framework,version,device,op_name,kernel_source,mla_dtype,kv_cache_dtype,num_heads,batch_size,isl,tp_size,step,latency
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,1,1,0,0.10204800367355346
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,4,1,0,0.10333119630813599
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,8,1,0,0.10308480262756348
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,16,1,0,0.10326720476150512
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,32,1,0,0.11165119409561157
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,64,1,0,0.10990079641342163
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,128,1,0,0.10964800119400024
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,256,1,0,0.10975359678268433
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,512,1,0,0.10945919752120972
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,1024,1,0,0.1096735954284668
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,2048,1,0,0.11169600486755371
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,4096,1,0,0.11459200382232666
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,8192,1,0,0.11970239877700806
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1,16384,1,0,0.1297055959701538
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,1,1,0,0.10240319967269898
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,4,1,0,0.1022879958152771
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,8,1,0,0.10223040580749512
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,16,1,0,0.10306240320205688
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,32,1,0,0.1126911997795105
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,64,1,0,0.10977599620819092
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,128,1,0,0.10841920375823974
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,256,1,0,0.10882560014724732
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,512,1,0,0.1087488055229187
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,1024,1,0,0.11102080345153809
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,2048,1,0,0.11292799711227416
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,4096,1,0,0.11506240367889405
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,8192,1,0,0.12182079553604126
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,2,16384,1,0,0.12983039617538453
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,1,1,0,0.10249919891357422
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,4,1,0,0.10294400453567505
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,8,1,0,0.1025920033454895
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,16,1,0,0.10358719825744629
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,32,1,0,0.11361600160598755
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,64,1,0,0.11187520027160644
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,128,1,0,0.11038719415664673
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,256,1,0,0.11106239557266236
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,512,1,0,0.11220159530639648
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,1024,1,0,0.11518720388412476
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,2048,1,0,0.115721595287323
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,1,1,0,0.0723583996295929
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,4096,1,0,0.12463040351867676
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,4,1,0,0.07216320037841797
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,8,1,0,0.07256320118904114
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,8192,1,0,0.13181760311126708
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,16,1,0,0.07269759774208069
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,32,1,0,0.07240319848060608
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,4,16384,1,0,0.14718400239944457
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,64,1,0,0.0742143988609314
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,1,1,0,0.10349440574645996
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,128,1,0,0.07521600127220154
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,256,1,0,0.07964159846305847
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,4,1,0,0.10371840000152588
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,512,1,0,0.08058879971504211
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,8,1,0,0.10335040092468262
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,1,1,0,0.038806399703025816
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,1024,1,0,0.08203840255737305
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,16,1,0,0.10337280035018921
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,4,1,0,0.038873600959777835
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,2048,1,0,0.08421440124511718
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,8,1,0,0.03812159895896912
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,32,1,0,0.11578240394592285
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,4096,1,0,0.08608639836311341
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,16,1,0,0.03895359933376312
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,32,1,0,0.03900479972362518
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,8192,1,0,0.09172800183296204
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,64,1,0,0.11308799982070923
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,64,1,0,0.04044800102710724
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,128,1,0,0.11293439865112305
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1,16384,1,0,0.09757440090179444
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,128,1,0,0.04229759871959686
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,256,1,0,0.11472959518432617
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,256,1,0,0.042310398817062375
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,512,1,0,0.04405440092086792
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,512,1,0,0.11577600240707397
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,1024,1,0,0.04470719993114471
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,1024,1,0,0.1184000015258789
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,1,1,0,0.05574079751968384
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,2048,1,0,0.045638400316238406
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,2048,1,0,0.12568000555038453
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,4,1,0,0.056278401613235475
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,4096,1,0,0.04745280146598816
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,8,1,0,0.055871999263763426
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,8192,1,0,0.05265920162200928
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,4096,1,0,0.13325439691543578
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,16,1,0,0.05560320019721985
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1,16384,1,0,0.06356160044670105
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,32,1,0,0.05614079833030701
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,1,1,0,0.03447999954223633
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,64,1,0,0.05809599757194519
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,1,1,0,0.06902400255203248
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,8192,1,0,0.14838080406188964
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,1,1,0,0.1059615969657898
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,4,1,0,0.03452480137348175
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,128,1,0,0.05885760188102722
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,4,1,0,0.06916159987449647
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,8,1,0,0.034390398859977724
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,256,1,0,0.06256960034370422
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,4,1,0,0.10542080402374268
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,8,1,0,0.0689631998538971
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,512,1,0,0.06410560011863708
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,16,1,0,0.03446080088615418
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,8,1,0,0.10563199520111084
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,16,1,0,0.06986560225486756
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,8,16384,1,0,0.178656005859375
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,32,1,0,0.03724479973316193
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,1024,1,0,0.06507840156555175
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,16,1,0,0.10533440113067627
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,32,1,0,0.0782368004322052
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,1,1,0,0.07252159714698792
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,64,1,0,0.036399999260902406
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,2048,1,0,0.06593919992446899
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,1,1,0,0.10522880554199218
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,32,1,0,0.10534720420837403
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,64,1,0,0.07657279968261718
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,4,1,0,0.07255359888076782
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,128,1,0,0.036099201440811156
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,4096,1,0,0.06789759993553161
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,4,1,0,0.1053887963294983
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,128,1,0,0.07495999932289124
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,64,1,0,0.10724159479141235
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,256,1,0,0.03772160112857818
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,8,1,0,0.0725600004196167
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,8192,1,0,0.07268800139427185
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,256,1,0,0.07497280240058898
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,8,1,0,0.10533440113067627
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,512,1,0,0.0390751987695694
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,128,1,0,0.10882560014724732
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,16,1,0,0.07285760045051574
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1,16384,1,0,0.07715839743614197
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,512,1,0,0.07489920258522034
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,1024,1,0,0.04295360147953033
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,16,1,0,0.10548160076141358
