framework,version,device,op_name,kernel_source,bmm_dtype,num_tokens,num_heads,latency
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1,128,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1,32,0.006649599969387054
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1,64,0.01096000000834465
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1,16,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1,8,0.01085439994931221
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1,4,0.013308799266815186
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1,2,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1,1,0.014662399888038635
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1,128,0.011052799969911575
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1,2,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1,8,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1,1,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1,64,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1,32,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1,16,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2,64,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2,128,0.009216000139713288
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2,32,0.006672000139951706
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2,16,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2,128,0.010035199671983719
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2,8,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2,64,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2,4,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2,32,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2,4,0.006864000111818314
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2,16,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2,8,0.007372800260782242
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2,2,0.006966400146484375
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2,2,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2,1,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2,1,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4,128,0.010444799810647965
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4,64,0.006867200136184692
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4,128,0.009404800087213516
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4,32,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4,64,0.008499199897050858
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4,32,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4,16,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4,8,0.007577600330114365
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4,8,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4,16,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4,4,0.007372800260782242
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4,4,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4,2,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4,2,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4,1,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4,1,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8,128,0.009318400174379349
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8,128,0.011264000087976456
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8,64,0.008905600011348724
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8,32,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8,32,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8,16,0.0065600000321865085
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8,16,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8,64,0.014233599603176116
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8,8,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8,8,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8,4,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8,2,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8,2,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8,1,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,16,128,0.009625600278377533
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,16,64,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,16,32,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8,1,0.007161600142717361
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,16,128,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,16,16,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,16,16,0.007577600330114365
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,16,64,0.009625600278377533
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,16,32,0.007884799689054488
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,16,8,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,16,8,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,16,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,16,2,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,16,2,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,16,1,0.007062400132417679
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,16,4,0.007379200309514999
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,16,1,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,32,128,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,32,128,0.01669120043516159
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,32,64,0.00697920024394989
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,32,64,0.011059200018644333
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,32,32,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,32,32,0.008499199897050858
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,32,8,0.008294399827718735
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,32,16,0.007158400118350982
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,32,8,0.006553599983453751
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,32,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,32,4,0.0068512000143527985
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,32,2,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,32,16,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,32,2,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,32,1,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,32,1,0.006566400080919266
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,48,128,0.018329599499702455
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,48,128,0.021401600539684297
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,48,64,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,48,64,0.013209599256515502
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,48,16,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,48,32,0.009523200243711472
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,48,8,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,48,16,0.007782399654388428
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,48,32,0.0072512000799179075
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,48,8,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,48,4,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,48,4,0.006646399945020675
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,48,2,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,48,2,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,48,1,0.006553599983453751
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,48,1,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,64,128,0.025600001215934753
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,64,64,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,64,64,0.014643199741840363
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,64,128,0.020377600193023683
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,64,32,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,64,32,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,64,16,0.007884799689054488
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,64,16,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,64,8,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,64,8,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,64,4,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,64,2,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,64,1,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,64,2,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,80,128,0.025308799743652344
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,64,1,0.006646399945020675
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,64,4,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,80,64,0.014630399644374847
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,80,128,0.028672000765800475
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,80,64,0.01822720021009445
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,80,32,0.008499199897050858
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,80,32,0.012390399724245072
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,80,16,0.006867200136184692
