,1024,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,768,0.017987199624379478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,1536,0.010410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,512,0.016822399695714314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,768,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,512,0.005153066913286845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,512,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,768,0.007608533402283986
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,256,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,128,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,128,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,64,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,64,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,256,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,256,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,128,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,64,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,10240,32,0.013346133629480996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,10240,32,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,10240,32,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,16384,0.3935914675394694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,65536,0.8873589197794596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,12288,0.08277333577473958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,16384,0.22644054094950355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,16384,0.2389674663543701
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,65536,0.8962730407714844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,12288,0.3238911946614583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,65536,1.6111615498860676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,10240,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,10240,0.23398399353027344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,12288,0.040209066867828366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,10240,0.03580586512883504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,8192,0.018602667252222697
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,8192,0.031061333417892457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,8192,0.21637120246887206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,7168,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,7168,0.17097387313842774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,6144,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,7168,0.03283626635869344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,6144,0.030242133140563964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,6144,0.028501333793004353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,5120,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,4096,0.02505386670430501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,5120,0.028535467386245728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,5120,0.024780799945195518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,4096,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,3584,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,3584,0.017951999107996622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,4096,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,3584,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,3072,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,2560,0.02187946637471517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,3072,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,2560,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,2560,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,3072,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,2048,0.010990933577219645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,2048,0.007642666498819987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,1536,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,2048,0.020309333006540933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,1536,0.006857599814732869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,1024,0.017885865767796834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,1024,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,768,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,1536,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,1024,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,768,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,512,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,256,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,512,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,512,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,768,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,128,0.015524267156918844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,256,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,256,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,128,0.004126933217048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,128,0.00347626656293869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,64,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,64,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,8192,32,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,64,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,8192,32,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,8192,32,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,65536,0.7784789403279622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,65536,1.4069482167561849
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,16384,0.34041067759195964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,65536,0.7798101425170898
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,16384,0.21179733276367188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,12288,0.024098134040832518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,10240,0.03402986526489258
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,8192,0.02955840031305949
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,12288,0.038809601465861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,16384,0.1950719992319743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,12288,0.2783232053120931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,10240,0.20497066179911294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,10240,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,8192,0.170359468460083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,8192,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,7168,0.031641600529352824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,6144,0.025702399015426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,6144,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,7168,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,6144,0.029320534070332843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,7168,0.0497322678565979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,5120,0.02723840077718099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,5120,0.021605332692464195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,5120,0.013205333550771078
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,4096,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,4096,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,3584,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,3072,0.022835199038187662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,4096,0.010683733224868774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,3072,0.01480959951877594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,3584,0.024132267634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,3584,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,2560,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,3072,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,2560,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,2560,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,2048,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,2048,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,2048,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,1024,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,1024,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,1536,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,1536,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,1536,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,1024,0.005898666878541311
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,768,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,512,0.01692906618118286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,768,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,512,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,512,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,768,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,256,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,256,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,256,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,128,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,128,0.004090666770935059
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,64,0.004092800120512644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,64,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,128,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,7168,32,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,7168,32,0.003958400090535482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,7168,32,0.013346133629480996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,65536,0.6738261540730794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,65536,1.203404744466146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,65536,0.6722901026407878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,16384,0.04682986736297608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,12288,0.03607893387476603
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,16384,0.05922133525212606
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,16384,0.29341014226277673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,12288,0.02218666672706604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,10240,0.1875285307566325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,8192,0.028125866254170732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,10240,0.031470932563145954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,10240,0.018667733669281004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,12288,0.23579306602478028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,8192,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,8192,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,7168,0.030105600754419964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,7168,0.02867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,7168,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,6144,0.02795413335164388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,6144,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,5120,0.012117333213488261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,6144,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,4096,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,4096,0.015871999661127727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,5120,0.02590720057487488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,5120,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,4096,0.009796266754468281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,3584,0.02259626587231954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,3584,0.015016532937685647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,3072,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,3072,0.021845332781473794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,3072,0.008669867118199667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,3584,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,2560,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,2560,0.011878400047620138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,2560,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,2048,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,2048,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,2048,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,1536,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,1024,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,1024,0.0056650668382644655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,768,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,768,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,1536,0.018600533405939736
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,1024,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,1536,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,512,0.004570666452248891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,768,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,512,0.0047765334447224935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,512,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,256,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,256,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,256,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,128,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,128,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,64,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,64,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,128,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,64,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,6144,32,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,6144,32,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,16384,0.03935573498408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,65536,1.0018474578857421
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,65536,0.5636095682779948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,65536,0.5588309605916341
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,16384,0.24620374043782553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,6144,32,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,12288,0.19694719314575196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,12288,0.019933867454528808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,10240,0.030856533845265703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,16384,0.02362026572227478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,12288,0.034030934174855546
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,10240,0.06256426572799682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,10240,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,8192,0.027716267108917236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,8192,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,7168,0.029422932863235475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,8192,0.02635093331336975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,7168,0.027852799495061236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,6144,0.027170133590698243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,6144,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,7168,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,5120,0.02542933424313863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,5120,0.019285333156585694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,5120,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,6144,0.011844266454378765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,4096,0.014876799782117209
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,4096,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,4096,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,3584,0.02263039946556091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,3584,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,3584,0.014131200313568116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,3072,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,3072,0.007912533481915791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,2560,0.020309333006540933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,2048,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,3072,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,2048,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,2560,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,2560,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,2048,0.006484266618887584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,1536,0.018193066120147705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,1536,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,1536,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,1024,0.006004266440868378
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,1024,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,768,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,1024,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,768,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,512,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,768,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,512,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,512,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,128,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,128,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,128,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,64,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,256,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,256,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,256,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,64,0.013957333564758301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,64,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,5120,32,0.004126933217048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,5120,32,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,5120,32,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,65536,0.4457130750020345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,16384,0.03952533404032389
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,65536,0.4562944094340007
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,16384,0.19514026641845703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,16384,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,65536,0.7995733261108399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,12288,0.046250665187835695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,12288,0.03345066706339518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,12288,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,10240,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,8192,0.027409066756566364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,10240,0.03068586587905884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,10240,0.030344533920288085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,8192,0.01269760032494863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,7168,0.028943999608357744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,8192,0.024029866854349772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,7168,0.024473599592844644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,7168,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,6144,0.02696533401807149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,6144,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,5120,0.02488320072491964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,5120,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,6144,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,4096,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,4096,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,4096,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,5120,0.009761066238085429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,3584,0.01262933313846588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,3584,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,3584,0.022357332706451415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,3072,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,3072,0.01160533328851064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,2560,0.020138667027155558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,3072,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,2560,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,2560,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,2048,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,2048,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,1536,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,1536,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,2048,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,1536,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,1024,0.00508480022350947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,1024,0.009315199653307597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,768,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,768,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,768,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,1024,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,512,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,512,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,512,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,256,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,256,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,256,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,128,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,64,0.014367999633153281
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,128,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,4096,32,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,64,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,64,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,4096,32,0.0027978666126728057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,4096,32,0.0038890667259693147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,65536,0.4037290573120117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,16384,0.03898026545842488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,16384,0.16957440376281738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,12288,0.03351893424987793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,65536,0.7077887852986653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,16384,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,65536,0.3950592041015625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,12288,0.03536213239034017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,12288,0.01689493258794149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,10240,0.03020799954732259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,10240,0.029422932863235475
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,10240,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,7168,0.028842665751775104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,8192,0.02768213351567586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,8192,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,7168,0.023108265797297158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,7168,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,8192,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,6144,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,6144,0.026822400093078614
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,6144,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,5120,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,4096,0.023001599311828613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,5120,0.02529279987017314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,5120,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,4096,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,3584,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,4096,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,3584,0.007642666498819987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,3584,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,3072,0.021057067314783733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,2560,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,3072,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,3072,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,2560,0.010069333513577779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,2048,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,2048,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,2048,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,2560,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,1536,0.018153599898020425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,1024,0.017062399784723917
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,1024,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,1536,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,1536,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,768,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,1024,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,768,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,768,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,512,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,512,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,256,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,512,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,256,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,128,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,256,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,128,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,64,0.013892267147699991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,64,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3584,32,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3584,32,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3584,32,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,65536,0.35150505701700846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,65536,0.33989973068237306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,65536,0.5967530568440755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,16384,0.03911679983139038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,16384,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,16384,0.06017706791559855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,12288,0.0335530678431193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,12288,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,12288,0.034406399726867674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,10240,0.03020799954732259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,10240,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,10240,0.028940800825754804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,8192,0.027170133590698243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,7168,0.028842665751775104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,7168,0.011571199695269267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,8192,0.012663466731707254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,6144,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,8192,0.023654399315516154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,7168,0.02211839954058329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,6144,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,6144,0.010513066252072652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,5120,0.02505386670430501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,5120,0.016450132926305136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,4096,0.02286293307940165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,5120,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,3584,0.021912533044815063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,4096,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,4096,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,3584,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,3584,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,3072,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,3072,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,2560,0.01992853283882141
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,3072,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,2048,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,2048,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,2560,0.010069333513577779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,2048,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,1536,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,2560,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,1536,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,1024,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,1024,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,1536,0.005596800148487091
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,768,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,768,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,1024,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,512,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,512,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,768,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,256,0.0035807999471823373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,512,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,256,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,256,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,128,0.002997333308060964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,128,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,128,0.003921066721280416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,64,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,32,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,64,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,3072,64,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,3072,32,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,3072,32,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,65536,0.49585491816202804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,65536,0.2877098719278971
