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 GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,1,1,0,0.05623040199279785
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,4,1,0,0.056524801254272464
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,8,1,0,0.056380802392959596
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,16,1,0,0.055980801582336426
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,32,1,0,0.06427519917488098
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,64,1,0,0.06362239718437195
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,1,1,0,0.0782751977443695
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,128,1,0,0.062028801441192626
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,256,1,0,0.0603488028049469
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,512,1,0,0.06209920048713684
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,1024,1,0,0.0659775972366333
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,2048,1,0,0.06451839804649354
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,4096,1,0,0.07057279944419861
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,8192,1,0,0.08267199993133545
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1,16384,1,0,0.0951903998851776
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,4,1,0,0.07976319789886474
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,8,1,0,0.07837439775466919
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,16,1,0,0.07848320007324219
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,32,1,0,0.0867904007434845
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,64,1,0,0.08635839819908142
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,1,1,0,0.05647040009498596
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,4,1,0,0.056032001972198486
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,8,1,0,0.05811200141906738
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,16,1,0,0.05594559907913208
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,32,1,0,0.06448959708213806
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,64,1,0,0.06232960224151611
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,128,1,0,0.06213120222091675
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,256,1,0,0.06087679862976074
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,512,1,0,0.06084160208702087
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,1024,1,0,0.06084160208702087
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,2048,1,0,0.06661120057106018
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,4096,1,0,0.07418879866600037
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,8192,1,0,0.08568000197410583
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,128,1,0,0.08424959778785705
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,256,1,0,0.08471999764442444
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,512,1,0,0.08608319759368896
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,1024,1,0,0.08496320247650146
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,2048,1,0,0.08999040126800537
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,4096,1,0,0.09579839706420898
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,8192,1,0,0.10461440086364746
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,1,1,0,0.03956480026245117
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,4,1,0,0.04176959991455078
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,8,1,0,0.041126400232315063
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,16,1,0,0.03962239921092987
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,32,1,0,0.04355199933052063
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,64,1,0,0.04332799911499023
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,128,1,0,0.041606399416923526
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,256,1,0,0.04193600118160248
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,512,1,0,0.0457152009010315
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,1024,1,0,0.049702399969100954
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,2048,1,0,0.049932798743247984
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,4096,1,0,0.05242239832878113
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,8192,1,0,0.062118399143219
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,2,16384,1,0,0.09817280173301697
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1,16384,1,0,0.07027199864387512
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1,16384,1,0,0.11156799793243408
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,1,1,0,0.042124798893928526
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,4,1,0,0.0415583997964859
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,8,1,0,0.04211199879646301
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,16,1,0,0.04166400134563446
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,32,1,0,0.0437855988740921
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,64,1,0,0.04189119935035705
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,128,1,0,0.04213759899139404
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,256,1,0,0.0437855988740921
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,512,1,0,0.04564160108566284
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,1024,1,0,0.046172800660133365
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,2048,1,0,0.051849597692489625
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,4096,1,0,0.05594559907913208
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,8192,1,0,0.06225600242614746
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,2,16384,1,0,0.07032960057258605
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,1,1,0,0.0419295996427536
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,4,1,0,0.039987200498580934
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,8,1,0,0.04157440066337585
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,16,1,0,0.04166080057621002
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,32,1,0,0.04546239972114563
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,64,1,0,0.04154880046844482
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,128,1,0,0.04182080030441284
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,256,1,0,0.04365119934082031
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,512,1,0,0.043510401248931886
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,1024,1,0,0.0475488007068634
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,2048,1,0,0.0526528000831604
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,4096,1,0,0.05927039980888367
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,1,1,0,0.07858240008354186
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,8192,1,0,0.06436799764633179
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,4,1,0,0.07843520045280457
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,8,1,0,0.07996799945831298
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,4,16384,1,0,0.0897055983543396
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,16,1,0,0.07929279804229736
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,32,1,0,0.08729599714279175
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,64,1,0,0.08715199828147888
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,128,1,0,0.08555520176887513
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,256,1,0,0.08543999791145325
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,512,1,0,0.08457599878311158
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,1,1,0,0.041705599427223204
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,1024,1,0,0.08865280151367187
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,4,1,0,0.04043839871883392
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,2048,1,0,0.09833599925041199
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,8,1,0,0.042115199565887454
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,4096,1,0,0.09759359955787658
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,8192,1,0,0.1033471941947937
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,16,1,0,0.039868798851966855
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,32,1,0,0.043881601095199584
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,2,16384,1,0,0.12055360078811646
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,64,1,0,0.04299519956111908
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,128,1,0,0.04398080110549927
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,256,1,0,0.04216319918632507
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,512,1,0,0.04695680141448975
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,1024,1,0,0.052121597528457644
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,2048,1,0,0.05614079833030701
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,4096,1,0,0.059036797285079955
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,8192,1,0,0.0847648024559021
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,8,16384,1,0,0.12492159605026246
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,1,1,0,0.03959679901599884
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,4,1,0,0.03957119882106781
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,8,1,0,0.039443200826644896
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,16,1,0,0.04149760007858276
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,32,1,0,0.04348799884319306
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,64,1,0,0.04164159893989563
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,128,1,0,0.04167360067367554
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,256,1,0,0.04564799964427948
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,1,1,0,0.0597760021686554
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,4,1,0,0.05702080130577088
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,512,1,0,0.051795202493667605
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,8,1,0,0.057920002937316896
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,1024,1,0,0.05376960039138794
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,16,1,0,0.05663359761238098
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,2048,1,0,0.05655679702758789
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,32,1,0,0.07017279863357544
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,64,1,0,0.06399359703063964
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,4096,1,0,0.08367679715156555
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,128,1,0,0.0637440025806427
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,256,1,0,0.06260799765586852
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,8192,1,0,0.12154239416122437