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,32,1,0,0.07301440238952636
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,256,1,0,0.11260800361633301
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,1024,1,0,0.07983999848365783
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,2048,1,0,0.042777600884437564
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,64,1,0,0.07391039729118347
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,32,1,0,0.1180191993713379
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,512,1,0,0.11477119922637939
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,2048,1,0,0.07922880053520202
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,4096,1,0,0.04548160135746002
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,128,1,0,0.07625600099563598
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,64,1,0,0.11807039976119996
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,1,1,0,0.0526528000831604
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,1,1,0,0.03896960020065308
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,1024,1,0,0.11638720035552978
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,4096,1,0,0.08125119805335998
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,8192,1,0,0.05057600140571594
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,256,1,0,0.07985919713973999
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,128,1,0,0.12055360078811646
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,4,1,0,0.05292479991912842
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,4,1,0,0.039068800210952756
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,8192,1,0,0.08568639755249023
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,2048,1,0,0.11801600456237793
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1,16384,1,0,0.06236159801483154
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,512,1,0,0.08153600096702576
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,8,1,0,0.053065598011016846
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,8,1,0,0.03898879885673523
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,256,1,0,0.12186559438705444
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1,16384,1,0,0.09353600144386291
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,4096,1,0,0.12184959650039673
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,1024,1,0,0.08310400247573853
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,16,1,0,0.05302720069885254
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,16,1,0,0.039238399267196654
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,512,1,0,0.12127679586410522
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,8192,1,0,0.12702080011367797
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,32,1,0,0.057529598474502563
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,32,1,0,0.039062398672103885
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,2048,1,0,0.08456000089645385
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,64,1,0,0.04079360067844391
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,64,1,0,0.0564191997051239
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,1024,1,0,0.12861119508743285
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1,16384,1,0,0.13127679824829103
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,4096,1,0,0.08936319947242737
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,128,1,0,0.04223679900169373
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,128,1,0,0.055929601192474365
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,8192,1,0,0.0958303987979889
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,2048,1,0,0.13628799915313722
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,256,1,0,0.04240959882736206
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,256,1,0,0.055827200412750244
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,512,1,0,0.04425599873065948
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,512,1,0,0.057897597551345825
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,2,16384,1,0,0.10285760164260864
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,1024,1,0,0.046460801362991334
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,1024,1,0,0.061791998147964475
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,4096,1,0,0.15259840488433837
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,1,1,0,0.07280319929122925
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,2048,1,0,0.04875200092792511
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,2048,1,0,0.06178879737854004
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,1,1,0,0.05543680191040039
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,4,1,0,0.07276480197906494
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,4096,1,0,0.06340799927711487
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,4,1,0,0.05554559826850891
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,4096,1,0,0.053180801868438723
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,8,1,0,0.07278079986572265
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,8,1,0,0.05539839863777161
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,8192,1,0,0.06821439862251281
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,8192,1,0,0.18214399814605714
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,8192,1,0,0.06047359704971313
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,16,1,0,0.07286720275878907
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,16,1,0,0.05591359734535217
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1,16384,1,0,0.07594559788703918
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,32,1,0,0.05575039982795715
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,32,1,0,0.07326080203056336
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,2,16384,1,0,0.07106239795684814
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,1,1,0,0.03494080007076263
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,64,1,0,0.05742080211639404
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,1,1,0,0.03904640078544617
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,64,1,0,0.07459520101547241
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,4,1,0,0.03457599878311157
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,128,1,0,0.05889279842376709
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,1,1,0,0.06958400011062622
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,4,1,0,0.039139199256896975
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,8,1,0,0.035020801424980166
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,128,1,0,0.07645440101623535
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,256,1,0,0.06200640201568604
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,8,1,0,0.03903039991855621
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,4,1,0,0.06989120244979859
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,16,1,0,0.035046398639678955
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,16,16384,1,0,0.2414720058441162
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,512,1,0,0.06359360218048096
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,256,1,0,0.08124480247497559
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,16,1,0,0.03941439986228943
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,8,1,0,0.0704255998134613
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,1,1,0,0.10588480234146118
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,32,1,0,0.038217601180076596
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,1024,1,0,0.06443840265274048
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,32,1,0,0.0395552009344101
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,512,1,0,0.08311359882354737
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,1,1,0,0.10885759592056274
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,4,1,0,0.10580159425735473
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,16,1,0,0.07066239714622498
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,64,1,0,0.038134399056434634
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,2048,1,0,0.06606400012969971
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,64,1,0,0.04091199934482574
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,8,1,0,0.10601279735565186
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,1024,1,0,0.08414720296859741
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,4,1,0,0.10901440382003784
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,32,1,0,0.07901120185852051
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,128,1,0,0.037459200620651244
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,128,1,0,0.042742401361465454
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,16,1,0,0.10551999807357788
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,4096,1,0,0.06907200217247009
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,2048,1,0,0.0893664002418518
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,8,1,0,0.10823040008544922
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,64,1,0,0.07761600017547607
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,256,1,0,0.03808639943599701
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,256,1,0,0.04321599900722504
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,32,1,0,0.10616320371627808
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,8192,1,0,0.07339199781417846
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,16,1,0,0.10879679918289184
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,128,1,0,0.07681599855422974
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,512,1,0,0.04095999896526337
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,4096,1,0,0.09402239918708802
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,512,1,0,0.04565440118312836
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,64,1,0,0.1074560046195984
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,32,1,0,0.10956480503082275
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,2,16384,1,0,0.08067839741706848
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,1024,1,0,0.04653120040893555
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,1024,1,0,0.04028480052947998
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,256,1,0,0.07620800137519837
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,128,1,0,0.10911999940872193
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,8192,1,0,0.1024448037147522
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,1,1,0,0.05305600166320801
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,1,1,0,0.05592640042304993