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,16384,0.03898026545842488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,65536,0.30006612141927086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,16384,0.042700799306233723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,12288,0.03324586749076843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,16384,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,12288,0.03399680058161418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,10240,0.030173865954081218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,10240,0.02867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,12288,0.016858667135238647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,8192,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,8192,0.022937599817911783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,10240,0.014916266997655234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,7168,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,7168,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,8192,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,6144,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,7168,0.029047467311223346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,6144,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,6144,0.010342400272687275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,5120,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,5120,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,5120,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,4096,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,3584,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,4096,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,3584,0.021910399198532104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,4096,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,3072,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,3072,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,3584,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,3072,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,2560,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,2560,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,2560,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,2048,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,2048,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,2048,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,1536,0.005694933235645294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,1536,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,1536,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,1024,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,1024,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,768,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,1024,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,768,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,512,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,768,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,512,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,512,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,256,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,256,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,128,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,256,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,128,0.0029343999922275543
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,64,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,32,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2560,64,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,128,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2560,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2560,32,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,65536,0.25146026611328126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,16384,0.03870720068613688
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,65536,0.3944448153177897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,65536,0.23084373474121095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,12288,0.03321173389752706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,16384,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,16384,0.03942399819691976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,12288,0.030754133065541582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,12288,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,10240,0.029969066381454468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,10240,0.026210133234659833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,10240,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,8192,0.02703253428141276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,8192,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,8192,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,7168,0.011361066500345867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,7168,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,6144,0.027511467536290485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,7168,0.02887679934501648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,6144,0.016721065839131674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,6144,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,5120,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,4096,0.01167039970556895
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,5120,0.02491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,3584,0.02218666672706604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,4096,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,5120,0.014878933628400167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,3584,0.010922666390736897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,4096,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,3584,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,3072,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,3072,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,2560,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,3072,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,2560,0.008970666925112407
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,2048,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,2560,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,2048,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,2048,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,1536,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,1536,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,1024,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,1536,0.005562666555245717
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,768,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,768,0.004465066889921824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,768,0.004537599782148996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,1024,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,1024,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,512,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,512,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,256,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,512,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,256,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,128,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,256,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,128,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,64,0.01392213304837545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,2048,32,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,2048,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,64,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,2048,32,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,65536,0.15366400082906087
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,16384,0.03853653271993001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,65536,0.2933727900187174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,16384,0.018739199638366698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,16384,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,65536,0.10284372965494792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,12288,0.03293866713841756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,12288,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,12288,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,10240,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,10240,0.029934932788213093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,10240,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,8192,0.02693120042483012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,8192,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,7168,0.028330665826797486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,7168,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,8192,0.012458667159080505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,6144,0.026555732885996504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,6144,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,7168,0.01890986760457357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,6144,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,5120,0.00921493371327718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,5120,0.01395840048789978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,4096,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,4096,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,4096,0.022766933838526408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,5120,0.02481493353843689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,3584,0.010818133751551311
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,3584,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,3584,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,3072,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,3072,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,3072,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,2560,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,2560,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,2048,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,2560,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,2048,0.006996266543865204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,2048,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,1536,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,1536,0.006206933160622915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,1536,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,768,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,1024,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,768,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,768,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,1024,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,512,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,1024,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,512,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,256,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,128,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,256,0.01525759994983673
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,128,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,256,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,128,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,64,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1536,32,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1536,32,0.003786666691303253
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1536,32,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,65536,0.10810026327768962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,16384,0.03853653271993001
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,65536,0.07430826822916667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,16384,0.012012799580891926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,16384,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,65536,0.1925802707672119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,12288,0.03276693423589071
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,10240,0.030037333567937214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,12288,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,10240,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,12288,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,10240,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,8192,0.02693120042483012
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,8192,0.012215466300646464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,8192,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,7168,0.028398933013280232
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,7168,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,6144,0.007233066856861115
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,7168,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,6144,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,6144,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,5120,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,5120,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,4096,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,3584,0.02187946637471517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,5120,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,4096,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,4096,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,3584,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,3584,0.007403733332951863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,3072,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,3072,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,3072,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,2560,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,2560,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,2560,0.020002132654190062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,2048,0.006995200117429097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,2048,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,2048,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,1536,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,1536,0.005764266848564148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,1536,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,1024,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,768,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,1024,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,768,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,512,0.01590506633122762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,1024,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,768,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,512,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,512,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,256,0.014842666188875833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,256,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,128,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,256,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,128,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,64,0.013892267147699991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,128,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,64,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,1024,32,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,1024,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,16384,0.03829760154088338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,1024,32,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,65536,0.07430079778035482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,65536,0.10796373685201008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,65536,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,16384,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,16384,0.010476799805959065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,12288,0.032529066006342575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,12288,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,10240,0.029730133215586346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,12288,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,10240,0.009009066224098205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,8192,0.026826665798823042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,8192,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,8192,0.012151466806729634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,10240,0.014643200238545737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,7168,0.028433066606521607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,7168,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,6144,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,6144,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,7168,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,5120,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,6144,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,5120,0.009142399827639262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,5120,0.006929066777229309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,4096,0.022528000672658286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,4096,0.00784853349129359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,4096,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,3584,0.0055285334587097164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,3584,0.022016000747680665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,3072,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,3584,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,2560,0.019796266158421835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,3072,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,2560,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,3072,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,2560,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,2048,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,2048,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,2048,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,1536,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,1536,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,1536,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,1024,0.004845866560935974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,1024,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,1024,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,768,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,768,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,768,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,512,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,512,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,512,0.004261333247025808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,256,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,128,0.015218133727709452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,256,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,256,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,128,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,64,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,64,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,768,32,0.01320319970448812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,768,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,16384,0.03829760154088338
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,65536,0.07447893619537353
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,768,32,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,65536,0.10799787044525147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,65536,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,16384,0.009142399827639262
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,16384,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,12288,0.03246080080668132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,12288,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,12288,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,8192,0.02658453385035197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,10240,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,8192,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,10240,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,10240,0.029764266808827718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,8192,0.012219732999801636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,7168,0.028296534220377607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,7168,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,7168,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,6144,0.010136533776919048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,5120,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,6144,0.026726400852203368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,6144,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,5120,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,5120,0.024678399165471397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,4096,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,4096,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,4096,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,3584,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,3072,0.02068480054537455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,3584,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,3584,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,3072,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,3072,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,2560,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,2560,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,2560,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,2048,0.018705066045125326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,2048,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,2048,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,1536,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,1536,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,1024,0.016684800386428833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,1024,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,1024,0.0048096001148223875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,1536,0.01795413295427958
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,768,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,768,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,512,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,768,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,512,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,512,0.004364799956480662
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,256,0.015154133240381876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,256,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,128,0.0027637332677841187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,128,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,64,0.013858133554458618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,128,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,512,32,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,512,32,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,512,32,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,16384,0.038330666224161786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,65536,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,65536,0.10775893529256184
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,65536,0.07430826822916667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,16384,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,16384,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,12288,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,12288,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,10240,0.029627732435862225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,10240,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,12288,0.03259733319282532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,10240,0.01474346617857615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,8192,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,8192,0.026826665798823042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,8192,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,7168,0.028296534220377607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,7168,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,7168,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,6144,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,6144,0.010103467106819152
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,6144,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,5120,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,5120,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,5120,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,4096,0.022357332706451415
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,4096,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,3584,0.021606399615605672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,3584,0.007300266623497009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,3584,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,3072,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,4096,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,3072,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,3072,0.006996266543865204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,2560,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,2560,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,2560,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,2048,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,2048,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,1536,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,2048,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,1536,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,1536,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,1024,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,1024,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,1024,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,768,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,768,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,512,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,512,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,256,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,512,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,768,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,256,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,256,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,128,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,128,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,64,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,256,32,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,64,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,256,32,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,256,32,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,65536,0.0741376002629598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,65536,0.10765653451283771
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,65536,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,16384,0.038395734628041585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,16384,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,12288,0.03242239952087402
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,16384,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,12288,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,12288,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,10240,0.029627732435862225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,10240,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,10240,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,8192,0.004775466521581014
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,8192,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,7168,0.02815999984741211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,7168,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,7168,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,8192,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,6144,0.026419200499852497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,6144,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,6144,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,5120,0.0045045331120491024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,5120,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,5120,0.02471253275871277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,4096,0.022521599133809408
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,4096,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,4096,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,3584,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,3584,0.007334400216738384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,3584,0.02177706758181254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,3072,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,2560,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,3072,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,3072,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,2560,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,2048,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,2560,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,2048,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,2048,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,1536,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,1536,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,1536,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,1024,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,1024,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,768,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,768,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,768,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,1024,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,512,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,512,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,256,0.015084800124168397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,512,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,256,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,256,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,128,0.015359999736150107
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,128,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,64,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,64,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,64,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,128,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,128,32,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,128,32,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,65536,0.07406933307647705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,65536,0.10799787044525147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,16384,0.03846826553344727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,65536,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,16384,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,12288,0.032391466697057086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,16384,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,12288,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,12288,0.01651946703592936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,10240,0.005386666456858317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,10240,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,8192,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,10240,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,8192,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,8192,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,7168,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,7168,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,7168,0.028228267033894854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,6144,0.026726400852203368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,6144,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,5120,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,5120,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,4096,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,6144,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,4096,0.013926399747530618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,5120,0.017203199863433837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,3584,0.02167466680208842