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,512,1,0,0.062319999933242796
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,1024,1,0,0.06809920072555542
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,2048,1,0,0.07690560221672058
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,4096,1,0,0.08154240250587463
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,8192,1,0,0.09100800156593322
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,1,1,0,0.0809119999408722
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,4,1,0,0.07945280075073242
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,8,1,0,0.0803264021873474
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,16,1,0,0.08079680204391479
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,32,1,0,0.0930176019668579
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,64,1,0,0.08828480243682861
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,128,1,0,0.08706560134887695
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,256,1,0,0.0884768009185791
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,512,1,0,0.09270079731941223
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,1024,1,0,0.0981760025024414
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,2048,1,0,0.09967679977416992
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,4096,1,0,0.10024640560150147
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,8192,1,0,0.11934720277786255
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,1,1,0,0.043321600556373595
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,4,1,0,0.043945598602294925
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,8,1,0,0.0438944011926651
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,16,1,0,0.0436928004026413
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,32,1,0,0.04813440144062042
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,64,1,0,0.048044800758361816
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,128,1,0,0.04616959989070892
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,256,1,0,0.04588159918785095
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,512,1,0,0.04965440034866333
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,1024,1,0,0.05411199927330017
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,2048,1,0,0.054092800617218016
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,4096,1,0,0.05803520083427429
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,8192,1,0,0.06683520078659058
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1,16384,1,0,0.07645760178565979
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,1,1,0,0.04406079947948456
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,4,1,0,0.0440447986125946
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,8,1,0,0.045664000511169436
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,16,1,0,0.04381439983844757
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,32,1,0,0.047849598526954654
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,64,1,0,0.047542399168014525
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,128,1,0,0.047056001424789426
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,256,1,0,0.04764800071716309
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,512,1,0,0.049439999461174014
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,1024,1,0,0.04782719910144806
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,2048,1,0,0.05448319911956787
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,4096,1,0,0.05937280058860779
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,8192,1,0,0.06817280054092408
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,2,16384,1,0,0.07955840229988098
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,1,1,0,0.04582720100879669
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,4,1,0,0.04383040070533752
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,8,1,0,0.04535999894142151
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,16,1,0,0.04527679979801178
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,32,1,0,0.051948797702789304
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,64,1,0,0.04767679870128631
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,128,1,0,0.04767360091209412
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,256,1,0,0.04772160053253174
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,512,1,0,0.04776639938354492
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,1024,1,0,0.05173119902610779
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,2048,1,0,0.057715201377868654
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,4096,1,0,0.06389759778976441
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,8192,1,0,0.07423359751701356
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,4,16384,1,0,0.0973695993423462
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,1,1,0,0.04408319890499115
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,4,1,0,0.04340159893035889
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,8,1,0,0.04511359930038452
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,16,1,0,0.04351679980754852
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,32,1,0,0.049670401215553286
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,64,1,0,0.047654399275779726
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,128,1,0,0.04776319861412048
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,256,1,0,0.04590399861335755
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,512,1,0,0.05137280225753784
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,1024,1,0,0.05754240155220032
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,2048,1,0,0.06020799875259399
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,4096,1,0,0.06717119812965393
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,8192,1,0,0.09242879748344421
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,8,16384,1,0,0.12869759798049926
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,1,1,0,0.04167360067367554
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,4,1,0,0.04353919923305512
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,8,1,0,0.04331200122833252
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,16,1,0,0.04376319944858551
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,32,1,0,0.048147198557853696
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,64,1,0,0.04574399888515472
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,128,1,0,0.04758400022983551
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,256,1,0,0.051945602893829344
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,512,1,0,0.05788480043411255
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,1024,1,0,0.06004480123519897
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,2048,1,0,0.06791999936103821
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,4096,1,0,0.09007359743118286
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,8192,1,0,0.12538880109786987
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,4,16384,1,0,0.10733120441436768
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,4,16384,1,0,0.1535423994064331
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,1,1,0,0.05424320101737976
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,4,1,0,0.05385919809341431
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,8,1,0,0.05597119927406311
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,16,1,0,0.053939199447631835
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,32,1,0,0.0662335991859436
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,64,1,0,0.06413760185241699
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,128,1,0,0.06362239718437195
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,256,1,0,0.06191040277481079
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,512,1,0,0.0679423987865448
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,1024,1,0,0.074755197763443
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,2048,1,0,0.0793727993965149
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,4096,1,0,0.08277119994163513
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,8192,1,0,0.1006432056427002
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,1,1,0,0.0772704005241394
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,4,1,0,0.0765824019908905
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,8,1,0,0.07882239818572997
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,16,1,0,0.07657600045204163
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,32,1,0,0.08991039991378784
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,64,1,0,0.08622080087661743
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,128,1,0,0.08700799942016602
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,256,1,0,0.09092159867286682
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,512,1,0,0.09679679870605469
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,1024,1,0,0.10099200010299683
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,2048,1,0,0.10001920461654663
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,4096,1,0,0.115174400806427
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,8192,1,0,0.15198400020599365
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,8,16384,1,0,0.13838080167770386
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,8,16384,1,0,0.23372159004211426
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,1,1,0,0.05384960174560547
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,4,1,0,0.056086397171020506
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,8,1,0,0.055199998617172244
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,16,1,0,0.05578879714012146
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,32,1,0,0.06816319823265075
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,64,1,0,0.0642304003238678
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,128,1,0,0.06615679860115051
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,256,1,0,0.07041280269622803
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,512,1,0,0.08064320087432861
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,1024,1,0,0.0825984001159668
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,2048,1,0,0.08428159952163697
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,4096,1,0,0.10206079483032227
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,8192,1,0,0.1383039951324463
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,1,1,0,0.07820159792900086
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,4,1,0,0.07865599989891052
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,8,1,0,0.07842239737510681