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,2048,1,0,0.050095999240875246
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,64,1,0,0.11110719442367553
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,2048,1,0,0.043171200156211856
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,512,1,0,0.07579200267791748
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,4,1,0,0.05616639852523804
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,256,1,0,0.11285439729690552
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,4,1,0,0.05315840244293213
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,8,1,0,0.05588480234146118
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,4096,1,0,0.05914880037307739
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,128,1,0,0.11273280382156373
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,8,1,0,0.05335999727249145
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,4,16384,1,0,0.11600960493087768
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,512,1,0,0.11442879438400269
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,4096,1,0,0.047603198885917665
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,1024,1,0,0.07601919770240784
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,16,1,0,0.05674880146980286
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,16,1,0,0.053513598442077634
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,256,1,0,0.1283776044845581
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,1024,1,0,0.11641600131988525
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,1,1,0,0.07410879731178284
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,2048,1,0,0.07776319980621338
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,32,1,0,0.05676159858703613
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,8192,1,0,0.07003200054168701
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,8192,1,0,0.056092798709869385
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,32,1,0,0.05805439949035644
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,64,1,0,0.05788159966468811
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,4,1,0,0.07351040244102477
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,512,1,0,0.1373695969581604
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,64,1,0,0.05709120035171509
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,2048,1,0,0.11966400146484375
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,2,16384,1,0,0.07327359914779663
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,128,1,0,0.05969280004501343
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,4096,1,0,0.08176320195198059
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,4,16384,1,0,0.08308799862861634
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,1,1,0,0.035392001271247864
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,128,1,0,0.05671039819717407
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,4096,1,0,0.12435840368270874
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,256,1,0,0.0632863998413086
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,8,1,0,0.07525759935379028
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,1024,1,0,0.1460096001625061
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,1,1,0,0.03841600120067597
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,8192,1,0,0.08702719807624817
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,4,1,0,0.035359999537467955
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,256,1,0,0.05614079833030701
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,512,1,0,0.06524800062179566
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,16,1,0,0.07294719815254211
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,4,1,0,0.03792639970779419
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,8192,1,0,0.12890880107879638
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,8,1,0,0.03531840145587921
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,512,1,0,0.0581055998802185
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,1024,1,0,0.06693120002746582
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,8,1,0,0.03812800049781799
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,16,1,0,0.03553600013256073
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,2048,1,0,0.16242239475250245
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,32,1,0,0.07414720058441163
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,2,16384,1,0,0.1008031964302063
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,2048,1,0,0.06917440295219421
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,1024,1,0,0.05861759781837463
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,16,1,0,0.03837760090827942
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,2,16384,1,0,0.13870079517364503
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,32,1,0,0.037436801195144656
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,64,1,0,0.07522240281105042
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,1,1,0,0.07015680074691773
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,32,1,0,0.038422399759292604
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,4096,1,0,0.07223680019378662
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,64,1,0,0.03709439933300018
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,2048,1,0,0.06087679862976074
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,1,1,0,0.10668799877166749
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,4,1,0,0.06934080123901368
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,128,1,0,0.07679039835929871
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,64,1,0,0.03983359932899475
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,128,1,0,0.03726080060005188
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,8192,1,0,0.07927359938621521
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,32,4096,1,0,0.19174400568008423
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,4096,1,0,0.06373119950294495
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,4,1,0,0.10655360221862793
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,8,1,0,0.06935679912567139
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,128,1,0,0.04156160056591034
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,256,1,0,0.03809599876403809
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,256,1,0,0.08347839713096619
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,16,1,0,0.07014080286026
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,8,1,0,0.10645120143890381
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,512,1,0,0.037894400954246524
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,256,1,0,0.043859198689460754
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,8192,1,0,0.06913920044898987
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,1,1,0,0.1143231987953186
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,1024,1,0,0.040678399801254275
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,4,16384,1,0,0.09292160272598267
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,512,1,0,0.08418560028076172
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,32,1,0,0.07999039888381958
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,512,1,0,0.04576640129089356
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,4,1,0,0.11499520540237426
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,2,16384,1,0,0.08208960294723511
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,1024,1,0,0.048105600476264956
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,2048,1,0,0.0442111998796463
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,1,1,0,0.05335999727249145
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,64,1,0,0.07786880135536194
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,8,1,0,0.1147968053817749
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,1024,1,0,0.09016640186309814
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,1,1,0,0.056041598320007324
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,2048,1,0,0.05856000185012818
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,16,1,0,0.10707520246505738
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,4,1,0,0.053388798236846925
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,128,1,0,0.0762880027294159
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,4,1,0,0.05638719797134399
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,16,1,0,0.11442559957504272
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,4096,1,0,0.05299839973449707
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,256,1,0,0.0757856011390686
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,8192,1,0,0.06655359864234925
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,8,1,0,0.05318719744682312
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,8,1,0,0.056460797786712646
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,32,1,0,0.10688960552215576
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,32,1,0,0.11509439945220948
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,4096,1,0,0.06814720034599304
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,2048,1,0,0.09419839978218078
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,16,1,0,0.05358399748802185
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,4,16384,1,0,0.07882879972457886
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,16,1,0,0.05635200142860412
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,512,1,0,0.0763264000415802
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,64,1,0,0.10799360275268555
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,64,1,0,0.11638720035552978
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,32,1,0,0.058284801244735715
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,8192,1,0,0.0828000009059906
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,32,1,0,0.05649600028991699
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,1024,1,0,0.07801600098609925
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,4096,1,0,0.10195519924163818
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,1,1,0,0.03474240005016327
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,64,1,0,0.056857597827911374
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,128,1,0,0.1099392056465149
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,128,1,0,0.11829440593719483
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,64,1,0,0.05779520273208618
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,128,1,0,0.05693119764328003
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,2048,1,0,0.0808896005153656
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,4,1,0,0.03451519906520843