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,4096,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,3584,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,3584,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,3072,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,3072,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,3072,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,2560,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,2560,0.009794132908185323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,2048,0.01853440006573995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,2560,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,2048,0.008122666676839193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,2048,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,1536,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,1024,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,1536,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,1536,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,768,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,768,0.004296533266703288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,1024,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,1024,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,768,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,512,0.01593493322531382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,256,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,256,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,256,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,512,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,512,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,128,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,128,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,64,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,64,0.0023210667073726653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,64,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,64,32,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,65536,0.10741759936014812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,64,32,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,64,32,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,65536,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,65536,0.07406933307647705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,16384,0.03819520076115926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,16384,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,12288,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,16384,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,10240,0.029730133215586346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,12288,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,10240,0.004980266590913137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,10240,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,12288,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,8192,0.02648746569951375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,8192,0.02611200014750163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,7168,0.028296534220377607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,7168,0.01109333336353302
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,6144,0.026282666126887004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,8192,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,7168,0.023176532983779908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,6144,0.009966933727264404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,6144,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,5120,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,5120,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,5120,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,4096,0.022423466046651207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,4096,0.014165332913398743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,4096,0.007952000200748443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,3584,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,3584,0.012663466731707254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,3584,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,3072,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,3072,0.02058026591936747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,3072,0.00678719977537791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,2560,0.019797333081563315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,2560,0.006171733140945435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,2560,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,2048,0.008254933357238769
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,2048,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,1536,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,1536,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,2048,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,1024,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,1536,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,1024,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,1024,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,768,0.01641386648019155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,768,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,768,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,512,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,512,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,256,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,512,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,128,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,256,0.015324800213177999
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,128,0.0024512000381946564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,64,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,128,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,64,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,4,32,32,0.0023552000522613524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,4,32,32,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,4,32,32,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,12288,1.255936050415039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,16384,1.6689834594726562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,12288,2.5818089803059894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,12288,1.3451605478922526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,10240,1.0443093617757162
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,10240,2.012979253133138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,16384,3.2203435262044273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,10240,1.107421875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,8192,0.8393728256225585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,8192,1.6149162292480468
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,8192,0.8841215769449869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,7168,0.7315114974975586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,7168,0.7788543701171875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,6144,0.6699008305867513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,7168,1.4117205301920572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,6144,0.5761023839314778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,6144,1.2088661193847656
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,5120,0.5599903742472331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,5120,0.9246719996134439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,5120,0.5246272087097168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,4096,0.4481024106343587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,4096,0.8062976201375326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,3584,0.3969024022420248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,4096,0.42100054423014327
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,3584,0.3685717264811198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,3072,0.34092372258504233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,3584,0.7083690643310547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,3072,0.5564053217569987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,3072,0.2898602803548177
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,2560,0.24384853045145669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,2048,0.23244800567626953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,2560,0.2866175969441732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,2048,0.4054688135782878
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,1536,0.07383039792378744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,2560,0.46493012110392257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,2048,0.19445759455362957
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,1536,0.2805760065714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,1536,0.043622398376464845
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,1024,0.03242133259773254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,1024,0.024371200799942018
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,1024,0.18848427136739093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,768,0.028808534145355225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,768,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,512,0.02617599964141846
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,768,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,512,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,512,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,256,0.0211626668771108
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,256,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,256,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,128,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,64,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,64,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,128,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,128,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,64,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,65536,32,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,65536,32,0.0047082667549451195
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,65536,32,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,16384,0.46062399546305344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,16384,0.4333567937215169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,16384,0.8360618591308594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,65536,1.5140864054361978
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,65536,1.7980074564615887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,12288,0.6648106892903646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,12288,0.3590144157409668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,65536,3.234884389241537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,12288,0.3297589302062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,10240,0.2686976114908854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,10240,0.509713077545166
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,8192,0.236407470703125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,10240,0.26996053059895836
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,8192,0.41949866612752273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,8192,0.2161322593688965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,7168,0.2098858674367269
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,7168,0.17012054125467937
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,7168,0.32607574462890626
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,6144,0.08024746576944987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,6144,0.03351893424987793
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,5120,0.03331413269042969
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,6144,0.28675413131713867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,5120,0.23374506632486977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,4096,0.029525333642959596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,4096,0.18759679794311523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,4096,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,5120,0.023756800095240276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,3584,0.02792106668154399
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,3584,0.16414720217386883
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,3584,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,3072,0.026521599292755126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,3072,0.04792319933573405
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,3072,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,2560,0.024541866779327393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,2560,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,2048,0.02239146629969279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,2560,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,2048,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,1536,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,1536,0.01590506633122762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,1536,0.010575999816258747
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,2048,0.012356266379356384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,1024,0.01904426614443461
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,1024,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,1024,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,768,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,768,0.008425600330034892
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,512,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,768,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,512,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,512,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,256,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,256,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,128,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,256,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,128,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,64,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,64,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,128,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,64,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,16384,32,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,16384,32,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,16384,32,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,16384,0.35392853418986003
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,16384,0.6227626800537109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,65536,1.1438079833984376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,65536,1.3603839874267578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,65536,2.4235689798990885
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,16384,0.32802133560180663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,12288,0.27385174433390297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,12288,0.22408533096313477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,12288,0.4939093271891276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,10240,0.20520960489908852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,10240,0.38331734339396156
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,8192,0.3006122589111328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,10240,0.22527999877929689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,8192,0.05085866848627726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,8192,0.03993599812189738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,7168,0.03635199864705403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,7168,0.029013333717981975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,6144,0.03287039995193482
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,7168,0.2797909418741862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,6144,0.02542933424313863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,5120,0.029794132709503172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,6144,0.25180160204569496
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,5120,0.18633386294047039
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,5120,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,4096,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,4096,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,3584,0.02525866627693176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,3584,0.027716267108917236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,4096,0.030583467086156207
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,3584,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,3072,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,3072,0.02430293361345927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,3072,0.014916266997655234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,2560,0.020682666699091592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,2560,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,2560,0.02269759972890218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,2048,0.011946666240692138
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,2048,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,2048,0.02146986722946167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,1536,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,1536,0.013653332988421122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,1024,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,1536,0.009928533434867859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,1024,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,1024,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,768,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,512,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,768,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,768,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,512,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,512,0.0067221333583196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,256,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,128,0.014537599682807923
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,256,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,256,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,64,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,64,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,128,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,128,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,64,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,12288,32,0.013173333803812661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,12288,32,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,12288,32,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,16384,0.3010218620300293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,16384,0.4950357437133789
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,65536,1.1344554901123047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,65536,0.9680213292439779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,16384,0.250982395807902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,65536,2.0169728597005205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,12288,0.23231147130330404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,12288,0.40755198796590164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,12288,0.19066880544026693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,10240,0.13311999638875324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,10240,0.11560959815979004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,8192,0.03338239987691243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,10240,0.31993172963460287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,8192,0.2642944018046061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,8192,0.029047467311223346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,7168,0.20534613927205406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,7168,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,6144,0.03051519989967346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,7168,0.03321066697438558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,6144,0.18094080289204914
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,6144,0.02399466633796692
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,5120,0.02815999984741211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,5120,0.08710827032725016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,5120,0.020885332425435384
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,4096,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,4096,0.02215253313382467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,3584,0.024064000447591147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,4096,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,3584,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,3584,0.01983039975166321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,3072,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,3072,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,3072,0.022937599817911783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,2560,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,2560,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,2560,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,2048,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,2048,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,1536,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,2048,0.011603200435638427
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,1536,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,1536,0.010035199920336406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,1024,0.01791999936103821
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,1024,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,1024,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,768,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,512,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,768,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,512,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,512,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,768,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,256,0.01576959987481435
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,128,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,256,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,128,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,256,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,128,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,64,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,64,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,64,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,10240,32,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,10240,32,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,10240,32,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,16384,0.39304533004760744
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,65536,1.6112640380859375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,65536,0.880737050374349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,16384,0.2400597254435221
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,16384,0.22661120096842446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,65536,0.8928597132364908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,12288,0.07809706528981528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,12288,0.323857053120931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,12288,0.038293333848317464
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,10240,0.021845332781473794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,8192,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,10240,0.03515733480453491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,8192,0.030822400252024335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,10240,0.23558826446533204
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,7168,0.030001066128412884
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,8192,0.21695146560668946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,7168,0.16390825907389323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,7168,0.016961065928141277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,6144,0.02788693308830261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,6144,0.01518933375676473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,6144,0.0475818673769633
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,5120,0.02388906677563985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,5120,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,5120,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,4096,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,4096,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,4096,0.010990933577219645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,3584,0.010376532872517902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,3584,0.022869332631429037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,3584,0.01771519978841146
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,3072,0.02177706758181254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,3072,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,3072,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,2560,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,2560,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,2560,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,2048,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,2048,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,1536,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,2048,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,1536,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,1536,0.00883840024471283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,1024,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,1024,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,1024,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,768,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,768,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,512,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,768,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,256,0.015660799543062844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,512,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,512,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,256,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,256,0.00443200021982193
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,128,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,128,0.004127999891837438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,64,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,64,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,128,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,8192,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,8192,32,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,8192,32,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,16384,0.2103637377421061
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,65536,0.7722325642903646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,65536,1.4069077809651693
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,65536,0.694818115234375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,16384,0.19128319422403972
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,12288,0.037273601690928145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,16384,0.3638271967569987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,12288,0.023893332481384276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,12288,0.27446613311767576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,10240,0.03303893407185872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,8192,0.02959360082944234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,10240,0.23022933006286622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,10240,0.020889600118001304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,8192,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,8192,0.1693354606628418
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,7168,0.02877440055211385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,7168,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,7168,0.04904959996541341
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,6144,0.027306665976842243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,6144,0.025156267484029132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,6144,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,5120,0.02542933424313863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,5120,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,4096,0.02290346622467041
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,5120,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,4096,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,4096,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,3584,0.02218559980392456