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,80,16,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,80,8,0.007577600330114365
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,80,4,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,80,4,0.0070720002055168155
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,80,8,0.007184000313282013
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,80,2,0.007782399654388428
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,80,2,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,80,1,0.006956800073385239
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,80,1,0.007782399654388428
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,96,128,0.025804799795150758
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,96,128,0.031539198756217954
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,96,64,0.015360000729560851
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,96,32,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,96,16,0.006764800101518631
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,96,32,0.013209599256515502
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,96,8,0.006639999896287918
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,96,8,0.007680000364780426
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,96,16,0.009318400174379349
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,96,64,0.019542400538921357
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,96,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,96,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,96,2,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,96,2,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,96,1,0.007680000364780426
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,96,1,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,128,128,0.03358719944953918
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,128,64,0.019356800615787505
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,128,64,0.02222079932689667
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,128,32,0.010342399775981902
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,128,128,0.037376001477241516
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,128,32,0.014745600521564484
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,128,16,0.006470400094985962
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,128,16,0.010035199671983719
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,128,8,0.006553599983453751
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,128,8,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,128,4,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,128,4,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,128,2,0.007062400132417679
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,128,1,0.007798399776220322
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,128,1,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,160,128,0.04095999896526337
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,128,2,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,160,128,0.04669440090656281
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,160,64,0.02232320010662079
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,160,64,0.026624000072479247
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,160,32,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,160,32,0.01679359972476959
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,160,16,0.007577600330114365
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,160,16,0.011264000087976456
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,160,8,0.008803199976682663
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,160,4,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,160,4,0.007561600208282471
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,160,2,0.009318400174379349
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,160,8,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,160,1,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,160,2,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,160,1,0.006553599983453751
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,192,128,0.028672000765800475
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,192,128,0.05242879986763001
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,192,64,0.017715199291706084
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,192,64,0.02959359884262085
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,192,32,0.01822720021009445
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,192,16,0.008499199897050858
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,192,16,0.012083200365304947
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,192,8,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,192,32,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,192,4,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,192,8,0.00920960009098053
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,192,4,0.007782399654388428
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,192,2,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,192,2,0.008089599758386612
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,192,1,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,192,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,256,128,0.06932479739189149
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,256,64,0.01863359957933426
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,256,32,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,256,32,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,256,64,0.037376001477241516
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,256,128,0.03430399894714355
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,256,16,0.010444799810647965
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,256,16,0.013823999464511872
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,256,8,0.006857600063085556
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,256,8,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,256,4,0.006553599983453751
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,256,4,0.008089599758386612
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,256,1,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,256,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,320,128,0.04526079893112182
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,320,128,0.09297919869422913
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,256,2,0.007372800260782242
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,320,64,0.024278399348258973
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,320,64,0.0449535995721817
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,320,32,0.014336000382900237
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,320,32,0.026111999154090883
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,320,16,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,256,2,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,320,16,0.01679359972476959
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,320,8,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,320,4,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,320,4,0.008703999966382981
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,320,8,0.007561600208282471
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,320,2,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,320,2,0.007577600330114365
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,320,1,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,384,128,0.06707839965820313
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,384,128,0.10536960363388062
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,384,64,0.02539519965648651
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,384,32,0.014745600521564484
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,384,32,0.02908160090446472
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,384,16,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,384,64,0.05232639908790589
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,384,16,0.018636800348758698
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,384,8,0.008499199897050858
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,320,1,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,384,8,0.011980800330638886
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,384,4,0.007475200295448303
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,384,4,0.0095040000975132
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,384,2,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,384,1,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,384,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,384,2,0.007884799689054488
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,512,128,0.1505280017852783