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,128,1,0,0.059113597869873045
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,8192,1,0,0.11504000425338745
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,256,1,0,0.11692800521850585
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,256,1,0,0.05627520084381103
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,256,1,0,0.14991040229797364
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,4096,1,0,0.08299199938774109
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,8,1,0,0.03498240113258362
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,256,1,0,0.06450560092926025
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,8,16384,1,0,0.10742080211639404
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,512,1,0,0.05633280277252197
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,512,1,0,0.11838400363922119
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,512,1,0,0.1640447974205017
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,512,1,0,0.06581119894981384
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,1024,1,0,0.058524799346923825
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,16,1,0,0.035174399614334106
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,8192,1,0,0.09354879856109619
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,1,1,0,0.03865920007228851
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,8,16384,1,0,0.14089280366897583
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,2048,1,0,0.060844802856445314
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,32,1,0,0.037302398681640626
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,1024,1,0,0.06844480037689209
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,1024,1,0,0.1221824049949646
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,4,1,0,0.03932160139083862
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,1,1,0,0.0746720016002655
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,4096,1,0,0.06598719954490662
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,4,16384,1,0,0.10555200576782227
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,1024,1,0,0.18502399921417237
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,2048,1,0,0.07180479764938355
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,64,1,0,0.037110400199890134
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,8,1,0,0.03881919980049133
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,4,1,0,0.07463679909706115
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,1,1,0,0.06978240013122558
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,2048,1,0,0.12550400495529174
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,8192,1,0,0.07484800219535828
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,4096,1,0,0.07866880297660828
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,16,1,0,0.039241600036621097
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,128,1,0,0.03654719889163971
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,4,1,0,0.07024639844894409
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,8,1,0,0.07457919716835022
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,4096,1,0,0.12987840175628662
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,4,16384,1,0,0.08766400218009948
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,32,1,0,0.03912000060081482
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,2048,1,0,0.21412160396575927
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,256,1,0,0.03685440123081207
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,8,1,0,0.07067840099334717
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,8192,1,0,0.09207680225372314
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,1,1,0,0.053097599744796754
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,16,1,0,0.07434239983558655
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,64,1,0,0.04113920032978058
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,16,1,0,0.07032319903373718
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,512,1,0,0.039084801077842714
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,8192,1,0,0.13971199989318847
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,4,1,0,0.05261440277099609
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,32,1,0,0.0747488021850586
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,128,1,0,0.042342400550842284
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,1024,1,0,0.04180159866809845
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,8,16384,1,0,0.11910400390625
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,32,1,0,0.08101119995117187
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,8,1,0,0.05350720286369324
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,1,1,0,0.05646399855613708
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,64,1,0,0.07842239737510681
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,2048,1,0,0.052393597364425656
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,4,16384,1,0,0.15606080293655394
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,64,4096,1,0,0.27464001178741454
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,64,1,0,0.07651839852333069
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,256,1,0,0.0456959992647171
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,16,1,0,0.05386880040168762
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,4,1,0,0.05654079914093017
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,128,1,0,0.07816640138626099
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,4096,1,0,0.06673280000686646
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,8,1,0,0.056828802824020384
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,1,1,0,0.10807679891586304
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,1,1,0,0.1322816014289856
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,32,1,0,0.0588703989982605
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,128,1,0,0.0781279981136322
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,512,1,0,0.04984639883041382
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,16,1,0,0.056828802824020384
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,8192,1,0,0.07844799757003784
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,4,1,0,0.10794559717178345
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,64,1,0,0.057359999418258666
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,4,1,0,0.13310400247573853
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,256,1,0,0.0776639997959137
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,32,1,0,0.05691519975662231
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,1024,1,0,0.05726400017738342
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,256,1,0,0.089273601770401
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,128,1,0,0.05713919997215271
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,64,1,0,0.057974398136138916
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,8,1,0,0.13300479650497438
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,8,1,0,0.10794880390167236
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,512,1,0,0.07880640029907227
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,8,16384,1,0,0.10241600275039672
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,2048,1,0,0.06855999827384948
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,256,1,0,0.0574783980846405
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,128,1,0,0.05984640121459961
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,512,1,0,0.09118720293045043
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,1,1,0,0.03436799943447113
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,16,1,0,0.10806399583816528
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,1024,1,0,0.08277119994163513
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,512,1,0,0.0591264009475708
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,16,1,0,0.13337279558181764
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,256,1,0,0.06659839749336242
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,4,1,0,0.0343392014503479
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,1024,1,0,0.09465600252151489
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,4096,1,0,0.08285120129585266
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,2048,1,0,0.08383039832115173
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,1024,1,0,0.06176959872245789
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,32,1,0,0.10851520299911499
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,32,1,0,0.13424320220947267
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,8,1,0,0.03446080088615418
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,512,1,0,0.06926400065422059
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,2048,1,0,0.06393280029296874
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,64,1,0,0.10977280139923096
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,16,1,0,0.03514240086078644
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,2048,1,0,0.10300480127334595
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,4096,1,0,0.09567040205001831
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,1024,1,0,0.07227519750595093
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,8192,1,0,0.10727360248565673
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,64,1,0,0.13749439716339112
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,4096,1,0,0.07700160145759583
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,32,1,0,0.03719359934329987
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,128,1,0,0.11146880388259887
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,2048,1,0,0.07961279749870301
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,4096,1,0,0.11655360460281372
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,8192,1,0,0.10827840566635132
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,64,1,0,0.03643839955329895
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,256,1,0,0.12069439888000488
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,128,1,0,0.1408735990524292
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,8192,1,0,0.08891519904136658
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,128,1,0,0.03647040128707886
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,4096,1,0,0.09336640238761902
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,512,1,0,0.12223999500274658
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,256,1,0,0.041196799278259276
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,8192,1,0,0.14132159948349
