framework,version,device,op_name,kernel_source,bmm_dtype,num_tokens,num_heads,latency
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1,128,0.007491199672222138
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1,64,0.005532800033688545
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1,32,0.005110400170087815
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1,16,0.004668800160288811
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1,128,0.010435199737548828
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1,8,0.004704000055789947
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1,8,0.008540800213813782
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1,4,0.004825599864125252
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1,64,0.00857279971241951
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1,4,0.008371199667453765
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1,2,0.005587200075387955
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1,1,0.004639999940991402
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1,2,0.008483199775218964
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2,128,0.008393599838018417
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1,1,0.008643200248479843
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2,64,0.006547199934720993
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1,32,0.008627200126647949
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2,32,0.005177599936723709
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2,128,0.010809600353240967
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2,16,0.004758400097489357
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2,16,0.008566399663686752
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2,8,0.004774399846792221
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2,8,0.008483199775218964
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2,4,0.0047807998955249785
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2,4,0.008476799726486206
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2,2,0.0047680001705884935
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2,2,0.008403199911117553
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2,1,0.004796800017356872
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2,1,0.008537600189447403
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4,128,0.008425600081682205
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4,128,0.010815999656915664
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4,64,0.006060799956321717
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4,64,0.00843520015478134
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1,16,0.008883199840784072
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4,32,0.008816000074148178
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4,16,0.004540799930691719
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4,16,0.008748800307512284
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4,8,0.004796800017356872
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4,8,0.008726400136947633
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4,4,0.004783999919891357
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4,4,0.008771199733018875
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4,2,0.0046016000211238865
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4,2,0.00862400010228157
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4,1,0.004598399996757508
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4,1,0.006649599969387054
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8,128,0.006707199662923813
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8,128,0.010822399705648422
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8,64,0.004838399961590767
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2,64,0.008720000088214875
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8,32,0.00456320010125637
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8,32,0.008656000345945358
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8,16,0.0048096001148223875
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8,16,0.00857279971241951
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8,8,0.004540799930691719
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8,8,0.008761599659919739
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8,4,0.0047231998294591905
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8,4,0.008582399785518646
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8,2,0.004575999826192856
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8,2,0.008588799834251403
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2,32,0.008544000238180161
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8,1,0.004604800045490265
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,16,128,0.0068000003695487974
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4,32,0.0048767998814582825
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,16,64,0.006387200206518173
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,16,128,0.012895999848842621
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,16,32,0.0047680001705884935
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,16,64,0.010255999863147736
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,16,32,0.008752000331878663
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,16,16,0.004742399975657463
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,16,8,0.005049600079655647
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,16,8,0.008483199775218964
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,16,4,0.004726399853825569
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,16,4,0.008585599809885025
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8,64,0.008684799820184708
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,16,2,0.004812800139188766
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,16,1,0.004793599992990494
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,16,2,0.008703999966382981
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,16,1,0.008646400272846222
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,32,128,0.008428800106048583
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,32,128,0.01584320068359375
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,32,64,0.006361600011587143
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,32,64,0.010976000130176545
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,32,32,0.008787199854850769
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,32,16,0.004758400097489357
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,32,16,0.008975999802350998
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,32,8,0.0047391999512910845
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,32,8,0.008508799970149994
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,32,4,0.004812800139188766
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8,1,0.008422400057315826
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,32,2,0.005087999999523163
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,32,2,0.008716800063848496
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,32,1,0.004851200059056282
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,32,1,0.008595199882984161
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,48,128,0.008752000331878663
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,48,128,0.019040000438690186
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,48,64,0.007919999957084655
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,48,64,0.012665599584579468
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,48,32,0.00639680027961731
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,48,32,0.010543999820947647
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,48,16,0.005491200089454651
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,16,16,0.008796799927949905
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,48,16,0.008764799684286118
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,48,8,0.004678399860858917
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,48,8,0.00875839963555336
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,48,4,0.0046431999653577805
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,48,4,0.00873280018568039
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,48,2,0.0046847999095916745