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,3584,0.015871999661127727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,3584,0.009795199831326802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,3072,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,3072,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,2560,0.020308266083399452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,3072,0.01437013347943624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,2560,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,2560,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,2048,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,2048,0.010376532872517902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,2048,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,1536,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,1536,0.006549333532651265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,1024,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,1536,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,1024,0.006792533397674561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,1024,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,768,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,768,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,768,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,512,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,512,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,256,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,512,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,256,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,128,0.014436266819636025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,256,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,128,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,128,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,64,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,7168,32,0.01300373375415802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,64,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,7168,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,64,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,7168,32,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,65536,0.6656341552734375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,65536,1.2031317392985026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,16384,0.050756267706553136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,65536,0.6717098871866862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,16384,0.044202665487925216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,16384,0.2963456153869629
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,12288,0.0339626669883728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,12288,0.23924053510030113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,12288,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,10240,0.03092479904492696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,10240,0.18725546201070148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,8192,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,10240,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,8192,0.031914667288462324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,7168,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,8192,0.015495466192563376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,7168,0.025565866629282636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,7168,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,6144,0.025702399015426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,6144,0.02198186715443929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,6144,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,5120,0.024234666426976522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,5120,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,4096,0.02222080032030741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,5120,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,4096,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,4096,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,3584,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,3584,0.01437013347943624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,3584,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,3072,0.012663466731707254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,3072,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,2560,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,3072,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,2560,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,2560,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,2048,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,2048,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,2048,0.009623466928799947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,1536,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,1536,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,1024,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,1024,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,1536,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,1024,0.005529599885145823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,512,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,768,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,768,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,768,0.005356800059477488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,512,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,512,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,256,0.015526400009791056
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,256,0.0037151999771595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,256,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,64,0.013346133629480996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,64,0.0028949332733949023
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,128,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,128,0.014436266819636025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,128,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,6144,32,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,64,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,6144,32,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,6144,32,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,16384,0.03925333420435588
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,65536,0.5555882771809896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,65536,0.4965375900268555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,65536,1.0016426722208658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,16384,0.24244906107584635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,16384,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,12288,0.03327999909718831
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,12288,0.19391147295633954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,12288,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,10240,0.03031040032704671
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,10240,0.11403946876525879
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,10240,0.016827734311421712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,8192,0.02815999984741211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,8192,0.013890133301417033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,7168,0.026862933238347368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,8192,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,7168,0.023825067281723022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,7168,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,6144,0.011803733309110005
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,6144,0.025565866629282636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,6144,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,5120,0.023654399315516154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,5120,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,5120,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,4096,0.021811199188232423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,4096,0.008874666690826417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,4096,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,3584,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,3584,0.021401600042978922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,3072,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,3584,0.008567466338475545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,3072,0.007747200131416321
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,3072,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,2560,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,2560,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,2560,0.0106495996316274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,2048,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,2048,0.018943999210993448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,2048,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,1536,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,1536,0.005866666634877523
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,1536,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,1024,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,1024,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,1024,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,768,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,768,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,512,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,768,0.0049792001644770306
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,256,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,512,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,256,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,256,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,128,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,128,0.003957333415746689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,128,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,64,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,5120,32,0.013209600249926248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,5120,32,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,5120,32,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,65536,0.4497066815694173
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,65536,0.7991637547810873
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,65536,0.446395746866862
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,16384,0.03857066631317139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,16384,0.19295573234558105
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,16384,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,12288,0.03290453354517619
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,12288,0.02969599962234497
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,10240,0.03007146716117859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,12288,0.01771093408266703
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,10240,0.02934933304786682
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,10240,0.015018666783968607
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,8192,0.02744320034980774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,8192,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,8192,0.023819732666015624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,7168,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,7168,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,7168,0.02177706758181254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,6144,0.024985599517822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,6144,0.018874667088190713
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,6144,0.010752000411351522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,5120,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,4096,0.02174293398857117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,5120,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,4096,0.013346133629480996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,5120,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,4096,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,3584,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,3584,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,3584,0.020992000897725425
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,3072,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,3072,0.010820266604423524
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,2560,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,2560,0.01955839991569519
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,3072,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,2560,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,2048,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,2048,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,2048,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,1536,0.01764586567878723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,1536,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,1536,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,1024,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,1024,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,1024,0.005085866649945577
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,768,0.0049450665712356566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,768,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,512,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,768,0.005560533205668131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,512,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,256,0.015358933806419372
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,512,0.004538666705290476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,256,0.003583999971548716
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,256,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,128,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,128,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,128,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,64,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,64,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,4096,32,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,4096,32,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,64,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,4096,32,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,65536,0.3968000094095866
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,65536,0.39314772288004557
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,65536,0.7064917246500652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,16384,0.03884373505910237
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,16384,0.17148586908976238
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,16384,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,12288,0.03430399894714355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,12288,0.03283626635869344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,12288,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,10240,0.030003199974695843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,10240,0.029047467311223346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,8192,0.02696533401807149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,10240,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,8192,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,8192,0.012458667159080505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,7168,0.026685865720113118
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,7168,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,7168,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,6144,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,6144,0.024985599517822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,6144,0.0105813334385554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,5120,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,5120,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,4096,0.021572266022364298
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,4096,0.013208533326784769
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,5120,0.009386666615804036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,4096,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,3584,0.007642666498819987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,3584,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,3584,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,3072,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,3072,0.010752000411351522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,3072,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,2560,0.01925119956334432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,2560,0.009482666850090027
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,2560,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,2048,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,2048,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,2048,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,1536,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,1536,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,1024,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,1536,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,1024,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,1024,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,768,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,768,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,512,0.015871999661127727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,768,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,512,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,256,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,512,0.004534400006135305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,128,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,256,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,256,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,128,0.002867199977238973
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,64,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,64,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,64,0.0038880000511805216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3584,32,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3584,32,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3584,32,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,65536,0.34481493631998694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,16384,0.03850239912668864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,65536,0.3394901275634766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,65536,0.6066858927408855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,16384,0.021191465854644775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,12288,0.03273386756579082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,16384,0.05085866848627726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,12288,0.033928533395131424
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,10240,0.029832533995310467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,10240,0.028706133365631104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,12288,0.01699733336766561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,10240,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,8192,0.026895999908447266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,8192,0.023040000597635904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,8192,0.012522666652997335
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,7168,0.026828799645106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,7168,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,7168,0.011571199695269267
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,6144,0.025120000044504803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,6144,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,6144,0.01037440001964569
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,5120,0.02321066657702128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,5120,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,5120,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,4096,0.021538132429122926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,4096,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,3584,0.011975466211636861
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,4096,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,3584,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,3584,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,3072,0.020309333006540933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,3072,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,2560,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,3072,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,2560,0.0068256000677744556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,2048,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,2560,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,2048,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,1536,0.01767786741256714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,1536,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,2048,0.006144000093142191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,1024,0.01678826610247294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,1536,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,1024,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,1024,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,768,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,768,0.005217066903909048
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,768,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,512,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,512,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,512,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,256,0.014984533190727234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,256,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,256,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,128,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,128,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,128,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,64,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,64,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,64,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,3072,32,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,3072,32,0.00402453343073527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,3072,32,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,65536,0.4957866668701172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,65536,0.29409281412760413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,65536,0.2840575853983561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,16384,0.03833173513412476
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,16384,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,16384,0.025015467405319215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,12288,0.03259413242340088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,12288,0.01952426632245382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,12288,0.01689599951108297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,10240,0.017232000827789307
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,10240,0.030003199974695843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,10240,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,8192,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,8192,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,8192,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,7168,0.02645333409309387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,7168,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,7168,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,6144,0.017851734161376955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,6144,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,6144,0.010373333096504211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,5120,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,5120,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,5120,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,4096,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,4096,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,4096,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,3584,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,3584,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,3584,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,3072,0.007130666573842366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,3072,0.01993280053138733
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,3072,0.010376532872517902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,2560,0.01962453325589498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,2560,0.009079466263453167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,2560,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,2048,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,2048,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,2048,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,1536,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,1536,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,1536,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,1024,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,1024,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,1024,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,768,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,768,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,512,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,768,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,256,0.015286399920781454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,256,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,512,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,512,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,256,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,128,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,128,0.003920000046491623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,64,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,64,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,128,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2560,32,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2560,32,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2560,32,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,65536,0.2326869328816732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,65536,0.24552106857299805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,65536,0.3942399978637695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,16384,0.03863893349965413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,16384,0.039318398634592695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,16384,0.020889600118001304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,12288,0.030133332808812457
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,12288,0.032494932413101196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,12288,0.016724266608556113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,10240,0.029658667246500653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,10240,0.025702399015426636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,10240,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,8192,0.026862933238347368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,8192,0.01225279966990153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,8192,0.020923733711242676
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,7168,0.026385066906611125
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,7168,0.01884160041809082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,7168,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,6144,0.024541866779327393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,6144,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,5120,0.023101866245269775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,6144,0.010410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,5120,0.014131200313568116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,5120,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,4096,0.021538132429122926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,4096,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,4096,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,3584,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,3584,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,3584,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,3072,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,3072,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,2560,0.018974934021631876
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,3072,0.009591466188430786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,2560,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,2560,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,2048,0.018261333306630455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,2048,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,2048,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,1536,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,1536,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,1536,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,1024,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,1024,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,1024,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,768,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,768,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,768,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,512,0.004296533266703288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,512,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,512,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,256,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,256,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,128,0.014097066720326743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,256,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,128,0.002625066787004471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,128,0.004539733131726583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,64,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,2048,32,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,64,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,2048,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,2048,32,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,65536,0.1362570603688558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,65536,0.29242026011149086
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,16384,0.038024532794952395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,65536,0.09792853196461995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,16384,0.018636800845464072
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,16384,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,12288,0.032290132840474446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,12288,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,12288,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,10240,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,10240,0.02941653331120809