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,8,16384,1,0,0.13191360235214233
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,256,1,0,0.15012480020523072
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,512,1,0,0.04503040015697479
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,16,16384,1,0,0.1559167981147766
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,8,16384,1,0,0.11262719631195069
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,1024,1,0,0.12563519477844237
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,1,1,0,0.0701856017112732
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,8192,1,0,0.11840959787368774
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,1024,1,0,0.051974397897720334
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,1,1,0,0.04011200070381164
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,4,1,0,0.07133439779281617
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,512,1,0,0.16895680427551268
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,2048,1,0,0.13093119859695435
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,1,1,0,0.05389440059661865
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,4,1,0,0.04039359986782074
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,2048,1,0,0.06451200246810913
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,8,1,0,0.0709119975566864
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,4,1,0,0.053887999057769774
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,8,1,0,0.0399071991443634
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,4096,1,0,0.140774405002594
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,16,1,0,0.0716159999370575
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,8,1,0,0.053635197877883914
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,16,16384,1,0,0.18959039449691772
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,4096,1,0,0.07672320008277893
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,16,1,0,0.04027200043201447
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,16,1,0,0.05373439788818359
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,16,16384,1,0,0.1674080014228821
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,32,1,0,0.08267840147018432
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,32,1,0,0.04020799994468689
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,1024,1,0,0.19987839460372925
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,8192,1,0,0.1541632056236267
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,1,1,0,0.07710719704627991
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,32,1,0,0.05952640175819397
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,64,1,0,0.042208001017570496
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,64,1,0,0.08024320006370544
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,1,1,0,0.05899839997291565
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,8192,1,0,0.10073920488357543
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,64,1,0,0.05818560123443604
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,4,1,0,0.07753599882125854
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,128,1,0,0.04644480049610138
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,128,1,0,0.07961279749870301
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,4,1,0,0.05823040008544922
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,128,1,0,0.058396798372268674
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,8,1,0,0.07714560031890869
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,256,1,0,0.054041600227355956
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,8,16384,1,0,0.18196799755096435
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,256,1,0,0.08256959915161133
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,8,1,0,0.058620798587799075
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,256,1,0,0.060575997829437254
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,16,1,0,0.07760639786720276
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,512,1,0,0.06073279976844788
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,1,1,0,0.10970560312271119
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,16,1,0,0.0587552011013031
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,512,1,0,0.08415039777755737
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,512,1,0,0.0622111976146698
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,128,2048,1,0,0.2606015920639038
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,32,1,0,0.07724480032920837
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,32,1,0,0.05931839942932129
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,4,1,0,0.10973119735717773
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,16,16384,1,0,0.1475391983985901
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,1024,1,0,0.06509439945220948
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,1024,1,0,0.08629119992256165
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,1024,1,0,0.0697920024394989
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,64,1,0,0.07924479842185975
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,64,1,0,0.06068159937858582
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,1,1,0,0.035043200850486754
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,8,1,0,0.10961920022964478
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,2048,1,0,0.07553600072860718
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,2048,1,0,0.09676480293273926
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,256,1,1,0,0.1794656038284302
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,128,1,0,0.08082240223884582
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,2048,1,0,0.08275840282440186
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,4,1,0,0.03659839928150177
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,128,1,0,0.06229119896888733
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,16,1,0,0.10964800119400024
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,4096,1,0,0.08814079761505127
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,8,1,0,0.03607679903507233
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,256,1,0,0.09583680033683777
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,256,1,0,0.0709824025630951
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,4096,1,0,0.10900800228118897
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,32,1,0,0.1099392056465149
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,256,4,1,0,0.18037439584732057
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,16,1,0,0.036473599076271054
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,64,1,0,0.11123520135879517
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,512,1,0,0.09882240295410157
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,512,1,0,0.07486079931259156
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,32,4096,1,0,0.10707839727401733
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,32,1,0,0.038915199041366574
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,8192,1,0,0.11241919994354248
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,128,1,0,0.11284159421920777
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,64,1,0,0.04157440066337585
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,256,8,1,0,0.18046400547027588
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,8192,1,0,0.13335039615631103
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,1024,1,0,0.08110399842262268
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,1,1,0,0.043024000525474546
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,1024,1,0,0.10675840377807617
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,128,1,0,0.04222719967365265
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,256,1,0,0.1258463978767395
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,4,1,0,0.042761600017547606
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,256,1,0,0.04752640128135681
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,256,16,1,0,0.1803264021873474
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,2048,1,0,0.09504640102386475
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,512,1,0,0.12819839715957643
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,2048,1,0,0.12024960517883301
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,8,1,0,0.042403200268745424
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,512,1,0,0.05390080213546753
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,16,16384,1,0,0.1591423988342285
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,16,1,0,0.04343999922275543
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,1024,1,0,0.13420799970626832
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,256,32,1,0,0.18223999738693236
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,1024,1,0,0.06737279891967773
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,1,1,0,0.05453760027885437
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,16,16384,1,0,0.18043839931488037
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,32,1,0,0.04360640048980713
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,32,4096,1,0,0.11930559873580933
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,2048,1,0,0.1428096055984497
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,4,1,0,0.054636800289154054
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,32,4096,1,0,0.14340800046920776
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,256,64,1,0,0.18921279907226562
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,2048,1,0,0.0796671986579895
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,64,1,0,0.04564160108566284
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,1,1,0,0.07287359833717347
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,8,1,0,0.05480639934539795
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,1,1,0,0.06061440110206604
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,1,1,0,0.08097919821739197
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,128,1,0,0.047788798809051514
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,4096,1,0,0.15647679567337036
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,4,1,0,0.07308160066604615
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,16,1,0,0.05453439950942993
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,4,1,0,0.06058560013771057
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,4,1,0,0.08118720054626465
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,256,128,1,0,0.198089599609375
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,32,4096,1,0,0.102128005027771