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,48,2,0.00868159979581833
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,48,1,0.004931199923157692
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,32,32,0.00456320010125637
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,48,1,0.008575999736785888
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,64,128,0.02112320065498352
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,64,64,0.006681600213050842
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,64,64,0.015024000406265258
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,64,32,0.00859839990735054
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,32,4,0.008377599716186523
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,64,32,0.010924799740314484
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,64,16,0.004960000142455101
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,64,16,0.008985599875450135
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,64,8,0.005241600051522255
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,64,4,0.005023999884724617
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,64,4,0.00841279998421669
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,64,2,0.004800000041723251
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,64,2,0.00867839977145195
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,64,1,0.005356800183653832
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,64,1,0.008550400286912918
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,80,128,0.008419200032949447
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,80,128,0.023187200725078582
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,80,64,0.006639999896287918
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,80,64,0.016806399822235106
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,80,32,0.00655680000782013
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,80,32,0.012646399438381195
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,80,16,0.006358399987220764
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,80,16,0.010633599758148194
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,80,8,0.005948799848556519
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,80,8,0.009020800143480301
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,64,128,0.006857600063085556
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,80,4,0.004979199916124344
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,80,4,0.00857279971241951
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,80,2,0.008656000345945358
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,80,1,0.0048767998814582825
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,80,1,0.008550400286912918
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,96,128,0.00841279998421669
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,96,128,0.025417599081993102
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,96,64,0.006694400310516357
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,64,8,0.008668799698352814
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,96,64,0.017088000476360322
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,96,32,0.006652799993753433
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,96,32,0.012751999497413635
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,96,16,0.004816000163555145
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,96,16,0.01067200005054474
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,96,8,0.0049056001007556915
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,96,4,0.004889599978923798
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,96,4,0.008924800157546996
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,96,2,0.004944000020623207
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,96,2,0.008899199962615966
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,96,1,0.004998400062322617
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,96,1,0.008742400258779526
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,128,128,0.008780799806118011
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,128,128,0.031711998581886294
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,128,64,0.006752000004053116
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,128,32,0.006860800087451935
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,128,64,0.021772800385951994
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,128,32,0.01499519944190979
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,128,16,0.00820479989051819
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,128,16,0.010857599973678588
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,128,8,0.0075392000377178196
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,128,8,0.008793599903583527
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,80,2,0.004812800139188766
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,128,4,0.008761599659919739
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,128,2,0.0049727998673915865
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,128,1,0.004915200173854828
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,128,1,0.008479999750852585
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,160,128,0.008928000181913375
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,160,128,0.037363201379776
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,160,64,0.007056000083684922
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,160,64,0.023151999711990355
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,160,32,0.0069472000002861025
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,160,32,0.016726399958133697
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,160,16,0.0063231997191905975
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,160,16,0.012899200618267059
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,160,8,0.0062720000743865965
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,160,8,0.010688000172376633
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,160,4,0.005241600051522255
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,160,4,0.008687999844551087
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,160,2,0.004950400069355964
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,160,2,0.008585599809885025
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,160,1,0.004828799888491631
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,160,1,0.00868159979581833
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,192,128,0.010828799754381179
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,192,128,0.04162240028381348
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,192,64,0.00846719965338707
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,192,64,0.025180798768997193
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,128,4,0.007203199714422226
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,192,32,0.006876800209283829
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,192,16,0.006364800035953522
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,192,32,0.017833599448204042
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,192,16,0.012671999633312225
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,192,8,0.010633599758148194
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,192,4,0.004732799902558327
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,96,8,0.008956799656152726
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,192,4,0.008879999816417693
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,192,2,0.005529600009322166
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,192,2,0.008537600189447403
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,192,1,0.004867200180888176
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,192,1,0.008585599809885025
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,256,128,0.011017599701881408
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,256,128,0.05357760190963745
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,256,64,0.00836160033941269
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,256,64,0.030969598889350893
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,256,32,0.006726399809122085
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,256,32,0.019228799641132353
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,256,16,0.0060384001582860945