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,8192,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,10240,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,8192,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,8192,0.01088853379090627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,7168,0.018329600493113198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,7168,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,7168,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,6144,0.02446933388710022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,6144,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,6144,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,5120,0.023040000597635904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,5120,0.013892267147699991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,5120,0.009246933460235595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,4096,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,4096,0.011502933502197266
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,4096,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,3584,0.020889600118001304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,3584,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,3584,0.010544000069300334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,3072,0.009316266576449076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,3072,0.019722666343053183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,2560,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,3072,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,2560,0.008123733103275299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,2560,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,2048,0.018158932526906334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,2048,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,2048,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,1536,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,1536,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,1536,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,1024,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,1024,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,1024,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,768,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,768,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,768,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,512,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,512,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,256,0.01532266636689504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,512,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,256,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,128,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,256,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,128,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,64,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,128,0.003818666686614355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,64,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,64,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1536,32,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1536,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1536,32,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,65536,0.10649600028991699
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,65536,0.07420586744944255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,65536,0.19056639671325684
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,16384,0.03805866638819377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,16384,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,16384,0.01259519954522451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,12288,0.03222186764081319
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,12288,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,12288,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,10240,0.02945599953333537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,10240,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,8192,0.026692267258961993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,10240,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,8192,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,8192,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,7168,0.02635093331336975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,7168,0.011264000336329143
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,7168,0.0075765331586201985
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,6144,0.024334933360417685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,6144,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,6144,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,5120,0.022937599817911783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,5120,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,5120,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,4096,0.021333332856496176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,4096,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,4096,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,3584,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,3584,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,3584,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,3072,0.005727999905745188
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,3072,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,3072,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,2560,0.005594666798909505
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,2560,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,2560,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,2048,0.018295466899871826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,2048,0.007031466563542683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,2048,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,1536,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,1536,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,1536,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,1024,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,1024,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,1024,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,768,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,768,0.004398933549722036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,768,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,512,0.003618133316437403
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,512,0.015460266669591268
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,512,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,256,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,256,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,256,0.004198400179545084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,128,0.014233600099881491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,128,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,64,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,64,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,1024,32,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,1024,32,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,1024,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,65536,0.10615253448486328
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,65536,0.027989333868026732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,65536,0.07410346666971843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,16384,0.037853864828745525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,16384,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,12288,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,16384,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,12288,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,12288,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,10240,0.02921813329060872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,10240,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,10240,0.014503467082977294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,8192,0.007677866518497467
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,8192,0.026658133665720625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,7168,0.026146133740743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,8192,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,7168,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,7168,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,6144,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,6144,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,5120,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,6144,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,5120,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,5120,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,4096,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,4096,0.007850666840871174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,4096,0.006038400034109751
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,3584,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,3072,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,3584,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,3584,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,3072,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,2560,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,2560,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,3072,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,2560,0.006348800162474315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,2048,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,2048,0.004845866560935974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,2048,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,1536,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,1536,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,1536,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,1024,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,1024,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,1024,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,768,0.016587733229001363
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,768,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,512,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,768,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,512,0.0036159999668598174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,512,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,256,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,256,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,256,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,128,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,64,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,128,0.014131200313568116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,64,0.0038549333810806276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,64,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,768,32,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,768,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,768,32,0.0038549333810806276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,65536,0.10584746996561686
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,16384,0.03805866638819377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,65536,0.0741376002629598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,65536,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,16384,0.00921493371327718
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,16384,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,12288,0.03242666721343994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,12288,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,12288,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,10240,0.02959360082944234
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,10240,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,10240,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,8192,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,8192,0.026521599292755126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,8192,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,7168,0.026009599367777508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,7168,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,7168,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,6144,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,6144,0.024573866526285806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,6144,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,5120,0.023037866751352946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,5120,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,5120,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,4096,0.02109439969062805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,4096,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,4096,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,3584,0.020546134312947592
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,3584,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,3584,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,3072,0.019694934288660683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,3072,0.00678719977537791
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,3072,0.004949333270390829
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,2560,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,2560,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,2560,0.0065194666385650635
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,2048,0.005734399954477946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,2048,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,2048,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,1536,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,1536,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,1536,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,1024,0.01665066679318746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,1024,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,1024,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,768,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,512,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,768,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,768,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,512,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,512,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,256,0.015253333250681558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,256,0.0029696000119050344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,256,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,128,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,128,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,64,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,64,0.0025557334224383037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,64,0.003852800031503042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,512,32,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,512,32,0.0037546666959921518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,512,32,0.0024906667570273083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,65536,0.105949862798055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,65536,0.016520532965660095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,16384,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,65536,0.07417173385620117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,16384,0.008188800017038981
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,12288,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,16384,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,12288,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,12288,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,10240,0.029525333642959596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,10240,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,10240,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,8192,0.026415999730428057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,8192,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,7168,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,7168,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,7168,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,8192,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,6144,0.024473599592844644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,6144,0.004844800134499868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,6144,0.01016533374786377
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,5120,0.023005867004394533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,5120,0.004806399842103322
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,4096,0.021504000822703043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,4096,0.007987200220425924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,4096,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,5120,0.009284266829490661
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,3584,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,3584,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,3584,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,3072,0.01949013272921244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,3072,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,3072,0.006963199873765309
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,2560,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,2560,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,2560,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,2048,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,2048,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,2048,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,1536,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,1536,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,1536,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,1024,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,1024,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,768,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,1024,0.004710400104522705
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,768,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,768,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,512,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,256,0.015223466356595359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,512,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,512,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,256,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,256,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,128,0.0026602665583292644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,128,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,128,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,64,0.013275733590126038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,64,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,64,0.004060799876848856
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,256,32,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,256,32,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,65536,0.105949862798055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,256,32,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,65536,0.010752000411351522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,65536,0.07417173385620117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,16384,0.037956265608469646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,16384,0.006041599810123444
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,16384,0.020753065745035805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,12288,0.03211946686108907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,12288,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,12288,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,10240,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,10240,0.02955946723620097
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,10240,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,8192,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,8192,0.026589866479237872
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,8192,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,7168,0.02573653260866801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,7168,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,7168,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,6144,0.024576000372568765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,6144,0.020070399840672812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,6144,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,5120,0.023005867004394533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,5120,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,5120,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,4096,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,4096,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,4096,0.014028799533843995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,3584,0.020718934138615926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,3072,0.020104533433914183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,3584,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,3584,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,3072,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,3072,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,2560,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,2560,0.006345599889755249
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,2560,0.00962559978167216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,2048,0.018090667327245076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,2048,0.008635733524958293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,2048,0.005905066430568695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,1536,0.017271467049916587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,1536,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,1536,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,1024,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,1024,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,1024,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,768,0.004469333092371622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,768,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,768,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,512,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,512,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,512,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,256,0.015155200163523355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,256,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,128,0.014233600099881491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,128,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,128,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,64,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,64,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,128,32,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,128,32,0.002384000023206075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,128,32,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,65536,0.10625600020090739
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,65536,0.008089600006739299
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,65536,0.07478613058725993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,16384,0.03799039920171102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,16384,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,12288,0.032085333267847696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,16384,0.020787199338277183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,12288,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,10240,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,12288,0.016622933745384216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,10240,0.02935466567675273
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,10240,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,8192,0.026521599292755126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,8192,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,8192,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,7168,0.0229695995648702
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,7168,0.026009599367777508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,7168,0.01129813293615977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,6144,0.024337067206700643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,6144,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,6144,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,5120,0.02272640069325765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,5120,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,5120,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,4096,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,4096,0.014095999797185264
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,3584,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,3584,0.012694399555524191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,4096,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,3584,0.007338666419188182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,3072,0.011059199770291645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,3072,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,3072,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,2560,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,2560,0.009727999567985535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,2560,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,2048,0.01812480092048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,2048,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,2048,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,1536,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,1536,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,1536,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,1024,0.016793600718180337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,768,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,1024,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,1024,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,768,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,768,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,512,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,512,0.0036864000062147772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,256,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,256,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,128,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,128,0.004331733286380768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,128,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,64,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,64,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,64,32,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,64,32,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,64,32,0.003819733361403147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,65536,0.105949862798055
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,65536,0.07403519948323568
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,65536,0.0076799998680750535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,16384,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,16384,0.037853864828745525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,16384,0.020821332931518555
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,12288,0.005188266436258951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,12288,0.032153600454330446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,12288,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,10240,0.014300800363222756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,10240,0.02945706645647685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,10240,0.004915200173854828
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,8192,0.02645333409309387
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,8192,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,8192,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,7168,0.026009599367777508
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,7168,0.02307413419087728
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,7168,0.011294933160146077
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,6144,0.024576000372568765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,5120,0.022971733411153158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,6144,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,6144,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,5120,0.009113599856694538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,5120,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,4096,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,4096,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,3584,0.02065066695213318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,4096,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,3584,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,3584,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,3072,0.011195733149846395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,3072,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,3072,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,2560,0.018773333231608073
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,2560,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,2560,0.006382933259010315
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,2048,0.017988266547520955
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,2048,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,2048,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,1536,0.01761173407236735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,1536,0.006683733562628429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,1536,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,1024,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,1024,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,768,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,1024,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,512,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,768,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,768,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,512,0.0035146666069825493
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,512,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,256,0.015428266922632852