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,32,1,0,0.06090559959411621
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,256,1,0,0.06298239827156067
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,8,1,0,0.0725600004196167
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,8,1,0,0.06084160208702087
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,8,1,0,0.08113600015640259
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,64,1,0,0.06092159748077393
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,16,1,0,0.07298880219459533
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,1,1,0,0.03675200045108795
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,16,1,0,0.06080639958381653
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,8192,1,0,0.18397760391235352
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,512,1,0,0.07205119729042053
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,16,1,0,0.08164160251617432
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,128,1,0,0.06386880278587341
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,256,256,1,0,0.21370880603790282
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,32,1,0,0.0847487986087799
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,4,1,0,0.037724798917770384
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,32,1,0,0.06130560040473938
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,32,1,0,0.08161919713020324
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,256,1,0,0.06459199786186218
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,1024,1,0,0.08526399731636047
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,64,1,0,0.08577600121498108
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,8,1,0,0.0365664005279541
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,64,1,0,0.0634112000465393
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,64,1,0,0.08360959887504578
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,512,1,0,0.06683520078659058
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,128,1,0,0.08813440203666686
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,16,1,0,0.03835200071334839
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,128,1,0,0.06517760157585144
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,128,1,0,0.08528000116348267
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,256,512,1,0,0.2485152006149292
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,1024,1,0,0.0794655978679657
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,256,1,0,0.08956159949302674
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,2048,1,0,0.10926079750061035
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,32,1,0,0.03930239975452423
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,16,16384,1,0,0.23960959911346436
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,256,1,0,0.07775359749794006
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,256,1,0,0.10468800067901611
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,512,1,0,0.08973760008811951
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,64,1,0,0.04238399863243103
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,2048,1,0,0.09198079705238342
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,1,1,0,0.11295679807662964
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,512,1,0,0.0851967990398407
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,512,1,0,0.11210240125656128
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,128,1,0,0.04604479968547821
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,1024,1,0,0.10392320156097412
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,4,1,0,0.11270079612731934
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,256,1,0,0.05575360059738159
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,1024,1,0,0.09807360172271729
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,32,4096,1,0,0.1139456033706665
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,1024,1,0,0.12541760206222535
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,8,1,0,0.11321280002593995
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,2048,1,0,0.11679359674453735
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,64,4096,1,0,0.1589408040046692
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,256,1024,1,0,0.3124383926391602
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,512,1,0,0.06852160096168518
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,1,1,0,0.056620800495147706
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,16,1,0,0.11370240449905396
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,2048,1,0,0.12278079986572266
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,32,1,0,0.11332160234451294
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,1,1,0,0.04774720072746277
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,4,1,0,0.05659840106964111
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,1024,1,0,0.08254079818725586
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,2048,1,0,0.14967360496520996
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,32,4096,1,0,0.13969600200653076
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,64,1,0,0.11529920101165772
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,8,1,0,0.056591999530792234
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,4,1,0,0.04884159862995148
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,512,1,1,0,0.3040735960006714
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,1,1,0,0.07582719922065735
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,128,1,0,0.11702719926834107
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,16,1,0,0.05654079914093017
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,8,1,0,0.04910080134868622
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,2048,1,0,0.1063423991203308
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,4,1,0,0.07571200132369996
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,32,1,0,0.05700799822807312
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,256,1,0,0.13630720376968383
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,16,1,0,0.05045440196990967
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,64,4096,1,0,0.17273279428482055
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,8,1,0,0.07577919960021973
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,512,4,1,0,0.30852479934692384
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,64,1,0,0.05821120142936707
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,32,1,0,0.049193599820137025
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,64,4096,1,0,0.2002880096435547
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,512,1,0,0.13910720348358155
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,16,1,0,0.07628160119056701
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,128,1,0,0.060127997398376466
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,1,1,0,0.06616640090942383
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,64,1,0,0.055504000186920165
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,32,1,0,0.07654399871826172
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,256,1,0,0.07047680020332336
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,4,1,0,0.06590080261230469
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,1024,1,0,0.14749759435653687
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,1,1,0,0.09003199934959412
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,512,8,1,0,0.30744318962097167
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,128,1,0,0.06041600108146668
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,64,4096,1,0,0.1533023953437805
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,64,1,0,0.0784063994884491
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,8,1,0,0.0661408007144928
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,4,1,0,0.08971199989318848
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,512,1,0,0.08229759931564332
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,128,1,0,0.07990400195121765
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,1,1,0,0.04143680036067963
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,2048,1,0,0.16241919994354248
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,16,1,0,0.06665279865264892
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,256,1,0,0.07370880246162415
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,8,1,0,0.090447998046875
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,1024,1,0,0.09732159972190857
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,512,16,1,0,0.3063647985458374
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,4,1,0,0.04240640103816986
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,256,1,0,0.0958400011062622
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,32,1,0,0.0670144021511078
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,16,1,0,0.09008960127830505
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,8,1,0,0.04439040124416351
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,512,1,0,0.08678719997406006
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,64,1,0,0.06950079798698425
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,512,1,0,0.10857280492782592
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,32,1,0,0.09132159948348999
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,32,4096,1,0,0.1904736042022705
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,16,1,0,0.04529919922351837
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,2048,1,0,0.12131839990615845
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,512,32,1,0,0.308841609954834
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,128,1,0,0.07299839854240417
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,64,1,0,0.09263359904289245
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,32,1,0,0.04653440117835998
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,1024,1,0,0.12401920557022095
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,1,1,0,0.11986240148544311
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,1024,1,0,0.11159360408782959
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,256,1,0,0.08525760173797607
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,64,1,0,0.04893760085105896
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,128,1,0,0.09520320296287536
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,4,1,0,0.1196671962738037
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,512,64,1,0,0.32036800384521485
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,512,1,0,0.0981760025024414