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,256,16,0.014815999567508698
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,256,8,0.005644800141453743
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,256,8,0.010796800255775452
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,256,4,0.0050976000726222995
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,256,4,0.008902399986982345
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,256,2,0.005350400134921074
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,128,2,0.008873599767684936
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,256,2,0.008748800307512284
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,256,1,0.004796800017356872
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,256,1,0.00843840017914772
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,320,128,0.012796799838542938
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,320,128,0.06437119841575623
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,320,64,0.009046400338411332
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,320,64,0.03737280070781708
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,320,32,0.006876800209283829
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,320,16,0.006828799843788147
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,320,16,0.014899200201034546
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,320,8,0.00560000017285347
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,320,8,0.010918399691581726
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,320,4,0.005158400163054466
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,320,4,0.010127999633550645
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,320,2,0.005142400041222572
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,320,2,0.008844800293445587
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,320,1,0.0050592001527547835
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,320,1,0.00897279977798462
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,384,128,0.015027199685573579
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,384,128,0.07648959755897522
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,384,64,0.010815999656915664
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,384,64,0.04164159893989563
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,384,32,0.008822400122880936
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,384,32,0.025183999538421632
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,192,8,0.009459199756383896
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,384,16,0.007840000092983246
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,384,8,0.007308799773454666
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,384,4,0.007020799815654755
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,384,8,0.013686400651931763
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,384,2,0.006377600133419037
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,384,4,0.011270400136709213
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,384,2,0.009424000233411788
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,384,1,0.008963199704885483
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,512,128,0.01879359930753708
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,512,128,0.09982399940490723
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,512,64,0.012870399653911591
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,512,64,0.05354239940643311
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,512,32,0.0088639996945858
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,512,32,0.03130879998207092
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,512,16,0.00687360018491745
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,512,16,0.018985599279403687
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,512,8,0.006380800157785416
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,512,8,0.014790399372577668
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,512,4,0.006227200105786324
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,512,4,0.010911999642848969
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,512,2,0.004969599843025208
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,512,2,0.008905600011348724
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,512,1,0.005404800176620483
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,512,1,0.008982399851083756
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,768,128,0.027212798595428467
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,384,16,0.016883200407028197
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,768,128,0.14599039554595947
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,768,64,0.07616320252418518
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,320,32,0.022937600314617158
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,768,32,0.010860799998044967
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,384,1,0.006908799707889557
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,768,16,0.008499199897050858
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,768,32,0.0434112012386322
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,768,8,0.0067840002477169035
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,768,8,0.016774399578571318
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,768,16,0.025177600979804992
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,768,4,0.0063680000603199005
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,768,4,0.012604799866676331
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,768,2,0.010896000266075134
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,768,1,0.005075199902057648
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,768,1,0.008761599659919739
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1024,128,0.03443520069122315
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1024,64,0.017628799378871917
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1024,128,0.19300479888916017
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1024,64,0.0989408016204834
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1024,32,0.01154239997267723
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1024,16,0.00883520022034645
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1024,16,0.031206399202346802
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1024,8,0.006723199784755707
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1024,8,0.020895999670028687
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1024,4,0.006393600255250931
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1024,4,0.014470399916172027
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1024,2,0.006374400109052658
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1024,2,0.010771200060844421
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1024,1,0.005398400127887726
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,768,64,0.014822399616241455
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1024,1,0.008799999952316284
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1536,128,0.046003198623657225
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1536,128,0.2873152017593384
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,768,2,0.006118400022387505
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1536,64,0.1457919955253601
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1536,32,0.014582400023937226
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1536,32,0.07636160254478455
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1536,16,0.011180800199508668
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1024,32,0.054688000679016115
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1536,8,0.008460800349712371
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1536,16,0.04357439875602722
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1536,4,0.006870400160551071
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1536,8,0.025382399559020996
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1536,4,0.016899199783802034
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1536,2,0.006348799914121628
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1536,2,0.01297599971294403
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,1536,1,0.010886400192975997