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,128,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,128,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,256,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,128,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,64,0.012969600160916648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,64,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,64,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,2,32,32,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,2,32,32,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,2,32,32,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,16384,1.5143253326416015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,16384,3.129241689046224
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,16384,1.7638741811116536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,12288,1.1252394358317057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,12288,1.3473108927408854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,12288,2.3776255289713544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,10240,1.1031893412272136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,10240,1.9733503977457683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,10240,1.0481664021809896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,8192,0.8838144302368164
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,8192,1.402026621500651
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,8192,0.8356864293416342
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,7168,0.7765674591064453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,7168,0.7345834732055664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,7168,1.3780651092529297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,6144,0.6682624181111654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,6144,1.0457087834676106
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,6144,0.6296576182047526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,5120,0.5579434712727864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,5120,0.9243989308675131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,5120,0.4717877388000488
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,4096,0.7140010833740235
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,4096,0.44677120844523116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,4096,0.4173823992411296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,3584,0.3959808031717936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,3584,0.3311957359313965
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,3584,0.69573974609375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,3072,0.3137194633483887
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,3072,0.5296469370524088
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,3072,0.28566080729166665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,2560,0.264465061823527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,2560,0.23883093198140465
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,2560,0.43318611780802413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,2048,0.2154815991719564
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,2048,0.3400362650553385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,2048,0.21248000462849936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,1536,0.06683306694030762
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,1536,0.2532693386077881
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,1536,0.032051199674606325
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,1024,0.032358400026957196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,1024,0.17131519317626953
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,1024,0.02321066657702128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,768,0.028467200199762982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,768,0.11953492959340413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,512,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,768,0.02071573336919149
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,512,0.07526400089263915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,512,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,256,0.03679573138554891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,256,0.020035199324289956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,256,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,128,0.017237333456675212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,64,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,128,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,128,0.012151466806729634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,64,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,64,0.011844266454378765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,65536,32,0.004673066735267639
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,65536,32,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,65536,32,0.011809066931406657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,16384,0.4599125226338704
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,16384,0.7581696192423503
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,16384,0.4335957209269206
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,65536,1.5124139149983724
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,12288,0.35607894261678064
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,12288,0.5742933273315429
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,65536,3.147332255045573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,12288,0.2935466766357422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,10240,0.29170347849527994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,10240,0.24255146980285644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,10240,0.47513599395751954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,8192,0.3692202568054199
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,8192,0.23592960039774574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,8192,0.21572267214457191
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,7168,0.20838400522867837
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,7168,0.29071359634399413
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,7168,0.1885866641998291
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,6144,0.24565332730611167
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,6144,0.03696639935175578
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,6144,0.031948800881703696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,5120,0.03249386747678121
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,5120,0.023653332392374674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,5120,0.19698346455891927
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,4096,0.028637866179148357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,4096,0.1531562646230062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,4096,0.019421867529551187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,3584,0.02764799992243449
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,3584,0.1363263924916585
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,3584,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,3072,0.016207999984423318
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,3072,0.02611200014750163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,3072,0.11045546531677246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,2560,0.024166399240493776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,2560,0.09594879945119222
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,2560,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,2048,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,2048,0.06492160161336263
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,2048,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,1536,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,1536,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,1024,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,1536,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,1024,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,1024,0.01460906664530436
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,768,0.018363734086354576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,768,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,768,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,512,0.0168938676516215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,512,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,512,0.007609599828720092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,256,0.01495039959748586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,256,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,256,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,128,0.013686399658521017
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,128,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,128,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,64,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,64,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,64,0.0054613331953684485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,16384,32,0.013241599996884665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,16384,32,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,16384,32,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,16384,0.35304107666015627
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,65536,1.1443541208902994
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,16384,0.5632341384887696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,16384,0.3279189427693685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,65536,2.3524693806966144
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,65536,1.3576821645100912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,12288,0.2753194808959961
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,12288,0.41837228139241533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,12288,0.2500266710917155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,10240,0.22435305913289388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,10240,0.34877440134684246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,8192,0.039185067017873124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,10240,0.18496853510538738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,8192,0.26961921056111654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,8192,0.031573333342870075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,7168,0.034508800506591795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,7168,0.2404010613759359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,7168,0.028569600979487103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,6144,0.025634133815765382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,6144,0.03174399932225545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,6144,0.17694719632466632
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,5120,0.028569600979487103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,5120,0.19281919797261554
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,5120,0.021435733636220297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,4096,0.026073600848515826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,4096,0.11171840031941731
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,4096,0.01751040021578471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,3584,0.02505386670430501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,3584,0.1050282637278239
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,3584,0.015837867061297098
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,3072,0.02338026762008667
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,3072,0.08410453001658122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,3072,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,2560,0.022388267517089843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,2560,0.0567957321802775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,2560,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,2048,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,2048,0.019831466674804687
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,2048,0.011400533715883891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,1536,0.01966080069541931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,1536,0.016448000073432924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,1536,0.00976213316122691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,1024,0.017749333381652833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,1024,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,1024,0.008499200145403545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,768,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,512,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,768,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,512,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,768,0.007714133461316426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,512,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,256,0.015052800377209982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,256,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,256,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,128,0.013346133629480996
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,128,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,64,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,128,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,64,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,64,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,12288,32,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,12288,32,0.00279573326309522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,12288,32,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,16384,0.30064640045166013
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,65536,0.9667925516764322
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,16384,0.5012480099995931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,65536,1.1324415842692057
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,65536,1.9551231384277343
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,16384,0.25091412862141926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,12288,0.23101439476013183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,12288,0.3458026568094889
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,12288,0.21114880243937173
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,10240,0.1337685267130534
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,10240,0.2499242623647054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,10240,0.11400533517201741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,8192,0.03324586749076843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,8192,0.19848532676696778
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,8192,0.029047467311223346
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,7168,0.03126613299051921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,7168,0.17274880409240723
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,6144,0.02945706645647685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,7168,0.026726400852203368
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,6144,0.1387818654378255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,6144,0.023586134115854897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,5120,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,5120,0.15213227272033691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,5120,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,4096,0.024439465999603272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,4096,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,4096,0.09574399789174398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,3584,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,3584,0.06980266571044921
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,3072,0.022323199113210044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,3072,0.05413440068562826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,3584,0.015803733468055726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,3072,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,2560,0.012900267044703165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,2560,0.031573333342870075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,2560,0.02187946637471517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,2048,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,2048,0.017097600301106772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,2048,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,1536,0.023756800095240276
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,1536,0.014472533265749613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,1536,0.009827199578285217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,1024,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,1024,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,1024,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,768,0.017339734236399333
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,768,0.006246399879455566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,512,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,512,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,512,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,768,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,256,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,256,0.008803199728329976
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,256,0.005493333439032236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,128,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,128,0.0052906667192777
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,128,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,64,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,64,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,32,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,10240,32,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,10240,64,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,10240,32,0.00508480022350947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,65536,0.8772607803344726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,65536,0.859443219502767
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,16384,0.23613440195719398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,65536,1.55600217183431
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,12288,0.053998935222625735
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,16384,0.3692885398864746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,16384,0.2226858615875244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,12288,0.2754901250203451
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,12288,0.026794666051864625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,10240,0.034065067768096924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,10240,0.0212991992632548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,10240,0.22302719751993813
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,8192,0.030343466997146608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,8192,0.15680853525797528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,8192,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,7168,0.0289792001247406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,7168,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,7168,0.1484117348988851
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,6144,0.026521599292755126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,6144,0.10523306528727214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,6144,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,5120,0.02484906713167826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,5120,0.012759466965993246
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,5120,0.08802986939748128
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,4096,0.02269866665204366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,4096,0.06932480335235595
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,3584,0.02211839954058329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,4096,0.010854400197664897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,3584,0.053282133738199865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,3584,0.010410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,3072,0.021191465854644775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,3072,0.009861333171526591
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,2560,0.020542933543523153
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,2560,0.018568533658981323
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,3072,0.021367466449737547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,2560,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,2048,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,2048,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,1536,0.018531199296315512
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,2048,0.007947733501593272
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,1536,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,1536,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,1024,0.009108266234397889
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,1024,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,1024,0.006109866499900818
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,768,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,768,0.005870933334032694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,768,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,512,0.016178133090337117
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,512,0.005016533533732096
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,512,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,256,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,256,0.0045738667249679565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,256,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,128,0.01330880026022593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,128,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,64,0.012795733412106833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,64,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,128,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,64,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,8192,32,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,8192,32,0.004158933212359746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,8192,32,0.002628266563018163
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,16384,0.20947626431783042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,65536,0.7552000045776367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,65536,0.7697749455769857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,65536,1.3585408528645835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,16384,0.19309226671854657
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,16384,0.3207167943318685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,12288,0.03590293327967326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,12288,0.24241493542989095
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,12288,0.022664533058802287
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,10240,0.019592533508936562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,10240,0.032255999247233075
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,10240,0.18981547355651857
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,8192,0.028808534145355225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,8192,0.13277866840362548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,8192,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,7168,0.10894827047983806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,7168,0.014847999811172486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,7168,0.027374933163324993
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,6144,0.08796160221099854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,6144,0.02573546568552653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,6144,0.013755733768145243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,5120,0.026555732885996504
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,5120,0.024064000447591147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,5120,0.011980799833933513
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,4096,0.02208426594734192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,4096,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,3584,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,4096,0.048946134249369305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,3584,0.01867093245188395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,3584,0.009693866968154908
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,3072,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,3072,0.01925119956334432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,3072,0.008840533097585042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,2560,0.020170666774113975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,2048,0.01884160041809082
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,2560,0.01634986698627472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,2048,0.0073728000124295544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,2048,0.014131200313568116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,2560,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,1536,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,1536,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,1536,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,1024,0.017063466707865398
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,1024,0.008328533172607422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,1024,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,768,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,768,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,768,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,512,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,512,0.015940266847610473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,512,0.0051136001944541935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,256,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,256,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,256,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,128,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,128,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,128,0.0035103999078273775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,64,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,64,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,64,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,7168,32,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,7168,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,7168,32,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,65536,1.1598847707112632
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,65536,0.6639946619669597
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,16384,0.2736810684204102
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,16384,0.05874346494674683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,65536,0.6449151992797851
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,16384,0.027477333943049114
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,12288,0.03386026620864868
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,12288,0.2069162686665853
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,12288,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,10240,0.03054933349291484
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,10240,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,8192,0.027067732810974122
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,8192,0.014882133404413859
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,8192,0.10400426387786865
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,10240,0.13858133951822918
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,7168,0.02621440092722575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,7168,0.09908479849497477
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,7168,0.013823999961217245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,6144,0.024541866779327393
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,5120,0.023142399390538533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,6144,0.07420586744944255
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,5120,0.023108265797297158
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,6144,0.012082133690516155
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,5120,0.011229866743087768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,4096,0.021332265933354695
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,4096,0.017442133029301962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,4096,0.009448533256848652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,3584,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,3584,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,3584,0.01641813317934672
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,3072,0.01976213256518046
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,3072,0.01699519952138265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,3072,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,2560,0.01962666710217794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,2560,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,2560,0.01532586713631948
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,2048,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,2048,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,2048,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,1536,0.011124266187349956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,1536,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,1024,0.016657066345214844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,1536,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,1024,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,1024,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,768,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,768,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,768,0.0053247998158137
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,512,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,512,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,512,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,256,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,256,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,256,0.004403199752171834
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,128,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,128,0.0034815999368826545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,128,0.003991466760635376