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,512,64,0.03266560137271881
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,512,64,0.06778879761695862
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,512,32,0.03706879913806915
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,512,32,0.018739199638366698
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,512,16,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,512,8,0.013823999464511872
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,512,8,0.010239999741315842
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,512,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,512,128,0.09595199823379516
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,512,4,0.010342399775981902
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,512,16,0.022118400037288665
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,512,2,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,512,2,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,512,1,0.006870400160551071
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,768,128,0.15902719497680665
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,512,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,768,64,0.04894720017910004
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,768,32,0.02539519965648651
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,768,32,0.05232639908790589
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,768,16,0.014745600521564484
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,768,16,0.029388800263404846
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,768,8,0.010035199671983719
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,768,64,0.10833920240402221
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,768,128,0.27013120651245115
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,768,8,0.018723200261592864
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,768,4,0.008499199897050858
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,768,2,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,768,2,0.009523200243711472
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,768,4,0.011980800330638886
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,768,1,0.006553599983453751
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1024,128,0.37345280647277834
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,768,1,0.007680000364780426
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1024,64,0.06932479739189149
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1024,128,0.2861056089401245
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1024,64,0.1491968035697937
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1024,32,0.031948798894882204
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1024,16,0.01852159947156906
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1024,32,0.06768640279769897
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1024,16,0.037171199917793274
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1024,8,0.022015999257564544
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1024,4,0.010441599786281586
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1024,2,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1024,4,0.013823999464511872
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1024,8,0.010649599879980088
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1024,2,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1024,1,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1024,1,0.008284799754619598
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1536,128,0.4348927974700928
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1536,128,0.5716991901397706
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1536,64,0.17233920097351074
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1536,64,0.26460158824920654
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1536,32,0.10854400396347046
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1536,16,0.06318079829216003
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1536,32,0.0494592010974884
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1536,16,0.02529279887676239
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1536,8,0.014847999811172486
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1536,4,0.010239999741315842
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1536,4,0.01884160041809082
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1536,8,0.02959359884262085
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1536,2,0.008483199775218964
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1536,2,0.012083200365304947
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2048,128,0.5708799839019776
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,1536,1,0.009113600105047226
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2048,128,0.7599103927612305
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2048,64,0.2687999963760376
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,1536,1,0.007254400104284286
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2048,32,0.08509439826011658
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2048,64,0.36392960548400877
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2048,32,0.14510079622268676
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2048,16,0.03256320059299469
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2048,16,0.06778879761695862
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2048,8,0.01903039962053299
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2048,8,0.037376001477241516
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2048,4,0.022015999257564544
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2048,2,0.010444799810647965
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2048,2,0.013820800185203552
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2048,4,0.010652799904346467
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,2048,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,3072,128,0.8514559745788575
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,3072,128,1.163980770111084
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,2048,1,0.010025600343942643
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,3072,64,0.5647359848022461
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,3072,64,0.41891841888427733
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,3072,32,0.17236160039901732
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,3072,16,0.0477183997631073
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,3072,16,0.10926079750061035
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,3072,8,0.025804799795150758
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,3072,4,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,3072,4,0.029491201043128967
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,3072,32,0.25815041065216066
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,3072,8,0.05242879986763001
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,3072,2,0.010035199671983719
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,3072,2,0.018739199638366698
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,3072,1,0.006943999975919724
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,3072,1,0.011878400295972823
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4096,128,1.1408384323120118
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4096,128,1.5622143745422363
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4096,64,0.7518208026885986
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4096,64,0.5606400012969971
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4096,32,0.25722880363464357
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4096,32,0.36188158988952634
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4096,16,0.0737280011177063
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4096,16,0.14510079622268676
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4096,8,0.032358399033546446
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4096,8,0.06809599995613098
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4096,4,0.018739199638366698
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4096,2,0.01064639985561371
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4096,2,0.022015999257564544
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,4096,1,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4096,1,0.013823999464511872
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,4096,4,0.037376001477241516
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,6144,128,1.7101823806762695
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,6144,128,2.3427072525024415
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,6144,64,1.144422435760498
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,6144,32,0.3983360052108765
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,6144,32,0.5566463947296143