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2048,128,0.06087679862976074
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2048,128,0.3787904024124146
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2048,64,0.0341728001832962
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2048,64,0.1926367998123169
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2048,32,0.017577600479125977
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2048,32,0.10053759813308716
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2048,16,0.011363200098276138
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2048,16,0.053673601150512694
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2048,8,0.008659200370311737
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2048,8,0.031219199299812317
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2048,4,0.006729599833488464
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1536,64,0.02717759907245636
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2048,4,0.020601600408554077
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2048,2,0.006518399715423584
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2048,2,0.014672000706195832
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,2048,1,0.006185600161552429
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,2048,1,0.010655999928712846
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,3072,128,0.08278719782829284
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,3072,64,0.04764800071716309
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,3072,32,0.026480001211166383
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,3072,64,0.2867327928543091
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,3072,128,0.5634016036987305
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,3072,16,0.014847999811172486
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,3072,32,0.14555840492248534
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,3072,8,0.010665600001811982
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,3072,4,0.008617600053548813
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,3072,8,0.043996798992156985
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,3072,4,0.025302401185035704
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,3072,2,0.007145600020885467
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,3072,1,0.006569600105285645
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,3072,1,0.013142399489879608
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4096,128,0.11103359460830689
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4096,128,0.7448575973510743
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4096,64,0.060329598188400266
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4096,64,0.37707200050354006
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4096,32,0.03429119884967804
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4096,32,0.19286400079727173
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4096,16,0.01786559969186783
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4096,16,0.10110080242156982
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4096,8,0.011769600212574005
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4096,8,0.05380799770355225
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4096,4,0.009033600240945816
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4096,4,0.03134399950504303
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4096,2,0.007065600156784058
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,3072,16,0.07628160119056701
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4096,2,0.021222400665283202
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,4096,1,0.006489600241184235
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,4096,1,0.014953599870204925
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,3072,2,0.01682240068912506
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,6144,64,0.08161280155181885
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,6144,128,1.1164608001708984
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,6144,64,0.5616352081298828
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,6144,32,0.046265599131584165
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,6144,16,0.026976001262664796
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,6144,32,0.2863647937774658
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,6144,8,0.014815999567508698
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,6144,8,0.07624639868736267
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,6144,4,0.010847999900579452
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,6144,4,0.04345279932022095
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,6144,2,0.008524800091981888
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,6144,2,0.025097599625587462
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,6144,1,0.006451199948787689
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,6144,1,0.016918399930000307
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8192,128,0.209168004989624
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8192,128,1.4950559616088868
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8192,64,0.11075839996337891
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,6144,128,0.16203199625015258
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8192,32,0.06099519729614258
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8192,64,0.7461631774902344
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8192,16,0.033215999603271484
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8192,32,0.37836480140686035
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8192,8,0.017628799378871917
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8192,16,0.19113919734954835
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8192,8,0.10111360549926758
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8192,4,0.05384320020675659
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,6144,16,0.14479999542236327
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8192,2,0.030969598889350893
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8192,1,0.006630399823188781
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,fp8,8192,1,0.019177600741386414
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,1536,1,0.005878400057554245
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8192,4,0.01064639985561371
SGLang,0.5.9,NVIDIA GB200,mla_gen_pre,default,float16,8192,2,0.009065599739551544
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1,128,0.006470400094985962
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1,64,0.005811199918389321
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1,32,0.00639680027961731
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1,64,0.008588799834251403
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1,16,0.006511999666690827
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1,128,0.008633600175380706
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1,16,0.008550400286912918
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1,8,0.00867839977145195
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1,4,0.004720000177621841
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1,4,0.00868159979581833
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1,2,0.0055263999849557875
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1,2,0.008659200370311737
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1,1,0.004793599992990494
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1,1,0.006742399930953979
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2,128,0.006390400230884552
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1,8,0.006451199948787689
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2,64,0.005382400006055832
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2,128,0.008947200328111648
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2,32,0.006412799656391144
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2,32,0.008703999966382981
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2,16,0.0064351998269557955
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2,16,0.008780799806118011
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2,64,0.00867839977145195
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2,8,0.00621119998395443