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,64,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,64,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,64,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,6144,32,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,6144,32,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,6144,32,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,65536,0.5529258728027344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,65536,0.539306640625
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,65536,0.9624575932820638
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,16384,0.03890879948933919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,16384,0.02187946637471517
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,12288,0.03293866713841756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,16384,0.2274986743927002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,12288,0.17489919662475586
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,12288,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,10240,0.030037333567937214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,10240,0.12694186369578045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,10240,0.015530666708946228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,8192,0.027306665976842243
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,8192,0.0869376023610433
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,8192,0.013071999947230021
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,7168,0.06744746367136636
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,7168,0.012083199620246888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,7168,0.02553173303604126
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,6144,0.024268800020217897
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,6144,0.05215573310852051
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,6144,0.011331199606259664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,5120,0.022937599817911783
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,5120,0.010171733299891154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,5120,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,4096,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,4096,0.014774399995803832
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,4096,0.008772266904513042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,3584,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,3584,0.017476266622543334
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,3584,0.008396800359090168
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,3072,0.01976319948832194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,3072,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,3072,0.007782400151093801
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,2560,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,2560,0.014131200313568116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,2560,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,2048,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,2048,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,2048,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,1536,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,1536,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,1536,0.009864532947540283
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,1024,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,1024,0.016691199938456216
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,1024,0.0055637334783871974
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,768,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,768,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,768,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,512,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,512,0.005083733300367991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,512,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,256,0.014365866780281067
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,256,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,256,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,128,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,128,0.0031061333914597826
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,128,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,64,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,64,0.003067733347415924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,5120,32,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,32,0.004124800115823746
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,5120,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,5120,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,65536,0.4471466700236003
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,65536,0.43287893931070964
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,65536,0.7640746434529622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,16384,0.038126933574676516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,16384,0.15171732902526855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,16384,0.021265067656834922
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,12288,0.03283626635869344
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,12288,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,12288,0.03577173153559367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,10240,0.030037333567937214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,10240,0.09065813223520915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,8192,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,10240,0.014813866217931113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,8192,0.05812906821568807
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,8192,0.01276586651802063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,7168,0.02529279987017314
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,7168,0.03560106754302979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,7168,0.011741866668065388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,6144,0.010645332932472228
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,6144,0.024132267634073893
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,6144,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,5120,0.022562134265899658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,5120,0.015086932977040609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,5120,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,4096,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,4096,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,4096,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,3584,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,3584,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,3072,0.014506666858990987
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,3584,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,3072,0.01966080069541931
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,2560,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,3072,0.007333333293596904
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,2560,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,2560,0.006826666494210561
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,2048,0.010240000486373902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,2048,0.017845332622528076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,1536,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,2048,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,1536,0.009352533022562663
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,1024,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,1024,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,1536,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,1024,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,768,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,768,0.016247466206550598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,768,0.004709333181381226
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,512,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,512,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,512,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,256,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,256,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,256,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,128,0.013004799683888754
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,128,0.00310506671667099
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,128,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,64,0.012970667084058127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,64,0.003885866701602936
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,64,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,4096,32,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,4096,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,4096,32,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,65536,0.3942058563232422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,65536,0.3805184046427409
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,65536,0.6714015960693359
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,16384,0.03809279998143514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,16384,0.12612266540527345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,12288,0.032494932413101196
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,16384,0.02129279971122742
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,12288,0.08669866720835367
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,12288,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,10240,0.02990079919497172
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,10240,0.014745600024859109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,10240,0.0740010658899943
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,8192,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,8192,0.040072532494862874
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,8192,0.01256106694539388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,7168,0.02512213389078776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,7168,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,6144,0.0237226665019989
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,7168,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,6144,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,6144,0.01050879955291748
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,5120,0.02249386707941691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,5120,0.013619200388590495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,4096,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,5120,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,4096,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,4096,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,3584,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,3584,0.014643200238545737
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,3584,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,3072,0.013038933277130127
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,2560,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,3072,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,2560,0.006724266707897187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,2560,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,3072,0.007264000177383423
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,2048,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,2048,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,2048,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,1536,0.017544533809026083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,1536,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,1024,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,1536,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,1024,0.005120000243186951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,1024,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,768,0.016450132926305136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,768,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,768,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,512,0.015564800302187601
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,512,0.0068256000677744556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,512,0.004534400006135305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,256,0.014472533265749613
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,256,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,256,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,128,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,128,0.013209600249926248
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,128,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,64,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,64,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,64,0.003992533435424169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3584,32,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3584,32,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3584,32,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,65536,0.34218241373697916
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,65536,0.5655210494995118
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,65536,0.32805547714233396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,16384,0.037956265608469646
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,16384,0.10202453136444092
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,12288,0.032458666960398355
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,16384,0.021230934063593547
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,12288,0.02757973273595174
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,12288,0.01699840029080709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,10240,0.029934932788213093
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,10240,0.048674134413401286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,10240,0.014779733618100485
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,8192,0.02658560077349345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,8192,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,7168,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,7168,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,8192,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,7168,0.011639466881752015
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,6144,0.023995733261108397
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,6144,0.013482667009035745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,6144,0.010444800059000652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,5120,0.022459733486175536
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,5120,0.00955733358860016
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,5120,0.01634666621685028
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,4096,0.013277866442998252
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,4096,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,4096,0.008362666765848795
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,3584,0.012184533476829528
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,3584,0.020275199413299562
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,3584,0.007748266557852428
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,3072,0.01935360034306844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,3072,0.011025066177050274
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,3072,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,2560,0.018978132804234823
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,2560,0.010752000411351522
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,2560,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,2048,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,2048,0.017746132612228394
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,2048,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,1536,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,1536,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,1536,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,1024,0.016452266772588094
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,1024,0.0050805335243542995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,1024,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,768,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,768,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,768,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,512,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,512,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,256,0.014062933127085366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,512,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,256,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,256,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,128,0.013206400473912559
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,128,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,128,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,64,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,64,0.0027594665686289472
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,3072,32,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,32,0.004057600100835165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,3072,64,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,3072,32,0.0026591998835404714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,65536,0.291430409749349
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,65536,0.4659200032552083
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,16384,0.037887998421986896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,65536,0.27054080963134763
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,16384,0.021128533283869426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,16384,0.08563840389251709
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,12288,0.050756267706553136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,12288,0.03232426643371582
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,12288,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,10240,0.029525333642959596
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,10240,0.014707199732462563
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,10240,0.019729065895080566
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,8192,0.02658560077349345
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,8192,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,8192,0.012390399972597759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,7168,0.025156267484029132
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,7168,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,7168,0.011673600474993388
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,6144,0.023552000522613525
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,6144,0.010478933652242024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,6144,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,5120,0.022050132354100548
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,5120,0.014404267072677612
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,4096,0.021026132504145305
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,5120,0.009659733374913533
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,4096,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,4096,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,3584,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,3584,0.011261866490046183
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,3584,0.007645866771539052
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,3072,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,3072,0.009898666540781658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,2560,0.018875734011332194
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,3072,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,2560,0.010035199920336406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,2560,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,2048,0.0180565337340037
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,2048,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,1536,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,2048,0.00610453337430954
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,1536,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,1024,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,1536,0.0055978665749231975
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,1024,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,1024,0.005154133339722951
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,768,0.016315733393033348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,512,0.015496533115704855
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,768,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,768,0.005256533126036326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,512,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,256,0.014165332913398743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,512,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,256,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,256,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,128,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,128,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,128,0.0038912000755469004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,64,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,64,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2560,32,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,64,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2560,32,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2560,32,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,65536,0.23958187103271483
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,65536,0.21896533966064452
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,65536,0.3716437339782715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,16384,0.03775146802266439
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,12288,0.03211946686108907
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,16384,0.020957867304484047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,16384,0.05717333157857259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,12288,0.019217065970102944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,12288,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,10240,0.029320534070332843
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,10240,0.01686186591784159
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,10240,0.014711466431617738
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,8192,0.026316799720128375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,8192,0.01378986636797587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,8192,0.012322133779525757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,7168,0.024985599517822265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,7168,0.013107200463612875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,7168,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,6144,0.01546239952246348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,6144,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,5120,0.022323199113210044
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,6144,0.010376532872517902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,5120,0.009384533762931824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,5120,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,4096,0.02106026609738668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,4096,0.01703146696090698
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,4096,0.008260266482830047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,3584,0.020206934213638304
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,3584,0.010205866893132527
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,3584,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,3072,0.019214934110641478
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,3072,0.0076117331782976794
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,3072,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,2560,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,2560,0.009147733449935913
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,2560,0.006587733328342438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,2048,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,2048,0.005973333120346069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,2048,0.007236266632874806
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,1536,0.017578667402267455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,1536,0.006549333532651265
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,1536,0.0054954667886098225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,1024,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,1024,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,768,0.016179200013478598
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,1024,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,768,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,768,0.00481279989083608
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,512,0.015735466281572977
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,512,0.005118933320045471
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,512,0.004469333092371622
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,256,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,256,0.014165332913398743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,256,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,128,0.013141333063443502
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,128,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,128,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,64,0.003852800031503042
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,64,0.0130730668703715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,64,0.002833066632350286
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,2048,32,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,2048,32,0.002423466742038727
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,2048,32,0.003925333420435587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,65536,0.11352746486663819
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,65536,0.28917760848999025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,65536,0.07799146970113119
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,16384,0.024029866854349772
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,16384,0.020855466524759926
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,16384,0.03761279980341593
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,12288,0.03174399932225545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,12288,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,12288,0.017066667477289833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,10240,0.02945706645647685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,10240,0.01802240014076233
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,10240,0.014571733276049294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,8192,0.026316799720128375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,8192,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,8192,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,7168,0.024951465924580894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,7168,0.009898666540781658
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,7168,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,6144,0.023347200949986775
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,6144,0.008869333068529765
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,6144,0.010376532872517902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,5120,0.021947733561197915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,5120,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,5120,0.01153706709543864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,4096,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,4096,0.01016960044701894
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,4096,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,3584,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,3584,0.009318400422732036
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,3584,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,3072,0.019319466749827065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,3072,0.007266133526961009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,3072,0.00846506655216217
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,2560,0.019012266397476198
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,2560,0.008294400076071422
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,2560,0.006690133114655812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,2048,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,2048,0.0065536002318064375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,2048,0.006075733403364817
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,1536,0.017373865842819212
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,1536,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,1536,0.006178133189678192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,1024,0.016008533040682473
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,1024,0.003549866626660029
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,1024,0.004983466863632202
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,768,0.0159061332543691
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,768,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,768,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,512,0.015701333681742348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,512,0.004331733286380768
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,512,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,256,0.014163200060526529
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,256,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,128,0.013174399733543396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,256,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,128,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,128,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,64,0.012731732924779257
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,64,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,64,0.0038890667259693147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1536,32,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1536,32,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1536,32,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,65536,0.10550613403320312
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,65536,0.07331840197245279
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,65536,0.1955498695373535