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,16,1,0,0.0791808009147644
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,32,1,0,0.09222720265388488
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,64,1,0,0.09141119718551635
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,128,1,0,0.09733440279960633
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,256,1,0,0.10126719474792481
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,512,1,0,0.10351999998092651
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,1024,1,0,0.10792959928512573
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,2048,1,0,0.12742079496383668
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,4096,1,0,0.16237759590148926
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,8192,1,0,0.2872191905975342
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,16,16384,1,0,0.20313599109649658
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,16,16384,1,0,0.20710399150848388
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,16,16384,1,0,0.2197887897491455
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,16,16384,1,0,0.47769598960876464
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,1,1,0,0.03971840143203735
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,4,1,0,0.041231998801231386
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,8,1,0,0.03937279880046844
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,16,1,0,0.03952000141143799
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,32,1,0,0.04365760087966919
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,64,1,0,0.04368639886379242
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,128,1,0,0.045664000511169436
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,256,1,0,0.05162879824638367
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,512,1,0,0.055667197704315184
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,1024,1,0,0.06402559876441956
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,2048,1,0,0.0884223997592926
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,32,4096,1,0,0.1292639970779419
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,1,1,0,0.043424001336097716
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,4,1,0,0.043459200859069826
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,8,1,0,0.04152320027351379
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,16,1,0,0.04353919923305512
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,32,1,0,0.049769601225852965
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,64,1,0,0.04987519979476929
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,128,1,0,0.05340800285339355
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,256,1,0,0.0598143994808197
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,512,1,0,0.06335359811782837
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,1024,1,0,0.07627840042114258
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,2048,1,0,0.09719039797782898
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,32,4096,1,0,0.135862398147583
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,1,1,0,0.05839999914169312
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,4,1,0,0.059494400024414064
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,8,1,0,0.058192002773284915
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,16,1,0,0.05942720174789429
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,32,1,0,0.07269759774208069
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,64,1,0,0.07401279807090759
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,128,1,0,0.07713919878005981
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,256,1,0,0.08455680012702942
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,512,1,0,0.08675839900970458
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,1024,1,0,0.09332799911499023
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,2048,1,0,0.11050879955291748
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,32,4096,1,0,0.1484992027282715
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,1,1,0,0.08042880296707153
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,4,1,0,0.08047999739646912
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,8,1,0,0.07876480221748353
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,16,1,0,0.08079360127449035
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,32,1,0,0.08136640191078186
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,64,1,0,0.0849727988243103
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,128,1,0,0.09008960127830505
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,256,1,0,0.10954879522323609
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,512,1,0,0.1127776026725769
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,1024,1,0,0.1310912013053894
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,2048,1,0,0.16718080043792724
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,32,4096,1,0,0.25765440464019773
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,1,1,0,0.03985919952392578
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,4,1,0,0.039552000164985654
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,8,1,0,0.0398144006729126
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,16,1,0,0.041740798950195314
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,32,1,0,0.03970879912376404
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,64,1,0,0.043699198961257936
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,128,1,0,0.04992319941520691
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,256,1,0,0.05599679946899414
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,512,1,0,0.06429439783096313
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,1024,1,0,0.09148160219192505
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,2048,1,0,0.1300287961959839
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,64,4096,1,0,0.2178368091583252
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,1,1,0,0.04353919923305512
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,4,1,0,0.043433600664138795
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,8,1,0,0.04377920031547546
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,16,1,0,0.04412479996681214
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,32,1,0,0.0438400000333786
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,64,1,0,0.04766719937324524
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,128,1,0,0.053744000196456906
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,256,1,0,0.0666208028793335
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,512,1,0,0.07850880026817322
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,1024,1,0,0.0998528003692627
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,2048,1,0,0.1377951979637146
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,64,4096,1,0,0.2236704111099243
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,1,1,0,0.08467199802398681
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,4,1,0,0.08447999954223633
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,8,1,0,0.08488960266113281
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,16,1,0,0.08522239923477173
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,32,1,0,0.08489279747009278
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,64,1,0,0.08755199909210205
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,128,1,0,0.0967903971672058
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,256,1,0,0.13573440313339233
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,512,1,0,0.146288001537323
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,1024,1,0,0.18083200454711915
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,2048,1,0,0.27292799949645996
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,64,4096,1,0,0.42801599502563475
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,1,1,0,0.06104959845542908
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,4,1,0,0.05986559987068176
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,8,1,0,0.06211199760437012
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,16,1,0,0.062438398599624634
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,32,1,0,0.06248639822006226
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,64,1,0,0.06607679724693298
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,128,1,0,0.07606719732284546
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,256,1,0,0.09185600280761719
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,512,1,0,0.09804480075836182
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,1024,1,0,0.11509120464324951
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,2048,1,0,0.15387519598007202
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,64,4096,1,0,0.2421567916870117
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,1,1,0,0.04162240028381348
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,4,1,0,0.041571199893951416
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,8,1,0,0.041596800088882446
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,16,1,0,0.04143039882183075
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,32,1,0,0.041468799114227295
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,64,1,0,0.0457152009010315
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,128,1,0,0.049728000164031984
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,256,1,0,0.07226880192756653
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,512,1,0,0.10078719854354859
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,1024,1,0,0.13693759441375733
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,128,2048,1,0,0.2298719882965088
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,1,1,0,0.047737601399421695
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,4,1,0,0.0474592000246048
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,8,1,0,0.04772160053253174
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,16,1,0,0.04685760140419006
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,32,1,0,0.049081599712371825
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,64,1,0,0.05394880175590515
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,128,1,0,0.061945599317550656
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,256,1,0,0.09220479726791382
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,512,1,0,0.11213760375976563
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,1024,1,0,0.1475711941719055
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,128,2048,1,0,0.2389535903930664
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,1,1,0,0.10555200576782227