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2,4,0.0048096001148223875
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2,4,0.008700799942016602
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2,2,0.004560000076889992
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2,2,0.008607999980449676
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2,1,0.004707200080156326
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2,1,0.0074143998324871065
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4,128,0.0064351998269557955
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4,128,0.008777599781751633
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4,64,0.006364800035953522
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4,64,0.008515200018882752
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4,32,0.006572800129652024
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4,32,0.008646400272846222
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4,16,0.0066431999206542965
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4,16,0.008729600161314011
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4,8,0.006623999774456024
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4,8,0.008687999844551087
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4,4,0.004691199958324432
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4,4,0.00859839990735054
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4,2,0.004720000177621841
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4,2,0.008671999722719193
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4,1,0.00469760000705719
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4,1,0.008591999858617782
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8,128,0.0065311998128890995
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8,128,0.008656000345945358
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8,64,0.005110400170087815
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8,64,0.008607999980449676
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8,32,0.006598400324583054
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8,32,0.008537600189447403
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8,16,0.006496000289916992
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8,16,0.008553600311279297
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8,8,0.006329599767923355
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8,8,0.008671999722719193
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8,4,0.004646399989724159
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8,4,0.00854720026254654
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8,2,0.004614400118589402
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2,8,0.008799999952316284
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8,1,0.004566400125622749
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8,1,0.008665599673986436
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16,128,0.0064191997051239015
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16,128,0.010768000036478043
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16,64,0.006508799642324448
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16,64,0.008774399757385254
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16,32,0.005680000036954879
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1,32,0.00868159979581833
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16,32,0.008710400015115739
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16,16,0.005680000036954879
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16,8,0.006297600269317627
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16,8,0.008793599903583527
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16,4,0.006153599917888641
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16,4,0.008665599673986436
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16,2,0.006102399900555611
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16,2,0.00838399976491928
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8,2,0.008553600311279297
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16,1,0.006044799834489823
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16,1,0.008633600175380706
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,32,128,0.006748799979686737
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,32,64,0.006227200105786324
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,32,64,0.010873600095510482
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,32,32,0.006294400244951248
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,32,32,0.009763199836015701
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,32,16,0.0060479998588562015
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,32,16,0.008563199639320373
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,32,8,0.0063008002936840056
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,32,8,0.008793599903583527
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,32,4,0.006246399879455566
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,32,4,0.008691199868917466
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16,16,0.008646400272846222
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,32,2,0.00634239986538887
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,32,1,0.004844800010323525
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,32,1,0.008265600353479386
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,48,128,0.006620799750089645
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,48,128,0.016969600319862367
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,48,64,0.006361600011587143
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,48,64,0.012736000120639801
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,48,32,0.006307200342416763
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,48,32,0.010633599758148194
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,32,128,0.01475519984960556
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,48,16,0.008723200112581254
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,48,8,0.005683200061321258
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,48,8,0.008499199897050858
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,48,4,0.0047807998955249785
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,48,4,0.008582399785518646
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,48,2,0.004783999919891357
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,48,2,0.008630400151014328
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,48,1,0.00461760014295578
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,48,1,0.008511999994516373
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,64,128,0.006540799885988236
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,64,128,0.018966400623321535
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,64,64,0.006534399837255478
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,64,64,0.014732800424098969
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,32,2,0.00859839990735054
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,64,32,0.006364800035953522
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,64,32,0.01080320030450821
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,48,16,0.0063680000603199005
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,64,16,0.008720000088214875
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,64,8,0.005977600067853928
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,64,8,0.008611200004816055
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,64,4,0.005683200061321258
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,64,4,0.008675199747085572
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,64,2,0.005516799911856651
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,64,2,0.008508799970149994
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,64,1,0.005273599922657013
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,64,1,0.008550400286912918
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,80,128,0.006857600063085556
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,80,128,0.02086080014705658
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,80,64,0.014921599626541137
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,80,32,0.00631679967045784
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,80,32,0.012035199999809265