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,6144,64,0.8172543525695801
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,6144,16,0.15892479419708253
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,6144,8,0.04904960095882416
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,6144,8,0.10240000486373901
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,6144,16,0.2582528114318848
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,6144,4,0.026214399933815004
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,6144,4,0.052019202709198
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,6144,2,0.017407999932765962
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,6144,1,0.010035199671983719
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,6144,2,0.030924800038337707
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,6144,1,0.01822720021009445
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8192,128,2.287215995788574
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8192,64,1.1149312019348145
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8192,64,1.5377408027648927
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8192,32,0.5450751781463623
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8192,32,0.7440383911132813
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8192,16,0.3559423923492432
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8192,16,0.2324415922164917
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8192,128,3.174604797363281
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8192,8,0.066457599401474
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8192,8,0.14305280447006224
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8192,4,0.05140479803085327
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8192,4,0.06809599995613098
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8192,2,0.01884160041809082
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8192,2,0.03665919899940491
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,float16,8192,1,0.010444799810647965
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_pre,default,fp8,8192,1,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1,128,0.008396799862384795
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1,64,0.007587199658155441
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1,64,0.013209599256515502
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1,16,0.006966400146484375
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1,32,0.006966400146484375
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1,4,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1,2,0.007379200309514999
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1,8,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1,4,0.007283200323581695
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1,2,0.013209599256515502
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1,128,0.009312000125646591
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1,1,0.012809599936008453
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2,128,0.008703999966382981
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1,32,0.011366400122642516
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1,8,0.010547199845314026
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1,16,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2,128,0.010649599879980088
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2,32,0.013107199966907502
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1,1,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2,64,0.01085439994931221
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2,32,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2,16,0.01372160017490387
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2,8,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2,16,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2,64,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2,4,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2,4,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2,2,0.007056000083684922
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2,2,0.006761600077152252
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2,1,0.006553599983453751
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4,128,0.008892799913883209
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4,128,0.009206400066614152
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2,8,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4,64,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4,64,0.007475200295448303
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4,32,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4,16,0.006966400146484375
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4,16,0.021196800470352172
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4,32,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4,8,0.006668800115585327
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4,8,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4,4,0.007372800260782242
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4,2,0.006972800195217133
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4,2,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4,1,0.006454399973154068
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4,1,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8,128,0.008806400001049042
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8,64,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8,128,0.012185599654912949
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8,32,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8,32,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8,64,0.007987199723720551
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8,16,0.006348799914121628
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8,16,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8,8,0.006652799993753433
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8,8,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8,4,0.0069472000002861025
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8,4,0.0071552000939846035
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8,2,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8,2,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8,1,0.0064640000462532045
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16,128,0.009094399958848953
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16,64,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16,32,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16,8,0.006675200164318084
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16,16,0.006665600091218948
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16,128,0.012185599654912949
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16,8,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16,64,0.008793599903583527
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16,32,0.008089599758386612
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16,16,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16,4,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16,2,0.006854400038719177
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16,2,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16,1,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16,1,0.007046400010585785
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,32,128,0.009833600372076035
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,32,128,0.014947199821472168
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,32,64,0.007772800326347351
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,32,64,0.010444799810647965
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,32,32,0.006646399945020675
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,32,16,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,32,32,0.008908800035715102
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,32,16,0.008073599636554718
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,32,4,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,32,8,0.007158400118350982
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,32,8,0.007999999821186066
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,32,4,0.0067552000284194945
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,32,2,0.006553599983453751
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,32,1,0.0071712002158164975
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,32,1,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,32,2,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,48,128,0.015872000157833098
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,48,64,0.009836799651384353
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,48,128,0.017715199291706084
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,48,64,0.011878400295972823