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,4,1,0,0.10493439435958862
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,8,1,0,0.10673919916152955
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,16,1,0,0.10465919971466064
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,32,1,0,0.10533440113067627
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,64,1,0,0.11240639686584472
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,128,1,0,0.12488640546798706
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,256,1,0,0.1413375973701477
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,512,1,0,0.17577919960021973
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,1024,1,0,0.24462080001831055
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,1,1,0,0.07013440132141113
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,128,2048,1,0,0.38534400463104246
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,4,1,0,0.07063999772071838
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,8,1,0,0.07170559763908387
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,16,1,0,0.0691424012184143
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,32,1,0,0.07187520265579224
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,64,1,0,0.07471359968185425
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,128,1,0,0.08222399950027466
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,256,1,0,0.11331839561462402
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,512,1,0,0.13233599662780762
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,1024,1,0,0.16900800466537474
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,128,2048,1,0,0.2611999988555908
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,256,1,1,0,0.0498879998922348
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,256,4,1,0,0.04982720017433166
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,256,8,1,0,0.049744001030921935
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,256,16,1,0,0.051811200380325315
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,256,32,1,0,0.049718400835990904
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,256,64,1,0,0.05987200140953064
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,256,128,1,0,0.07219840288162231
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,256,256,1,0,0.10099519491195678
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,256,512,1,0,0.13576960563659668
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,256,1024,1,0,0.20522239208221435
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,256,1,1,0,0.060083198547363284
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,256,4,1,0,0.062224000692367554
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,256,8,1,0,0.06198400259017944
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,256,16,1,0,0.06403520107269287
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,256,32,1,0,0.06429759860038757
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,256,64,1,0,0.07455679774284363
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,256,128,1,0,0.0896448016166687
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,256,256,1,0,0.1090880036354065
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,256,512,1,0,0.14180799722671508
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,256,1024,1,0,0.21081919670104982
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,256,1,1,0,0.13564800024032592
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,256,4,1,0,0.13609600067138672
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,256,8,1,0,0.13604799509048462
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,256,16,1,0,0.1379263997077942
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,256,32,1,0,0.13660160303115845
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,256,64,1,0,0.15082240104675293
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,256,128,1,0,0.17429440021514891
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,256,256,1,0,0.21054720878601074
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,256,512,1,0,0.2777600049972534
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,256,1024,1,0,0.42015681266784666
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,256,1,1,0,0.08879039883613586
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,256,4,1,0,0.08855999708175659
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,256,8,1,0,0.08833919763565064
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,256,16,1,0,0.09031680226325989
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,256,32,1,0,0.08881919980049133
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,256,64,1,0,0.09560959935188293
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,256,128,1,0,0.10850880146026612
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,256,256,1,0,0.126803195476532
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,256,512,1,0,0.1604032039642334
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,256,1024,1,0,0.23043520450592042
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,512,1,1,0,0.0725055992603302
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,512,4,1,0,0.07247359752655029
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,512,8,1,0,0.07270399928092956
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,512,16,1,0,0.0741823971271515
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,512,32,1,0,0.07824959754943847
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,512,64,1,0,0.09454079866409301
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,512,128,1,0,0.12146559953689576
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,512,256,1,0,0.15677759647369385
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,512,512,1,0,0.22900159358978273
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,512,1,1,0,0.09574400186538697
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,512,4,1,0,0.095523202419281
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,512,8,1,0,0.09574080109596253
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,512,16,1,0,0.09677760004997253
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,512,32,1,0,0.09924160242080689
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,512,64,1,0,0.1092479944229126
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,512,128,1,0,0.1355679988861084
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,512,256,1,0,0.17011840343475343
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,512,512,1,0,0.2421056032180786
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,512,1,1,0,0.19455679655075073
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,512,4,1,0,0.19477759599685668
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,512,8,1,0,0.19449280500411986
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,512,16,1,0,0.19567999839782715
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,512,32,1,0,0.19741439819335938
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,512,64,1,0,0.2182368040084839
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,512,128,1,0,0.25857279300689695
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,512,256,1,0,0.3200000047683716
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,512,512,1,0,0.44341440200805665
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,512,1,1,0,0.12283840179443359
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,512,4,1,0,0.12297279834747314
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,512,8,1,0,0.12293119430541992
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,512,16,1,0,0.12347520589828491
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,512,32,1,0,0.12548160552978516
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,512,64,1,0,0.1377120018005371
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,512,128,1,0,0.1627519965171814
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,512,256,1,0,0.19832320213317872
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,512,512,1,0,0.2707808017730713
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1024,1,1,0,0.11371840238571167
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1024,4,1,0,0.11640959978103638
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1024,8,1,0,0.11577279567718506
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1024,16,1,0,0.11713919639587403
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1024,32,1,0,0.11939840316772461
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1024,64,1,0,0.14407999515533448
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1024,128,1,0,0.1832095980644226
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,16,1024,256,1,0,0.24735360145568847
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1024,1,1,0,0.13790080547332764
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1024,4,1,0,0.1398527979850769
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1024,8,1,0,0.1391487956047058
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1024,16,1,0,0.1407807946205139
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1024,32,1,0,0.1422816038131714
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1024,64,1,0,0.1674720048904419
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1024,128,1,0,0.20581119060516356
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,32,1024,256,1,0,0.2698080062866211
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,1,1,0,0.07881919741630554
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,4,1,0,0.08007360100746155
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,8,1,0,0.07907519936561584
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,16,1,0,0.07859839797019959
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,32,1,0,0.07904319763183594
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,64,1,0,0.07982400059700012
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,128,1,0,0.07872639894485474
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,256,1,0,0.08093119859695434
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,512,1,0,0.08484159708023072
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,1024,1,0,0.08496000170707703
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,2048,1,0,0.08776000142097473
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,4096,1,0,0.09466879963874816
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,8192,1,0,0.09336640238761902
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1,16384,1,0,0.10751359462738037
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,1,1,0,0.08109120130538941
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,4,1,0,0.07827839851379395
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,8,1,0,0.0800704002380371