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,80,16,0.006492800265550614
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,80,16,0.01051200032234192
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,80,8,0.006511999666690827
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,80,8,0.00859839990735054
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,80,4,0.006022400036454201
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,80,4,0.008560000360012055
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,80,2,0.004630399867892265
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,80,2,0.008633600175380706
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,80,1,0.004636799916625023
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,80,1,0.008582399785518646
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,96,128,0.006547199934720993
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,96,128,0.024905599653720856
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,96,64,0.006521599739789963
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,96,64,0.016710400581359863
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,96,32,0.00650240033864975
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,96,32,0.012691199779510498
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,96,16,0.0064640000462532045
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,96,16,0.01064319983124733
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,96,8,0.00647680014371872
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,64,16,0.006467200070619583
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,96,4,0.005385600030422211
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,96,4,0.008745600283145905
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,96,2,0.0046847999095916745
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,96,2,0.008495999872684479
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,96,1,0.004579199850559235
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,96,1,0.008505599945783615
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,80,64,0.006431999802589417
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,128,128,0.029094401001930236
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,128,64,0.018748800456523895
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,128,64,0.006972800195217133
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,128,32,0.006489600241184235
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,128,16,0.006233600154519081
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,128,16,0.010780800133943558
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,128,8,0.006329599767923355
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,128,8,0.00873280018568039
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,128,4,0.004700800031423568
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,128,4,0.008534400165081025
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,128,2,0.004665600135922432
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,128,2,0.008579199761152267
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,128,1,0.0044511999934911724
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,128,1,0.008544000238180161
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,160,128,0.008470399677753449
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,160,128,0.03526400029659271
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,160,64,0.006524799764156342
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,160,64,0.02107200026512146
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,160,32,0.00642239972949028
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,160,32,0.014847999811172486
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,160,16,0.006518399715423584
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,160,16,0.012716799974441528
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,160,8,0.006537599861621857
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,160,8,0.010566399991512298
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,160,4,0.00456320010125637
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,160,4,0.008822400122880936
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,160,2,0.004633599892258644
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,160,2,0.008640000224113464
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,160,1,0.0046431999653577805
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,160,1,0.008566399663686752
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,192,128,0.008560000360012055
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,128,128,0.008326400071382523
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,192,128,0.03928320109844208
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,192,64,0.0068000003695487974
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,192,64,0.024988800287246704
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,192,32,0.0065760001540184024
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,192,32,0.016864000260829924
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,192,16,0.006377600133419037
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,192,16,0.012700800597667695
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,192,8,0.010691200196743012
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,192,4,0.004611200094223023
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,96,8,0.008726400136947633
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,192,4,0.008499199897050858
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,192,2,0.0047807998955249785
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,192,2,0.008640000224113464
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,192,1,0.004665600135922432
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,192,1,0.008710400015115739
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,256,128,0.00862400010228157
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,256,128,0.05160639882087707
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,256,64,0.0068351998925209045
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,256,64,0.029334399104118346
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,192,8,0.006185600161552429
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,256,32,0.006601600348949433
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,256,32,0.019036799669265747
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,256,16,0.014854399859905243
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,256,8,0.006588800251483918
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,256,8,0.011023999750614166
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,256,4,0.006652799993753433
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,256,4,0.01056319996714592
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,256,2,0.006527999788522721
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,256,2,0.008963199704885483
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,256,1,0.00589120015501976
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,256,1,0.009753599762916565
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,320,128,0.010480000078678131
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,320,128,0.06208639740943909
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,320,64,0.00862400010228157
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,320,64,0.03544960021972656
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,320,32,0.006988800317049027
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,320,32,0.023097600042819976
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,320,16,0.0068512000143527985
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,320,16,0.016096000373363496
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,128,32,0.015184000134468079
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,320,8,0.012758399546146392
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,320,4,0.010873600095510482
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,320,4,0.006377600133419037
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,320,2,0.0057599999010562895
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,320,1,0.005231999978423119
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,320,2,0.008991999924182892