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,128,1,0,0.05388479828834534
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,8,1,0,0.11989120244979859
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,256,1,0,0.10914880037307739
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,2048,1,0,0.14941120147705078
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,64,4096,1,0,0.16779520511627197
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,16,1,0,0.11976000070571899
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,256,1,0,0.07232319712638854
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,1,1,0,0.06124160289764404
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,512,128,1,0,0.33736960887908934
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,512,1,0,0.12318719625473022
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,32,1,0,0.11991679668426514
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,128,2048,1,0,0.16081600189208983
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,1024,1,0,0.12300479412078857
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,4,1,0,0.06104000210762024
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,512,1,0,0.09089919924736023
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,64,1,0,0.12183680534362792
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,8,1,0,0.061375999450683595
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,256,1,1,0,0.06428160071372986
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,128,1,0,0.1232416033744812
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,16,1,0,0.06147199869155884
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,1024,1,0,0.14707839488983154
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,64,4096,1,0,0.1962015986442566
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,256,4,1,0,0.0657920002937317
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,1024,1,0,0.11545920372009277
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,32,1,0,0.06145600080490112
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,256,1,0,0.13382400274276735
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,512,256,1,0,0.36957759857177735
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,256,8,1,0,0.06510720252990723
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,1,1,0,0.08398720026016235
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,128,2048,1,0,0.17145919799804688
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,64,1,0,0.06253759860992432
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,512,1,0,0.14007680416107177
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,4,1,0,0.08423039913177491
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,256,16,1,0,0.06613439917564393
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,128,1,0,0.06619200110435486
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,256,1,1,0,0.08121920228004456
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,8,1,0,0.08371840119361877
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,256,32,1,0,0.06795200109481811
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,1024,1,0,0.1548383951187134
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,128,2048,1,0,0.19701119661331176
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,256,4,1,0,0.08187519907951354
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,256,1,0,0.08821120262145996
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,128,2048,1,0,0.16183680295944214
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,16,1,0,0.08416640162467956
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,256,64,1,0,0.07313920259475708
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,256,8,1,0,0.08149120211601257
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,32,1,0,0.08517760038375854
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,256,1,1,0,0.05508480072021484
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,512,1,0,0.10834239721298218
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,256,1,1,0,0.11046719551086426
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,512,512,1,0,0.43703360557556153
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,2048,1,0,0.18271679878234864
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,256,128,1,0,0.08014400005340576
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,256,16,1,0,0.08222079873085023
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,64,1,0,0.08695679903030396
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,256,4,1,0,0.05620800256729126
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,256,4,1,0,0.11064319610595703
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,256,32,1,0,0.0825119972229004
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,1024,1,0,0.13305920362472534
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,256,256,1,0,0.10175360441207885
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,128,1,0,0.08822720050811768
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,256,8,1,0,0.057708799839019775
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,256,8,1,0,0.11149439811706544
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,256,64,1,0,0.08661440014839172
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,256,16,1,0,0.05897279977798462
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1024,1,1,0,0.554095983505249
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,256,1,0,0.12354559898376465
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,256,16,1,0,0.11144000291824341
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,64,4096,1,0,0.23946878910064698
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,256,32,1,0,0.0604095995426178
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,256,128,1,0,0.0946175992488861
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,256,512,1,0,0.12758400440216064
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,256,32,1,0,0.11296960115432739
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,512,1,0,0.14468159675598144
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,128,2048,1,0,0.18061439990997313
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,1,1,0,0.1374848008155823
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,256,64,1,0,0.06507840156555175
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,256,64,1,0,0.11717760562896729
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,256,256,1,0,0.11815680265426635
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,4,1,0,0.13768960237503053
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1024,4,1,0,0.5586080074310302
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,256,1,1,0,0.07422080039978027
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,256,128,1,0,0.07182719707489013
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,1024,1,0,0.1694591999053955
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,8,1,0,0.13815040588378907
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,256,128,1,0,0.12434240579605102
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,256,4,1,0,0.0744480013847351
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,256,1024,1,0,0.17522239685058594
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,256,256,1,0,0.08608319759368896
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,16,1,0,0.13799999952316283
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,256,512,1,0,0.14377599954605103
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,256,8,1,0,0.07520959973335266
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,256,256,1,0,0.15127999782562257
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,32,1,0,0.13888959884643554
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1024,8,1,0,0.5595871925354003
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,256,16,1,0,0.07517120242118835
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,512,1,1,0,0.09571840167045594
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,64,1,0,0.14214400053024293
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,256,512,1,0,0.11026879549026489
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,128,2048,1,0,0.21651840209960938
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,256,32,1,0,0.07629759907722473
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,512,4,1,0,0.09550719857215881
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,128,1,0,0.1468832015991211
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,256,512,1,0,0.17887680530548095
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,256,64,1,0,0.07979199886322022
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,256,1024,1,0,0.19244799613952637
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,256,1,1,0,0.10261119604110717
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,512,8,1,0,0.09744960069656372
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1024,16,1,0,0.5609312057495117
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,256,1,0,0.16491520404815674
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,256,4,1,0,0.10333440303802491
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,256,128,1,0,0.08659840226173401
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,256,1024,1,0,0.15722559690475463
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,512,16,1,0,0.09647679924964905
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,512,1,0,0.18122559785842896
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,512,1,1,0,0.11725759506225586
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,256,1024,1,0,0.22753920555114746
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,256,8,1,0,0.10322879552841187
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1024,32,1,0,0.5642943859100342
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,512,32,1,0,0.09902399778366089
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,256,256,1,0,0.10213760137557984
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,512,4,1,0,0.11876800060272216
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,512,1,1,0,0.08260800242424012
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,256,16,1,0,0.10443520545959473
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,1024,1,0,0.21074240207672118
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,512,1,1,0,0.18142720460891723
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,256,32,1,0,0.10663360357284546
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,512,8,1,0,0.11938240528106689
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,512,4,1,0,0.0834879994392395
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,512,64,1,0,0.1086400032043457