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,16,1,0,0.07900480031967164
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,32,1,0,0.0782975971698761
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,64,1,0,0.07896639704704285
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,128,1,0,0.07951679825782776
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,256,1,0,0.08100159764289856
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,512,1,0,0.08426880240440368
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,1024,1,0,0.08496320247650146
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,2048,1,0,0.09324480295181274
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,4096,1,0,0.09401599764823913
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,8192,1,0,0.10120639801025391
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,2,16384,1,0,0.09490560293197632
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,1,1,0,0.07825599908828736
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,4,1,0,0.07934399843215942
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,8,1,0,0.07895680069923401
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,16,1,0,0.07742720246315002
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,32,1,0,0.07917760014533996
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,64,1,0,0.07794240117073059
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,128,1,0,0.07845759987831116
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,256,1,0,0.08126720190048217
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,512,1,0,0.08568000197410583
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,1024,1,0,0.09254080057144165
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,2048,1,0,0.09633600115776061
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,4096,1,0,0.08758400082588196
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,8192,1,0,0.09084799885749817
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,4,16384,1,0,0.09650239944458008
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,1,1,0,0.07968000173568726
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,4,1,0,0.07925440073013305
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,8,1,0,0.07837439775466919
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,16,1,0,0.0802944004535675
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,32,1,0,0.07863039970397949
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,64,1,0,0.07937600016593933
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,128,1,0,0.08042560219764709
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,256,1,0,0.08054400086402894
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,512,1,0,0.09268800020217896
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,1024,1,0,0.09499520063400269
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,2048,1,0,0.0846015989780426
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,4096,1,0,0.08884479999542236
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,8192,1,0,0.09438719749450683
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,8,16384,1,0,0.10798720121383668
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,1,1,0,0.08459519743919372
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,4,1,0,0.08326079845428466
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,8,1,0,0.08432959914207458
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,16,1,0,0.08446080088615418
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,32,1,0,0.08432639837265014
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,64,1,0,0.08432959914207458
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,128,1,0,0.08406400084495544
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,256,1,0,0.08840640187263489
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,512,1,0,0.09224640130996704
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,1024,1,0,0.08696320056915283
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,2048,1,0,0.08876799941062927
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,4096,1,0,0.09710080027580262
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,8192,1,0,0.11142079830169678
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,16,16384,1,0,0.13712639808654786
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,1,1,0,0.07925440073013305
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,4,1,0,0.0804319977760315
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,8,1,0,0.07910079956054687
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,16,1,0,0.07880319952964783
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,32,1,0,0.0807200014591217
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,64,1,0,0.07864639759063721
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,128,1,0,0.08072959780693054
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,256,1,0,0.08854399919509888
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,512,1,0,0.08780800104141236
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,1024,1,0,0.09375039935111999
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,2048,1,0,0.09934399724006653
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,32,4096,1,0,0.11455359458923339
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,1,1,0,0.0866047978401184
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,4,1,0,0.08522560000419617
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,8,1,0,0.0859008014202118
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,16,1,0,0.08648959994316101
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,32,1,0,0.08608639836311341
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,64,1,0,0.08683840036392212
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,128,1,0,0.08829439878463745
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,256,1,0,0.08872320055961609
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,512,1,0,0.09317759871482849
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,1024,1,0,0.09889280200004577
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,2048,1,0,0.11350719928741455
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,64,4096,1,0,0.14151040315628052
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,1,1,0,0.10463039875030518
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,4,1,0,0.10644799470901489
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,8,1,0,0.10523840188980102
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,16,1,0,0.10486079454421997
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,32,1,0,0.10505599975585937
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,64,1,0,0.1051360011100769
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,128,1,0,0.10900800228118897
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,256,1,0,0.11082559823989868
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,512,1,0,0.11886080503463745
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,1024,1,0,0.13369920253753662
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,128,2048,1,0,0.16254719495773315
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,256,1,1,0,0.1314687967300415
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,256,4,1,0,0.13213440179824829
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,256,8,1,0,0.13168959617614745
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,256,16,1,0,0.13262399435043334
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,256,32,1,0,0.13231040239334108
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,256,64,1,0,0.13234879970550537
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,256,128,1,0,0.1399359941482544
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,256,256,1,0,0.1471392035484314
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,256,512,1,0,0.1632863998413086
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,256,1024,1,0,0.1926144003868103
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,512,1,1,0,0.18909120559692383
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,512,4,1,0,0.18940479755401612
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,512,8,1,0,0.18981759548187255
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,512,16,1,0,0.18929280042648317
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,512,32,1,0,0.1887935996055603
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,512,64,1,0,0.19302079677581788
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,512,128,1,0,0.20241279602050782
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,512,256,1,0,0.2185663938522339
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,512,512,1,0,0.24600319862365722
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1024,1,1,0,0.32727999687194825
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1024,4,1,0,0.32931840419769287
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1024,8,1,0,0.32760000228881836
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1024,16,1,0,0.326963210105896
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1024,32,1,0,0.3284768104553223
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1024,64,1,0,0.3333791971206665
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1024,128,1,0,0.35327041149139404
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,128,1024,256,1,0,0.3844320058822632
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,1,1,0,0.05230399966239929
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,4,1,0,0.0525056004524231
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,8,1,0,0.05232639908790589
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,16,1,0,0.052035200595855716
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,32,1,0,0.053491199016571046
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,64,1,0,0.05310720205307007
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,128,1,0,0.051999998092651364
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,256,1,0,0.05589759945869446
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,512,1,0,0.05780479907989502
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,1024,1,0,0.05620800256729126
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,2048,1,0,0.060012799501419065
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,4096,1,0,0.060012799501419065
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,8192,1,0,0.06849279999732971
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1,16384,1,0,0.07696639895439147
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,1,1,0,0.05220479965209961
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,4,1,0,0.05380799770355225
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,8,1,0,0.052211201190948485