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,48,32,0.009625600278377533
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,48,16,0.008287999778985977
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,48,8,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,48,16,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,48,32,0.007372800260782242
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,48,8,0.009318400174379349
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,48,4,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,48,4,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,48,2,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,48,2,0.006777600198984146
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,48,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,48,1,0.008396799862384795
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,64,64,0.00981760025024414
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,64,128,0.020787200331687926
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,64,128,0.016998399794101716
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,64,64,0.013414399325847625
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,64,32,0.007673600316047668
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,64,32,0.010239999741315842
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,64,16,0.007161600142717361
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,64,8,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,64,16,0.008806400001049042
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,64,8,0.007577600330114365
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,64,4,0.006553599983453751
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,64,4,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,64,2,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,64,2,0.006943999975919724
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,64,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,80,128,0.025600001215934753
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,80,128,0.021913599967956544
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,64,1,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,80,64,0.029491201043128967
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,80,64,0.01730560064315796
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,80,32,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,80,32,0.011561600118875503
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,80,16,0.007372800260782242
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,80,8,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,80,16,0.009216000139713288
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,80,8,0.007782399654388428
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,80,2,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,80,4,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,80,4,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,80,2,0.006966400146484375
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,80,1,0.0070592001080513
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,80,1,0.009724800288677216
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,96,128,0.02222079932689667
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,96,128,0.02815999984741211
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,96,64,0.013209599256515502
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,96,64,0.018432000279426576
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,96,32,0.008499199897050858
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,96,16,0.007475200295448303
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,96,32,0.012083200365304947
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,96,8,0.0069760002195835115
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,96,16,0.00942080020904541
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,96,4,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,96,8,0.007987199723720551
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,96,4,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,96,2,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,96,1,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,96,2,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,96,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,128,128,0.033792001008987424
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,128,128,0.027852800488471986
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,128,64,0.016601599752902985
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,128,64,0.020582400262355804
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,128,32,0.013414399325847625
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,128,32,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,128,16,0.007670400291681289
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,128,16,0.010342399775981902
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,128,8,0.007056000083684922
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,128,8,0.008396799862384795
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,128,4,0.007577600330114365
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,128,4,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,128,2,0.006758400052785873
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,128,2,0.007884799689054488
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,160,128,0.03665919899940491
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,128,1,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,128,1,0.006534399837255478
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,160,128,0.04403199851512909
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,160,64,0.01945600062608719
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,160,64,0.024780799448490144
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,160,32,0.011264000087976456
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,160,32,0.015667200088500977
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,160,16,0.008396799862384795
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,160,16,0.011059200018644333
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,160,8,0.008806400001049042
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,160,8,0.007257600128650665
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,160,4,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,160,2,0.0071712002158164975
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,160,2,0.008905600011348724
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,160,4,0.007680000364780426
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,160,1,0.007062400132417679
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,160,1,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,192,128,0.04229120016098022
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,192,64,0.02263039946556091
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,192,32,0.012902399897575379
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,192,64,0.027647998929023743
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,192,128,0.05027840137481689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,192,32,0.01822720021009445
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,192,16,0.008499199897050858
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,192,16,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,192,8,0.007475200295448303
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,192,8,0.009318400174379349
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,192,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,192,4,0.007971200346946716
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,192,2,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,192,1,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,192,2,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,192,1,0.00809279978275299
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,256,128,0.05775359869003296
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,256,128,0.05939199924468994
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,256,64,0.027750399708747864
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,256,64,0.0336896002292633
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,256,32,0.020787200331687926
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,256,32,0.01658879965543747
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,256,16,0.009728000313043595
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,256,16,0.013516800105571746
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,256,8,0.007884799689054488
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,256,4,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,256,4,0.009011200070381165
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,256,8,0.010239999741315842
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,256,2,0.026025599241256712