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,320,1,0.008671999722719193
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,384,128,0.011062400043010711
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,256,16,0.0065760001540184024
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,384,64,0.039324799180030824
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,384,32,0.006547199934720993
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,384,32,0.024383999407291412
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,384,16,0.006483200192451477
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,384,16,0.01679999977350235
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,384,8,0.00650240033864975
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,384,8,0.012591999769210816
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,384,4,0.006499200314283371
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,384,4,0.010678400099277497
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,384,2,0.005327999964356423
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,384,2,0.008745600283145905
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,384,1,0.004783999919891357
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,384,1,0.00854720026254654
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,512,128,0.013929599523544311
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,512,128,0.0988640010356903
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,512,64,0.008607999980449676
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,512,64,0.05120000243186951
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,512,32,0.006819199770689011
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,512,32,0.029235199093818665
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,512,16,0.006473600119352341
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,512,16,0.01900479942560196
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,512,8,0.006304000318050384
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,512,8,0.013894400000572205
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,512,4,0.005763199925422668
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,512,4,0.010681600123643876
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,512,2,0.005104000121355057
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,512,2,0.008870399743318557
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,384,128,0.07487360239028931
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,512,1,0.004614400118589402
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,512,1,0.00865280032157898
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,768,128,0.027379199862480164
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,768,128,0.1459391951560974
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,768,64,0.07427840232849121
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,768,32,0.008665599673986436
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,768,32,0.04068160057067871
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,768,16,0.006595200300216675
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,768,16,0.024979199469089507
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,768,8,0.006572800129652024
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,768,8,0.016857600212097167
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,768,4,0.006310400366783142
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,768,4,0.012591999769210816
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,768,2,0.006265600025653839
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,768,2,0.010716799646615982
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,768,1,0.004659200087189674
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,768,1,0.008767999708652496
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1024,128,0.0331743985414505
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1024,128,0.19205119609832763
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1024,64,0.012918399274349212
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1024,64,0.09872959852218628
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,320,8,0.006329599767923355
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1024,32,0.008780799806118011
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1024,16,0.006601600348949433
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1024,32,0.05157439708709717
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1024,16,0.029100799560546876
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1024,8,0.018915200233459474
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1024,4,0.00658240020275116
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,384,64,0.008595199882984161
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1024,4,0.01483519971370697
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1024,2,0.006195199862122536
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1024,2,0.0108255997300148
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1024,1,0.006473600119352341
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1024,1,0.010831999778747558
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1536,128,0.04731839895248413
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1536,128,0.287007999420166
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1536,64,0.02529920041561127
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1536,64,0.14587520360946654
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1536,32,0.010771200060844421
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1536,32,0.0743615984916687
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1536,16,0.008630400151014328
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1536,16,0.042752000689506534
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1536,8,0.007366400212049484
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1536,8,0.0252703994512558
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1536,4,0.0065311998128890995
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1536,4,0.01714559942483902
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1536,2,0.00655680000782013
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1024,8,0.00655680000782013
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,1536,1,0.0063391998410224915
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1536,2,0.013049599528312684
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,1536,1,0.010496000200510025
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2048,128,0.06539199948310852
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2048,64,0.03133760094642639
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2048,128,0.3784192085266113
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,768,64,0.010566399991512298
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2048,32,0.01297920048236847
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2048,16,0.00883840024471283
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2048,64,0.191977596282959
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2048,16,0.051622402667999265
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2048,8,0.007049600034952164
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2048,8,0.029388800263404846
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2048,4,0.006444799900054932
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2048,4,0.018915200233459474
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2048,2,0.006467200070619583
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,2048,1,0.006527999788522721
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2048,1,0.010979200154542923
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,3072,128,0.08371840119361877
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,3072,128,0.5786943912506104
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,3072,64,0.04609920084476471
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,3072,32,0.02436159998178482
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,3072,32,0.14609919786453246
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,3072,16,0.010620799660682679
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,3072,16,0.0743071973323822
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,3072,8,0.008416000008583068
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,3072,8,0.04090240001678467