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,16384,0.037546666463216145
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,16384,0.01693013310432434
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,16384,0.020753065745035805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,12288,0.032153600454330446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,12288,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,12288,0.01672533353169759
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,10240,0.0289792001247406
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,10240,0.013448533415794373
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,8192,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,10240,0.01454080045223236
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,8192,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,8192,0.012388267119725545
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,7168,0.024644267559051514
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,7168,0.00890880028406779
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,7168,0.011434666315714518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,6144,0.023517866929372154
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,6144,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,5120,0.02222080032030741
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,6144,0.01050986647605896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,4096,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,5120,0.010513066252072652
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,5120,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,4096,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,4096,0.008226133386294047
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,3584,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,3584,0.00897706647713979
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,3584,0.0075434664885203045
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,3072,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,3072,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,3072,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,2560,0.019182932376861573
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,2560,0.007816533247629803
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,2560,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,2048,0.017812265952428182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,2048,0.006212266782919565
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,2048,0.005899733304977417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,1536,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,1536,0.017169066270192466
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,1536,0.005393066505591074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,1024,0.016145066420237223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,1024,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,1024,0.003305600086847941
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,768,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,768,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,768,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,512,0.015633066495259605
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,512,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,512,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,256,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,256,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,128,0.013414399822553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,128,0.0025941332181294756
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,64,0.012622933586438498
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,64,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,64,0.003956266740957896
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,1024,32,0.002491733431816101
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,1024,32,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,1024,32,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,65536,0.10523306528727214
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,65536,0.10120533307393391
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,65536,0.07321600119272867
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,16384,0.03757973512013753
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,16384,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,16384,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,12288,0.031846400101979574
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,12288,0.012117333213488261
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,12288,0.016759467124938966
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,10240,0.028842665751775104
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,10240,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,8192,0.026077866554260254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,10240,0.01474346617857615
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,8192,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,8192,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,7168,0.02542933424313863
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,7168,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,7168,0.011331199606259664
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,6144,0.023415466149648033
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,6144,0.01027413308620453
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,5120,0.02198186715443929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,6144,0.008157866696516674
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,5120,0.00993280013402303
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,5120,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,4096,0.02058239976565043
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,4096,0.009181867043177288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,3584,0.019933867454528808
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,4096,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,3584,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,3072,0.019146666924158732
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,3584,0.0075093333919843035
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,3072,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,3072,0.007099733253320058
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,2048,0.017783466974894205
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,2560,0.006586666901906331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,2560,0.0184661328792572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,2560,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,2048,0.005666133264700572
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,2048,0.003647999962170919
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,1536,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,1536,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,1024,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,1536,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,1024,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,1024,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,768,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,768,0.0033791999022165934
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,768,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,512,0.0032373333970705668
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,512,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,256,0.013892267147699991
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,256,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,256,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,128,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,128,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,64,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,64,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,768,32,0.012800000111262002
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,768,32,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,768,32,0.0039594667653242745
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,65536,0.10533226331075032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,65536,0.048059733708699544
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,65536,0.07314773400624594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,16384,0.037478399276733396
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,16384,0.009523199995358785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,16384,0.02044586737950643
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,12288,0.031778132915496825
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,12288,0.008430932958920796
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,10240,0.028910932938257854
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,10240,0.008806399504343669
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,12288,0.01658560037612915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,10240,0.014677332838376364
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,8192,0.026316799720128375
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,8192,0.007406933108965556
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,8192,0.012049067020416259
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,7168,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,7168,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,7168,0.011468799908955891
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,6144,0.023483733336130776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,6144,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,6144,0.010410666465759277
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,5120,0.02177706758181254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,5120,0.008942932883898417
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,5120,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,4096,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,4096,0.008055466910203297
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,3584,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,4096,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,3584,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,3584,0.007577600081761678
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,3072,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,3072,0.004300799965858459
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,3072,0.0069930667678515124
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,2560,0.018807466824849448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,2560,0.006621866424878438
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,2560,0.0040554667512575785
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,2048,0.017407999436060587
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,2048,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,2048,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,1536,0.017681066195170084
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,1536,0.0034080001215140024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,1536,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,1024,0.01614293356736501
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,1024,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,1024,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,768,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,768,0.016280532876650492
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,768,0.0046079998215039575
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,512,0.015291733543078103
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,512,0.003310933212439219
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,512,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,256,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,256,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,128,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,256,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,128,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,64,0.012868266304334006
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,64,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,512,32,0.012526933352152506
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,512,32,0.0026261332134405774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,65536,0.10571093559265136
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,65536,0.013380266229311624
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,512,32,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,65536,0.07314773400624594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,16384,0.03741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,16384,0.00631466656923294
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,12288,0.00628053347269694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,12288,0.031948800881703696
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,16384,0.020411733786265054
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,12288,0.01648640036582947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,10240,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,10240,0.02887679934501648
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,10240,0.014574933052062988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,7168,0.02539520064989726
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,8192,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,8192,0.02604373296101888
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,8192,0.012014933427174886
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,7168,0.009454933802286784
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,7168,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,6144,0.023244800170262654
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,6144,0.009489066402117411
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,6144,0.010308266679445902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,4096,0.02034133275349935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,5120,0.009215999643007915
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,5120,0.02211839954058329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,5120,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,4096,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,4096,0.00784746656815211
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,3584,0.020036266247431437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,3584,0.00730453332265218
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,3584,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,3072,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,3072,0.00365226666132609
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,3072,0.007133866846561432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,2560,0.01911466717720032
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,2560,0.0034773332377274835
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,2560,0.006758399804433187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,2048,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,2048,0.0030720000465710956
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,2048,0.005939200023810069
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,1536,0.016964266697565712
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,1536,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,1536,0.005358933409055074
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,1024,0.015974400440851848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,1024,0.0026591998835404714
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,1024,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,768,0.016588800152142844
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,768,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,768,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,512,0.002935466667016347
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,512,0.015598932902018229
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,256,0.013960533340771995
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,512,0.004232533276081085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,256,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,256,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,128,0.013243732849756875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,128,0.004095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,128,0.0030378667016824085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,64,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,64,0.002218666672706604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,64,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,256,32,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,256,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,256,32,0.0038506666819254553
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,65536,0.10543786684672038
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,65536,0.01225386659304301
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,65536,0.07304533322652182
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,16384,0.03857066631317139
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,16384,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,12288,0.031914667288462324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,16384,0.020514132579167683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,12288,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,12288,0.005427200098832448
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,10240,0.028808534145355225
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,10240,0.005051733553409576
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,10240,0.014267733693122864
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,8192,0.026760532458623247
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,8192,0.012185600399971009
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,7168,0.02519039909044902
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,8192,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,7168,0.007167999943097432
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,7168,0.011161599556605022
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,6144,0.023483733336130776
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,6144,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,6144,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,5120,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,5120,0.022016000747680665
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,4096,0.020343466599782308
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,5120,0.009250133236249288
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,4096,0.0035157332817713416
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,4096,0.007918933530648549
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,3584,0.019899733861287437
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,3584,0.0037205333511034647
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,3584,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,3072,0.019080533583958944
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,3072,0.003345066557327906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,3072,0.0069909334182739254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,2560,0.018397865692774455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,2560,0.0032768001159032187
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,2560,0.006485333542029063
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,2048,0.018227199713389076
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,2048,0.00290133332212766
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,2048,0.00580266664425532
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,1536,0.017032533884048462
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,1536,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,1536,0.0030037333567937215
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,1024,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,1024,0.002457600086927414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,1024,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,768,0.016042666633923848
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,768,0.004437333345413208
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,768,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,512,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,256,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,512,0.0026933332284291584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,256,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,256,0.004164266586303711
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,128,0.01323946714401245
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,128,0.0023210667073726653
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,128,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,64,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,64,0.003822933385769526
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,64,0.0021503999829292296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,128,32,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,128,32,0.0026602665583292644
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,128,32,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,65536,0.10513066450754802
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,65536,0.008021333316961924
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,65536,0.07283946673075357
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,16384,0.037307735284169516
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,16384,0.00562666654586792
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,16384,0.02054826617240906
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,12288,0.031914667288462324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,12288,0.005017599960168203
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,12288,0.01655466655890147
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,10240,0.028835199276606244
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,10240,0.004881066580613455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,10240,0.014301866292953491
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,8192,0.025975465774536133
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,8192,0.00576853354771932
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,7168,0.024780799945195518
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,8192,0.012151466806729634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,7168,0.005700266857941946
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,7168,0.011127466956774395
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,6144,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,6144,0.0056320001681645715
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,5120,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,5120,0.02211839954058329
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,5120,0.009313066800435383
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,6144,0.01030293305714925
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,4096,0.020616533358891805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,4096,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,4096,0.007884799937407176
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,3584,0.0034474665919939675
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,3584,0.01996799906094869
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,3584,0.00747519979874293
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,3072,0.01904639999071757
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,3072,0.0032085334261258446
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,3072,0.006997333467006683
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,2560,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,2560,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,2560,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,2048,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,2048,0.006007466713587443
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,2048,0.017612799008687337
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,1536,0.01713493267695109
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,1536,0.005222400029500326
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,1536,0.0027647999425729113
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,1024,0.00238933339715004
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,1024,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,768,0.016110933820406594
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,1024,0.00477866679430008
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,768,0.002696533252795537
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,768,0.004676266511281332
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,512,0.015121066570281982
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,512,0.0027306665976842242
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,512,0.004266666869322458
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,256,0.014199466506640116
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,256,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,256,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,128,0.013175466656684875
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,128,0.0021503999829292296
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,64,0.002147199958562851
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,128,0.003788800040880839
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,64,0.012936533490816752
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,64,32,0.012663466731707254
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,64,0.003993600110212962
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,64,32,0.002525866776704788
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,64,32,0.0037834666669368743
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,65536,0.005559466779232025
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,65536,0.07321279843648275
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,65536,0.104857603708903
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,16384,0.03741013209025065
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,16384,0.004334933559099833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,12288,0.031914667288462324
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,16384,0.020753065745035805
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,12288,0.004126933217048645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,12288,0.01628159979979197
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,10240,0.02877440055211385
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,10240,0.01443839967250824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,10240,0.0040277334551016486
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,8192,0.02621120015780131
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,8192,0.004744533201058706
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,8192,0.012288000186284382
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,7168,0.02474666635195414
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,7168,0.0045056000351905824
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,7168,0.010990933577219645
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,6144,0.023381332556406655
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,6144,0.010615467031796774
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,6144,0.004642133414745331
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,5120,0.02208426594734192
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,5120,0.004130133241415024
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,4096,0.020309333006540933
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,5120,0.009045333663622538
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,4096,0.0034133332471052804
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,4096,0.00795306662718455
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,3584,0.020002132654190062
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,3584,0.007441066702206929
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,3584,0.0034111998975276947
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,3072,0.01914880077044169
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,3072,0.0032426667710145317
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,3072,0.006894933183987935
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,2560,0.01843199928601583
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,2560,0.006417066852251689
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,2560,0.00314026673634847
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,2048,0.01781760056813558
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,2048,0.005836800237496694
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,2048,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,1536,0.017026132345199584
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,1536,0.00266239990790685
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,1536,0.005253333350022634
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,1024,0.016076800227165223
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,1024,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,1024,0.004846933484077454
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,768,0.0044714664419492085
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,768,0.01621333360671997
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,768,0.0026943999032179515
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,512,0.01539413332939148
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,512,0.0025600001215934755
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,512,0.004369066655635833
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,256,0.0027989332874615988
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,256,0.013994666934013366
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,256,0.004054400076468786
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,128,0.01313920021057129
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,128,0.0021162666380405426
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,128,0.003857066730658213
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,64,0.012725333372751871
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,64,0.002218666672706604
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,64,0.0040618665516376495
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,int8_wo,1,32,32,0.012834133704503379
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,float16,1,32,32,0.0022528000175952912
SGLang,0.5.5.post3,NVIDIA L40S,gemm,sglang,fp8,1,32,32,0.003857066730658213