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,256,512,1,0,0.12688000202178956
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,512,4,1,0,0.1815775990486145
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,256,64,1,0,0.11101759672164917
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1024,64,1,0,0.5841728210449219
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,512,8,1,0,0.08429759740829468
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,512,16,1,0,0.12067199945449829
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,512,128,1,0,0.12245440483093262
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,256,128,1,0,0.11746879816055297
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,512,8,1,0,0.18275519609451293
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,128,2048,1,0,0.2653343915939331
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,256,1024,1,0,0.17359999418258668
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,512,32,1,0,0.12330880165100097
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,512,16,1,0,0.08590400218963623
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,256,256,1,0,0.13249599933624268
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,512,256,1,0,0.1592512011528015
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,256,1,1,0,0.1877344012260437
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,512,1,1,0,0.10675840377807617
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,512,32,1,0,0.08964800238609313
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,512,64,1,0,0.13262720108032228
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,512,16,1,0,0.1838111996650696
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1024,128,1,0,0.6173535823822022
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,256,512,1,0,0.158624005317688
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,256,4,1,0,0.18852479457855226
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,512,512,1,0,0.21098558902740477
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,512,32,1,0,0.18671040534973143
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,512,64,1,0,0.09609919786453247
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,512,4,1,0,0.10770239830017089
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,256,1024,1,0,0.2065727949142456
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,512,128,1,0,0.14564800262451172
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,128,1024,256,1,0,0.6926496028900146
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,512,128,1,0,0.1087615966796875
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1024,1,1,0,0.14340800046920776
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,256,8,1,0,0.1897279977798462
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,512,64,1,0,0.19287680387496947
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,512,8,1,0,0.10922240018844605
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,512,1,1,0,0.16856000423431397
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,256,16,1,0,0.18988480567932128
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1024,4,1,0,0.14469120502471924
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,512,256,1,0,0.1876863956451416
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,512,4,1,0,0.17039999961853028
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,512,256,1,0,0.13533439636230468
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,512,128,1,0,0.20562880039215087
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,512,16,1,0,0.11136000156402588
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,256,32,1,0,0.1915295958518982
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,512,8,1,0,0.170959997177124
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1024,8,1,0,0.14557440280914308
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,512,512,1,0,0.23931519985198973
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,512,32,1,0,0.11512000560760498
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,256,64,1,0,0.19600319862365723
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,512,256,1,0,0.25517759323120115
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,512,512,1,0,0.18510719537734985
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,512,16,1,0,0.1716863989830017
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1024,1,1,0,0.2020064115524292
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1024,16,1,0,0.14840320348739625
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,512,64,1,0,0.12202240228652954
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,256,128,1,0,0.2041599988937378
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1024,1,1,0,0.12714879512786864
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,512,512,1,0,0.3060960054397583
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,512,32,1,0,0.17443840503692626
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,512,128,1,0,0.13439040184020995
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1024,32,1,0,0.1520032048225403
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1024,4,1,0,0.20313599109649658
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1024,4,1,0,0.1294592022895813
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,256,256,1,0,0.23685760498046876
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1024,1,1,0,0.3193984031677246
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1024,64,1,0,0.16830079555511473
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,512,64,1,0,0.1806656002998352
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,512,256,1,0,0.16136640310287476
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1024,8,1,0,0.20475521087646484
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1024,4,1,0,0.3224800109863281
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,256,512,1,0,0.2687135934829712
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1024,8,1,0,0.13179199695587157
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,512,128,1,0,0.19425920248031617
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1024,16,1,0,0.20673599243164062
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1024,128,1,0,0.19000639915466308
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1024,8,1,0,0.3209120035171509
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,512,512,1,0,0.21215360164642333
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1024,16,1,0,0.14014400243759156
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,256,1024,1,0,0.3253216028213501
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1024,32,1,0,0.20894079208374022
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1024,16,1,0,0.32410879135131837
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,512,256,1,0,0.22135679721832274
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1024,1,1,0,0.190447998046875
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,512,1,1,0,0.32191359996795654
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,16,1024,256,1,0,0.25520639419555663
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1024,32,1,0,0.14363839626312255
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1024,64,1,0,0.2225696086883545
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,512,512,1,0,0.27311038970947266
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1024,32,1,0,0.3252000093460083
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,512,4,1,0,0.32447359561920164
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1024,4,1,0,0.1928704023361206
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1024,64,1,0,0.15274239778518678
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1024,128,1,0,0.24509119987487793
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1024,1,1,0,0.3020704030990601
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1024,8,1,0,0.19306559562683107
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1024,64,1,0,0.33639678955078123
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,512,8,1,0,0.32704958915710447
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1024,128,1,0,0.17475520372390746
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,32,1024,256,1,0,0.31850879192352294
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1024,4,1,0,0.3057919979095459
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1024,16,1,0,0.19672640562057495
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1024,128,1,0,0.3612512111663818
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,16,1024,256,1,0,0.22483839988708496
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,512,16,1,0,0.3295520067214966
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1024,8,1,0,0.30684800148010255
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1024,32,1,0,0.2010495901107788
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,512,32,1,0,0.3289344072341919
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,64,1024,256,1,0,0.4502880096435547
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1024,16,1,0,0.30601599216461184
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,512,64,1,0,0.33741440773010256
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1024,64,1,0,0.2101344108581543
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,512,128,1,0,0.3514080047607422
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1024,32,1,0,0.3096575975418091
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1024,128,1,0,0.2336479902267456
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1024,64,1,0,0.32187519073486326
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,512,256,1,0,0.41192002296447755
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1024,128,1,0,0.34628479480743407
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,512,512,1,0,0.47203841209411623
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,32,1024,256,1,0,0.28519039154052733
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1024,1,1,0,0.5838399887084961
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,flashinfer,fp8_block,fp8,64,1024,256,1,0,0.3992511987686157
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1024,4,1,0,0.5865056037902832
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1024,8,1,0,0.5884191989898682
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1024,16,1,0,0.5997536182403564
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1024,32,1,0,0.6059807777404785
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1024,64,1,0,0.6095168113708496
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1024,128,1,0,0.6452127933502197
SGLang,0.5.8.post1,NVIDIA H100 80GB HBM3,mla_generation,fa3,fp8_block,fp8,128,1024,256,1,0,0.7585087776184082