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,16,1,0,0.0529695987701416
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,32,1,0,0.05225920081138611
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,64,1,0,0.05284479856491089
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,128,1,0,0.05401279926300049
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,256,1,0,0.054153597354888915
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,512,1,0,0.05583680272102356
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,1024,1,0,0.05800319910049438
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,2048,1,0,0.0563103973865509
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,4096,1,0,0.06406400203704835
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,8192,1,0,0.0782144010066986
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,2,16384,1,0,0.08572480082511902
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,1,1,0,0.05425919890403748
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,4,1,0,0.052134400606155394
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,8,1,0,0.051881599426269534
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,16,1,0,0.052086400985717776
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,32,1,0,0.05210559964179993
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,64,1,0,0.0521888017654419
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,128,1,0,0.05245440006256104
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,256,1,0,0.055667197704315184
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,512,1,0,0.056841599941253665
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,1024,1,0,0.05635200142860412
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,2048,1,0,0.06208639740943909
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,4096,1,0,0.0782368004322052
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,8192,1,0,0.08253440260887146
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,4,16384,1,0,0.08675519824028015
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,1,1,0,0.05264639854431152
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,4,1,0,0.05139840245246887
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,8,1,0,0.052508801221847534
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,16,1,0,0.05165119767189026
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,32,1,0,0.05208960175514221
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,64,1,0,0.05223039984703064
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,128,1,0,0.05605440139770508
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,256,1,0,0.056329602003097536
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,512,1,0,0.060140800476074216
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,1024,1,0,0.06772480010986329
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,2048,1,0,0.07577599883079529
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,4096,1,0,0.07453759908676147
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,8192,1,0,0.08082879781723022
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,8,16384,1,0,0.09488639831542969
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,1,1,0,0.05462719798088074
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,4,1,0,0.05403519868850708
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,8,1,0,0.055199998617172244
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,16,1,0,0.054067200422286986
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,32,1,0,0.053699201345443724
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,64,1,0,0.0580672025680542
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,128,1,0,0.05785279870033264
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,256,1,0,0.06111680269241333
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,512,1,0,0.06839680075645446
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,1024,1,0,0.07887679934501649
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,2048,1,0,0.07396159768104553
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,4096,1,0,0.08104320168495179
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,8192,1,0,0.09488959908485413
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,16,16384,1,0,0.12089279890060425
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,1,1,0,0.06526079773902893
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,4,1,0,0.06440960168838501
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,8,1,0,0.06452800035476684
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,16,1,0,0.06427519917488098
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,32,1,0,0.06545600295066833
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,64,1,0,0.0655680000782013
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,128,1,0,0.06628159880638122
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,256,1,0,0.07239680290222168
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,512,1,0,0.07485759854316712
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,1024,1,0,0.0754144012928009
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,2048,1,0,0.08207039833068848
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,32,4096,1,0,0.09633920192718506
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,1,1,0,0.06466879844665527
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,4,1,0,0.06212480068206787
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,8,1,0,0.0635807991027832
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,16,1,0,0.06410239934921265
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,32,1,0,0.06316159963607788
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,64,1,0,0.06413760185241699
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,128,1,0,0.06558079719543457
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,256,1,0,0.07443519830703735
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,512,1,0,0.07870079874992371
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,1024,1,0,0.08544960021972656
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,2048,1,0,0.10042879581451417
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,64,4096,1,0,0.129312002658844
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,1,1,0,0.0723904013633728
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,4,1,0,0.07408959865570068
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,8,1,0,0.07280960083007812
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,16,1,0,0.07228479981422424
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,32,1,0,0.07460799813270569
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,64,1,0,0.07272319793701172
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,128,1,0,0.07696639895439147
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,256,1,0,0.08049920201301575
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,512,1,0,0.08852800130844116
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,1024,1,0,0.10334399938583375
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,128,2048,1,0,0.1313472032546997
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,256,1,1,0,0.09231680035591125
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,256,4,1,0,0.09190080165863038
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,256,8,1,0,0.091430401802063
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,256,16,1,0,0.0920799970626831
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,256,32,1,0,0.09192000031471252
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,256,64,1,0,0.09259200096130371
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,256,128,1,0,0.09921919703483581
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,256,256,1,0,0.10540800094604492
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,256,512,1,0,0.12303680181503296
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,256,1024,1,0,0.15421760082244873
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,512,1,1,0,0.12493760585784912
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,512,4,1,0,0.12666879892349242
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,512,8,1,0,0.12583359479904174
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,512,16,1,0,0.1257599949836731
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,512,32,1,0,0.1268064022064209
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,512,64,1,0,0.13078080415725707
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,512,128,1,0,0.1412992000579834
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,512,256,1,0,0.1558527946472168
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,512,512,1,0,0.18754559755325317
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1024,1,1,0,0.19876480102539062
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1024,4,1,0,0.19959039688110353
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1024,8,1,0,0.1993888020515442
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1024,16,1,0,0.19805439710617065
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1024,32,1,0,0.20002880096435546
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1024,64,1,0,0.2081439971923828
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1024,128,1,0,0.22713921070098878
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,64,1024,256,1,0,0.2534143924713135
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1024,1,1,0,0.3429631948471069
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1024,4,1,0,0.3460576057434082
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1024,8,1,0,0.3461472034454346
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1024,16,1,0,0.34844160079956055
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1024,32,1,0,0.35003199577331545
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1024,64,1,0,0.39203839302062987
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1024,1,1,0,0.19212160110473633
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1024,128,1,0,0.46788802146911623
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1024,4,1,0,0.19492800235748292
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1024,8,1,0,0.19417279958724976
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1024,16,1,0,0.19483840465545654
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1024,32,1,0,0.19731520414352416
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1024,64,1,0,0.22014079093933106
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1024,128,1,0,0.25874879360198977
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,128,1024,256,1,0,0.5911200046539307
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,flashinfer,fp8_block,fp8,64,1024,256,1,0,0.32282559871673583