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,3072,4,0.006867200136184692
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,3072,4,0.025196799635887147
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,3072,2,0.0064351998269557955
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,3072,2,0.01717440038919449
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,3072,1,0.006409599632024765
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,3072,1,0.013040000200271606
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4096,128,0.11673280000686645
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2048,32,0.0984607994556427
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4096,64,0.06001600027084351
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4096,128,0.7904160022735596
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,2048,2,0.014947199821472168
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4096,32,0.031231999397277832
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4096,16,0.012732799351215362
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4096,16,0.09920639991760254
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4096,8,0.008963199704885483
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4096,8,0.0515936017036438
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4096,4,0.006822399795055389
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4096,4,0.03057279884815216
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4096,2,0.0065600000321865085
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4096,2,0.019219200313091277
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,4096,1,0.006454399973154068
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4096,1,0.014844800531864166
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,6144,128,0.15643839836120604
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,6144,128,1.2522720336914062
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,6144,64,0.08058239817619324
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,6144,64,0.5750720024108886
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,6144,32,0.04481280148029328
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,6144,32,0.28504960536956786
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,6144,16,0.0239424005150795
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4096,32,0.19171839952468872
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,6144,16,0.14604159593582153
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,6144,8,0.07406399846076965
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,6144,4,0.008764799684286118
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,6144,4,0.04130240082740784
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,6144,2,0.006831999868154526
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,6144,2,0.025155198574066163
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,6144,1,0.006412799656391144
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,6144,1,0.01717440038919449
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,3072,64,0.28419520854949953
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8192,128,0.22035520076751708
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8192,64,0.1126431941986084
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8192,64,0.7974080085754395
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8192,128,1.7162559509277344
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8192,32,0.060252797603607175
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8192,16,0.03036159873008728
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8192,32,0.37842559814453125
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8192,16,0.19287680387496947
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8192,8,0.012783999741077422
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8192,4,0.008601599931716919
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8192,8,0.09891200065612793
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8192,2,0.00697920024394989
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8192,4,0.05151680111885071
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8192,2,0.03062720000743866
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,8192,1,0.006387200206518173
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,8192,1,0.019171200692653656
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,4096,64,0.37959680557250974
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,12288,128,0.3267904043197632
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,12288,64,0.1497599959373474
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,6144,8,0.010592000186443329
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,12288,32,0.07838400006294251
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,12288,32,0.5768383979797364
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,12288,64,1.23121919631958
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,12288,16,0.04361599981784821
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,12288,8,0.026956799626350402
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,12288,128,2.671526336669922
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,12288,16,0.2850944042205811
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,12288,8,0.14582719802856445
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,12288,2,0.008604799956083297
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,12288,4,0.07425280213356018
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,12288,1,0.006451199948787689
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,12288,1,0.025014400482177734
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16384,128,0.43144640922546384
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16384,128,3.5604801177978516
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16384,64,0.21189119815826415
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,12288,4,0.010550399869680404
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16384,32,0.11007360219955445
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,12288,2,0.041171199083328246
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16384,16,0.05921279788017273
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16384,8,0.03129279911518097
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16384,64,1.70283203125
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16384,32,0.7896096229553222
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16384,8,0.19189759492874145
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16384,4,0.012723200023174286
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16384,4,0.09890559911727906
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16384,2,0.05164480209350586
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16384,1,0.0070592001080513
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16384,1,0.029091200232505797
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,20480,128,0.5399392127990723
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,20480,64,0.2655679941177368
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,20480,64,2.240380859375
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,20480,32,0.13565119504928588
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,20480,128,4.480716705322266
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,20480,16,0.07041280269622803
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,20480,32,1.0299551963806153
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,20480,8,0.03872320055961609
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,16384,16,0.37538559436798097
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,20480,16,0.4707520008087158
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,20480,4,0.017155200242996216
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,20480,8,0.23994240760803223
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,20480,4,0.1215008020401001
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,20480,2,0.06419199705123901
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,fp8,20480,1,0.03531840145587921
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,16384,2,0.008953599631786347
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,20480,2,0.009174399822950364
SGLang,0.5.9,NVIDIA GB200,mla_gen_post,default,float16,20480,1,0.006668800115585327