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,256,2,0.007475200295448303
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,256,1,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,256,1,0.006553599983453751
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,320,128,0.07167999744415283
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,320,128,0.0766975998878479
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,320,64,0.036556801199913024
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,320,64,0.043007999658584595
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,320,32,0.019251200556755065
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,320,16,0.011264000087976456
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,320,16,0.015449599921703338
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,320,32,0.024166400730609893
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,320,8,0.008396799862384795
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,320,8,0.010956799983978272
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,320,2,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,320,4,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,320,4,0.026419198513031004
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,320,2,0.007987199723720551
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,384,128,0.09287679791450501
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,384,128,0.08458240032196045
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,384,64,0.04259839951992035
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,320,1,0.007065600156784058
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,320,1,0.0067552000284194945
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,384,64,0.050483202934265135
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,384,32,0.02293439954519272
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,384,32,0.027647998929023743
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,384,8,0.008499199897050858
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,384,16,0.0130048006772995
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,384,16,0.017097599804401398
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,384,4,0.009404800087213516
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,384,4,0.007286400347948074
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,384,2,0.007782399654388428
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,384,8,0.011878400295972823
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,384,1,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,384,1,0.007168000191450119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,384,2,0.008089599758386612
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,512,128,0.05222079753875732
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,512,128,0.235315203666687
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,512,64,0.0626688003540039
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,512,64,0.024166400730609893
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,512,32,0.027750399708747864
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,512,32,0.0336896002292633
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,512,16,0.01658879965543747
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,512,16,0.020883199572563172
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,512,8,0.013494400680065155
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,512,8,0.010035199671983719
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,512,4,0.007577600330114365
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,512,4,0.010239999741315842
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,512,2,0.006550399959087372
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,512,2,0.00838399976491928
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,512,1,0.006963200122117996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,512,1,0.007372800260782242
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,768,128,0.2327712059020996
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,768,128,0.431001615524292
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,768,64,0.034815999865531924
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,768,64,0.09103360176086425
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,768,32,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,768,32,0.048742398619651794
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,768,16,0.022118400037288665
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,768,16,0.02754560112953186
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,768,8,0.0130048006772995
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,768,8,0.017100800573825837
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,768,4,0.008601599931716919
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,768,4,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,768,2,0.009318400174379349
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,768,2,0.00727040022611618
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,768,1,0.006860800087451935
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,768,1,0.007987199723720551
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1024,128,0.3079071998596191
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1024,64,0.04567039906978607
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1024,128,0.6014976024627685
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1024,64,0.2527231931686401
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1024,32,0.024166400730609893
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1024,16,0.027750399708747864
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1024,32,0.05928959846496582
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1024,8,0.01658879965543747
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1024,16,0.033792001008987424
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1024,8,0.020684799551963805
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1024,4,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1024,2,0.007782399654388428
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1024,4,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1024,2,0.010239999741315842
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1024,1,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1536,128,0.4434944152832031
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1024,1,0.006451199948787689
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1536,128,0.9293824195861816
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1536,64,0.21524479389190673
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1536,64,0.42690558433532716
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1536,32,0.034406399726867674
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1536,32,0.09082880020141601
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1536,16,0.022835199534893037
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1536,16,0.048742398619651794
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1536,8,0.02232320010662079
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1536,4,0.016998399794101716
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1536,4,0.013203200697898865
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1536,8,0.027955201268196107
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1536,2,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1536,2,0.008499199897050858
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,1536,1,0.006943999975919724
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,1536,1,0.009113600105047226
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2048,128,1.246617603302002
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2048,64,0.30504961013793946
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2048,128,0.5808159828186035
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2048,64,0.5911551952362061
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2048,32,0.04627839922904968
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2048,32,0.24360959529876708
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2048,16,0.024780799448490144
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2048,16,0.062463998794555664
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2048,8,0.027750399708747864
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2048,8,0.0336896002292633
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2048,4,0.020684799551963805
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2048,4,0.01679359972476959
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2048,2,0.009830400347709656
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,2048,1,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2048,2,0.013312000036239623
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,2048,1,0.010137599706649781