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,1,1,0,0.04184640049934387
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,4,1,0,0.043907201290130614
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,8,1,0,0.04366079866886139
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,16,1,0,0.04357759952545166
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,32,1,0,0.043593600392341614
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,64,1,0,0.04364160001277924
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,128,1,0,0.04355520009994507
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,256,1,0,0.04573439955711365
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,512,1,0,0.04770239889621734
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,1024,1,0,0.047654399275779726
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,2048,1,0,0.04782080054283142
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,4096,1,0,0.047619199752807616
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,8192,1,0,0.05308799743652344
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1,16384,1,0,0.05751360058784485
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,1,1,0,0.04384639859199524
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,4,1,0,0.046003198623657225
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,8,1,0,0.04354240000247955
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,16,1,0,0.04435519874095917
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,32,1,0,0.043663999438285826
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,64,1,0,0.04357120096683502
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,128,1,0,0.043935999274253845
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,256,1,0,0.045603200793266296
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,512,1,0,0.04776639938354492
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,1024,1,0,0.047686401009559634
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,2048,1,0,0.04766719937324524
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,4096,1,0,0.052112001180648806
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,8192,1,0,0.05589119791984558
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,2,16384,1,0,0.06631360054016114
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,1,1,0,0.043663999438285826
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,4,1,0,0.04153920114040375
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,8,1,0,0.04532159864902496
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,16,1,0,0.04357759952545166
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,32,1,0,0.04176000058650971
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,64,1,0,0.043593600392341614
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,128,1,0,0.04340159893035889
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,256,1,0,0.04556800127029419
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,512,1,0,0.04767679870128631
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,1024,1,0,0.047865599393844604
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,2048,1,0,0.04759039878845215
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,4096,1,0,0.05758399963378906
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,8192,1,0,0.05994240045547485
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,4,16384,1,0,0.08070080280303955
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,1,1,0,0.04348160028457641
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,4,1,0,0.04559040069580078
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,8,1,0,0.04360640048980713
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,16,1,0,0.04315840005874634
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,32,1,0,0.04560959935188293
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,64,1,0,0.043536001443862916
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,128,1,0,0.043510401248931886
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,256,1,0,0.04707199931144714
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,512,1,0,0.047670400142669676
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,1024,1,0,0.04961600005626678
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,2048,1,0,0.052083200216293334
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,4096,1,0,0.06216959953308106
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,8192,1,0,0.07835839986801148
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,8,16384,1,0,0.09842560291290284
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,1,1,0,0.043619200587272644
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,4,1,0,0.043663999438285826
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,8,1,0,0.04567680060863495
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,16,1,0,0.043510401248931886
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,32,1,0,0.045552000403404236
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,64,1,0,0.043670400977134705
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,128,1,0,0.04359680116176605
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,256,1,0,0.047651201486587524
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,512,1,0,0.051862400770187375
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,1024,1,0,0.058220797777175905
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,2048,1,0,0.06813120245933532
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,4096,1,0,0.08097599744796753
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,8192,1,0,0.10187840461730957
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,16,16384,1,0,0.1420192003250122
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,1,1,0,0.04583680033683777
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,4,1,0,0.04564160108566284
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,8,1,0,0.04562239944934845
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,16,1,0,0.04760960042476654
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,32,1,0,0.04564799964427948
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,64,1,0,0.045763200521469115
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,128,1,0,0.04775039851665497
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,256,1,0,0.053148800134658815
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,512,1,0,0.06517760157585144
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,1024,1,0,0.07015039920806884
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,2048,1,0,0.0821183979511261
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,32,4096,1,0,0.10397759675979615
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,1,1,0,0.04756160080432892
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,4,1,0,0.047603198885917665
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,8,1,0,0.04768959879875183
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,16,1,0,0.04767679870128631
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,32,1,0,0.04763840138912201
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,64,1,0,0.04959999918937683
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,128,1,0,0.04952960014343262
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,256,1,0,0.058633601665496825
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,512,1,0,0.06804479956626892
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,1024,1,0,0.08170239925384522
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,2048,1,0,0.09975039958953857
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,64,4096,1,0,0.13808319568634034
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,1,1,0,0.05952320098876953
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,4,1,0,0.05819839835166931
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,8,1,0,0.059987199306488034
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,16,1,0,0.05791680216789245
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,32,1,0,0.059945601224899295
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,64,1,0,0.06387839913368225
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,128,1,0,0.06732800006866455
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,256,1,0,0.07958400249481201
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,512,1,0,0.09079999923706054
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,1024,1,0,0.1094208002090454
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,128,2048,1,0,0.14910399913787842
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,256,1,1,0,0.07991679906845092
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,256,4,1,0,0.08041599988937378
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,256,8,1,0,0.08033279776573181
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,256,16,1,0,0.0807424008846283
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,256,32,1,0,0.08041599988937378
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,256,64,1,0,0.08436480164527893
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,256,128,1,0,0.08866559863090515
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,256,256,1,0,0.10493439435958862
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,256,512,1,0,0.12495360374450684
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,256,1024,1,0,0.16450239419937135
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,512,1,1,0,0.11714880466461182
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,512,4,1,0,0.11747839450836181
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,512,8,1,0,0.11823999881744385
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,512,16,1,0,0.11719679832458496
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,512,32,1,0,0.1174239993095398
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,512,64,1,0,0.11833280324935913
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,512,128,1,0,0.12084159851074219
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,512,256,1,0,0.1511263966560364
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,512,512,1,0,0.18780800104141235
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1024,1,1,0,0.17662080526351928
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1024,4,1,0,0.17711999416351318
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1024,8,1,0,0.17788480520248412
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1024,16,1,0,0.17699520587921141
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1024,32,1,0,0.17845120429992675
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1024,64,1,0,0.18227200508117675
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1024,128,1,0,0.1859007954597473
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,32,1024,256,1,0,0.24600958824157715