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,3072,128,0.8519680023193359
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,3072,64,0.4193280220031738
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,3072,128,1.8749439239501953
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,3072,32,0.21001598834991456
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,3072,64,0.9255935668945312
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,3072,32,0.42588157653808595
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,3072,16,0.034492799639701845
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,3072,16,0.09144319891929627
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,3072,8,0.022732800245285033
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,3072,4,0.022118400037288665
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,3072,8,0.05058559775352478
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,3072,4,0.027750399708747864
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,3072,2,0.01719360053539276
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,3072,2,0.012992000579833985
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,3072,1,0.006656000018119812
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,3072,1,0.011776000261306763
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4096,128,1.1967488288879395
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4096,64,0.5719039916992188
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4096,64,1.2395520210266113
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4096,32,0.2804543972015381
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4096,128,2.5011199951171874
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4096,16,0.045372799038887024
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4096,32,0.5893119812011719
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4096,16,0.25077760219573975
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4096,8,0.024883200228214265
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4096,8,0.06318079829216003
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4096,4,0.027750399708747864
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4096,4,0.03399679958820343
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4096,2,0.01669120043516159
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4096,1,0.013513599336147309
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,4096,2,0.020684799551963805
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,4096,1,0.007046400010585785
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,6144,128,1.799577522277832
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,6144,64,0.8216575622558594
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,6144,32,0.41492481231689454
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,6144,64,1.8885631561279297
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,6144,128,3.794841766357422
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,6144,32,0.9200639724731445
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,6144,16,0.21032960414886476
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,6144,8,0.03532800078392029
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,6144,16,0.4221951961517334
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,6144,8,0.09287679791450501
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,6144,4,0.023449599742889404
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,6144,4,0.04782080054283142
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,6144,2,0.012390399724245072
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,6144,1,0.017407999932765962
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,6144,2,0.03061760067939758
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,6144,1,0.008191999793052674
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8192,128,2.2752256393432617
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8192,64,1.0901503562927246
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8192,32,0.5510144233703613
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8192,64,2.5276416778564452
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8192,128,5.049446487426758
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8192,32,1.2433407783508301
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8192,16,0.27975680828094485
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8192,16,0.5863423824310303
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8192,8,0.040345600247383116
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8192,8,0.26920959949493406
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8192,4,0.024883200228214265
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8192,4,0.06041600108146668
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8192,2,0.013823999464511872
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8192,2,0.03614720106124878
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,8192,1,0.009523200243711472
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,8192,1,0.021401600539684297
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,12288,32,0.8168448448181153
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,12288,64,1.6833536148071289
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,12288,16,0.41451520919799806
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,12288,32,1.8882560729980469
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,12288,64,3.7896190643310548
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,12288,128,3.6197376251220703
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,12288,16,0.921395206451416
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,12288,128,7.568281555175782
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,12288,8,0.2027519941329956
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,12288,4,0.03522560000419617
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,12288,4,0.09318400025367737
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,12288,8,0.44759039878845214
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,12288,2,0.022937600314617158
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,12288,2,0.04782080054283142
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,12288,1,0.012492799758911132
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,12288,1,0.02805759906768799
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16384,64,2.161356735229492
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16384,32,1.0933247566223145
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16384,128,4.771635055541992
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16384,64,5.047296142578125
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16384,16,1.2515328407287598
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16384,16,0.5502975940704345
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16384,32,2.527948760986328
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16384,128,10.049024200439453
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16384,8,0.28395519256591795
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16384,8,0.6240255832672119
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16384,4,0.04044800102710724
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16384,4,0.26705920696258545
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16384,2,0.02457599937915802
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16384,2,0.060313600301742556
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,16384,1,0.03430399894714355
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,16384,1,0.016383999586105348
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,20480,128,6.0065662384033205
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,20480,64,2.792140769958496
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,20480,32,1.3710335731506347
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,20480,16,0.6810624122619628
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,20480,32,3.159552001953125
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,20480,64,6.3056896209716795
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,20480,16,1.5566847801208497
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,20480,128,12.538982391357422
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,20480,8,0.3383296012878418
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,20480,8,0.8159232139587402
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,20480,4,0.08355839848518372
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,20480,4,0.3659775972366333
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,20480,2,0.03399679958820343
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,20480,2,0.08171520233154297
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,float16,20480,1,0.022937600314617158
SGLang,0.5.5.post3,NVIDIA L40S,mla_gen_post,default,fp8,20480,1,0.04341759979724884