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,1,1,0,0.039654400944709775
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,4,1,0,0.039664000272750854
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,8,1,0,0.04146240055561066
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,16,1,0,0.03964479863643646
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,32,1,0,0.03951359987258911
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,64,1,0,0.04023999869823456
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,128,1,0,0.03977279961109161
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,256,1,0,0.04166719913482666
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,512,1,0,0.043673598766326906
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,1024,1,0,0.04551360011100769
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,2048,1,0,0.04564160108566284
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,4096,1,0,0.04416959881782532
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,8192,1,0,0.047651201486587524
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1,16384,1,0,0.05002880096435547
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,1,1,0,0.04151679873466492
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,4,1,0,0.04157440066337585
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,8,1,0,0.039392000436782836
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,16,1,0,0.041577601432800294
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,32,1,0,0.04100799858570099
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,64,1,0,0.03951039910316467
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,128,1,0,0.04148800075054169
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,256,1,0,0.04360640048980713
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,512,1,0,0.04353919923305512
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,1024,1,0,0.04347519874572754
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,2048,1,0,0.0435808002948761
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,4096,1,0,0.04549440145492554
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,8192,1,0,0.049609598517417905
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,2,16384,1,0,0.051772797107696535
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,1,1,0,0.041494399309158325
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,4,1,0,0.03933440148830414
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,8,1,0,0.04149760007858276
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,16,1,0,0.041465601325035094
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,32,1,0,0.03943679928779602
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,64,1,0,0.04157440066337585
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,128,1,0,0.040003201365470885
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,256,1,0,0.043424001336097716
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,512,1,0,0.045238399505615236
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,1024,1,0,0.04343999922275543
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,2048,1,0,0.04552319943904877
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,4096,1,0,0.04959999918937683
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,8192,1,0,0.0515999972820282
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,4,16384,1,0,0.06687999963760376
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,1,1,0,0.03942080140113831
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,4,1,0,0.041440001130104064
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,8,1,0,0.04125120043754578
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,16,1,0,0.03935360014438629
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,32,1,0,0.04162560105323791
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,64,1,0,0.03938240110874176
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,128,1,0,0.03953920006752014
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,256,1,0,0.043699198961257936
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,512,1,0,0.04344319999217987
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,1024,1,0,0.045398399233818054
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,2048,1,0,0.04546880125999451
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,4096,1,0,0.04976319968700409
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,8192,1,0,0.06203839778900146
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,8,16384,1,0,0.08446400165557862
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,1,1,0,0.03943040072917938
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,4,1,0,0.04075199961662292
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,8,1,0,0.0414112001657486
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,16,1,0,0.03946239948272705
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,32,1,0,0.041465601325035094
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,64,1,0,0.04127680063247681
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,128,1,0,0.03923200070858002
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,256,1,0,0.043484801054000856
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,512,1,0,0.04505600035190582
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,1024,1,0,0.045510399341583255
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,2048,1,0,0.04968000054359436
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,4096,1,0,0.0624064028263092
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,8192,1,0,0.08428159952163697
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,16,16384,1,0,0.11047040224075318
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,1,1,0,0.041596800088882446
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,4,1,0,0.0414112001657486
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,8,1,0,0.04140160083770752
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,16,1,0,0.041868799924850465
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,32,1,0,0.041500800848007204
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,64,1,0,0.04126079976558685
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,128,1,0,0.04359039962291718
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,256,1,0,0.04351679980754852
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,512,1,0,0.047100800275802615
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,1024,1,0,0.04955199956893921
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,2048,1,0,0.0639360010623932
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,32,4096,1,0,0.08457279801368714
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,1,1,0,0.04380159974098206
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,4,1,0,0.041510400176048276
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,8,1,0,0.04355199933052063
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,16,1,0,0.04345600008964538
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,32,1,0,0.04346239864826203
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,64,1,0,0.043350398540496826
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,128,1,0,0.04349119961261749
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,256,1,0,0.045542401075363156
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,512,1,0,0.051609599590301515
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,1024,1,0,0.06875519752502442
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,2048,1,0,0.08729599714279175
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,64,4096,1,0,0.11476160287857055
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,1,1,0,0.04553920030593872
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,4,1,0,0.0456063985824585
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,8,1,0,0.04778560101985931
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,16,1,0,0.04357120096683502
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,32,1,0,0.04756160080432892
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,64,1,0,0.04552319943904877
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,128,1,0,0.0455487996339798
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,256,1,0,0.0515936017036438
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,512,1,0,0.06901760101318359
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,1024,1,0,0.08629440069198609
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,128,2048,1,0,0.11169919967651368
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,256,1,1,0,0.05375679731369019
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,256,4,1,0,0.05383679866790771
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,256,8,1,0,0.05397440195083618
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,256,16,1,0,0.0537056028842926
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,256,32,1,0,0.05553600192070007
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,256,64,1,0,0.05416319966316223
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,256,128,1,0,0.06458240151405334
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,256,256,1,0,0.0813535988330841
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,256,512,1,0,0.09454079866409301
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,256,1024,1,0,0.1238368034362793
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,512,1,1,0,0.08488960266113281
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,512,4,1,0,0.08590400218963623
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,512,8,1,0,0.08651520013809204
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,512,16,1,0,0.08534719944000244
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,512,32,1,0,0.0858847975730896
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,512,64,1,0,0.09101439714431762
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,512,128,1,0,0.0933184027671814
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,512,256,1,0,0.1160159945487976
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,512,512,1,0,0.14457600116729735
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1024,1,1,0,0.12022720575332642
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1024,4,1,0,0.11955840587615967
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1024,8,1,0,0.12122880220413208
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1024,16,1,0,0.11897920370101929
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1024,32,1,0,0.1207360029220581
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1024,64,1,0,0.1281440019607544
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1024,128,1,0,0.13287680149078368
SGLang,0.5.8.post1,NVIDIA GB200,mla_generation,trtllm_mla,fp8_block,fp8,16,1024,256,1,0,0.17242560386657715
