framework,version,device,op_name,kernel_source,bmm_dtype,num_tokens,num_heads,latency
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1,128,0.0058559998869895935
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1,128,0.008448000252246856
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1,64,0.0047231998294591905
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1,64,0.008236800134181977
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1,32,0.004473600164055824
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1,32,0.00796160027384758
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1,16,0.004185599833726883
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1,16,0.0075935997068881985
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1,8,0.003987200185656547
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1,8,0.007209599763154983
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1,4,0.00379519984126091
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1,4,0.006758400052785873
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1,2,0.004303999990224838
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1,2,0.007103999704122543
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1,1,0.003903999924659729
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1,1,0.00684799998998642
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2,128,0.007097599655389785
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2,128,0.009142400324344635
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2,64,0.005036799982190132
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2,64,0.00854720026254654
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2,32,0.0052767999470233916
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2,32,0.008054400235414505
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2,16,0.004131200164556504
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2,16,0.0075103998184204105
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2,8,0.003884800150990486
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2,8,0.007235199958086014
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2,4,0.0037983998656272886
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2,4,0.007254400104284286
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2,2,0.004425600171089172
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2,2,0.00732479989528656
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2,1,0.0038463998585939406
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2,1,0.006883200258016586
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4,128,0.006019200012087822
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4,128,0.009148799628019334
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4,64,0.00490880012512207
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4,64,0.00841279998421669
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4,32,0.004444799944758415
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4,32,0.008198399841785432
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4,16,0.004121600091457367
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4,16,0.007811199873685837
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4,8,0.003932800143957138
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4,8,0.007388799637556076
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4,4,0.003836800158023834
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4,4,0.0070880003273487095
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4,2,0.004348799958825112
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4,2,0.007235199958086014
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4,1,0.003852799907326698
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4,1,0.00684799998998642
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8,128,0.0060095999389886854
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8,128,0.010022400319576264
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8,64,0.00488319993019104
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8,64,0.00875839963555336
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8,32,0.004614400118589402
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8,32,0.0077344000339508055
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8,16,0.003958399966359138
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8,16,0.007945600152015685
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8,8,0.003888000175356865
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8,8,0.007683199644088745
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8,4,0.0038752000778913497
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8,4,0.0075103998184204105
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8,2,0.003920000046491623
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8,2,0.0071103997528553006
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8,1,0.003868800029158592
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8,1,0.006982400268316269
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,16,128,0.006041600182652473
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,16,128,0.011503999680280685
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,16,64,0.0048640001565217975
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,16,64,0.009468799829483033
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,16,32,0.00448639988899231
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,16,32,0.008691199868917466
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,16,16,0.004134399816393852
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,16,16,0.007964800298213958
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,16,8,0.003955199941992759
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,16,8,0.0077344000339508055
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,16,4,0.004220800101757049
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,16,4,0.007366400212049484
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,16,2,0.003807999938726425
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,16,2,0.007372800260782242
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,16,1,0.0039391998201608654
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,16,1,0.007372800260782242
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,32,128,0.006233600154519081
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,32,128,0.014550399780273438
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,32,64,0.005113599821925163
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,32,64,0.01120320037007332
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,32,32,0.0045056000351905824
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,32,32,0.009040000289678574
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,32,16,0.004204799979925155
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,32,16,0.008448000252246856
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,32,8,0.003987200185656547
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,32,8,0.007916799932718276
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,32,4,0.003888000175356865
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,32,4,0.0079584002494812
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,32,2,0.00379519984126091
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,32,2,0.007491199672222138
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,32,1,0.003913599997758865
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,32,1,0.007152000069618225
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,48,128,0.00692799985408783
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,48,128,0.01807360053062439
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,48,64,0.005420799925923347
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,48,64,0.012851199507713318
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,48,32,0.004787199944257736
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,48,32,0.010278400033712387
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,48,16,0.004428799822926521
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,48,16,0.008867199718952178
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,48,8,0.004108799993991852
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,48,8,0.008089599758386612
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,48,4,0.003971200063824653
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,48,4,0.007644800096750259
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,48,2,0.003916800022125244
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,48,2,0.007452800124883652
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,48,1,0.0038495998829603194
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,48,1,0.007177600264549255
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,64,128,0.007596799731254577
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,64,128,0.021961599588394165
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,64,64,0.005750399827957153
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,64,64,0.014825600385665893
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,64,32,0.004764800146222115
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,64,32,0.011155200004577637
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,64,16,0.004403200000524521
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,64,16,0.008848000317811966
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,64,8,0.004095999896526337
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,64,8,0.007983999699354172
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,64,4,0.003884800150990486
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,64,4,0.007644800096750259
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,64,2,0.0038463998585939406
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,64,2,0.007868800312280655
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,64,1,0.003888000175356865
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,64,1,0.007635200023651123
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,80,128,0.008057600259780884
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,80,128,0.024959999322891235
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,80,64,0.005987200140953064
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,80,64,0.016540800034999848
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,80,32,0.005017599835991859
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,80,32,0.012051200121641159
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,80,16,0.004527999833226204
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,80,16,0.009635200351476669
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,80,8,0.004233599826693535
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,80,8,0.00843520015478134
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,80,4,0.004025600105524063
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,80,4,0.007651200145483017
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,80,2,0.0038176000118255613
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,80,2,0.007344000041484833
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,80,1,0.003836800158023834
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,80,1,0.0075552001595497135
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,96,128,0.008627200126647949
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,96,128,0.028726398944854736
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,96,64,0.006371200084686279
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,96,64,0.01823039948940277
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,96,32,0.00517439991235733
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,96,32,0.012681600451469422
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,96,16,0.004598399996757508
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,96,16,0.009839999675750732
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,96,8,0.00416640006005764
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,96,8,0.008604799956083297
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,96,4,0.0040352001786232
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,96,4,0.008083199709653854
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,96,2,0.003948799893260002
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,96,2,0.007631999999284744
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,96,1,0.003891199827194214
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,96,1,0.007468800246715546
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,128,128,0.010460799932479859
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,128,128,0.03560320138931274
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,128,64,0.007196799665689468
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,128,64,0.021699200570583343
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,128,32,0.005587200075387955
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,128,32,0.01462399959564209
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,128,16,0.004678399860858917
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,128,16,0.011068800091743469
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,128,8,0.004323200136423111
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,128,8,0.00910080000758171
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,128,4,0.003961599990725517
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,128,4,0.00799039974808693
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,128,2,0.003958399966359138
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,128,2,0.007574400305747986
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,128,1,0.0038784001022577285
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,128,1,0.007744000107049942
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,160,128,0.011503999680280685
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,160,128,0.04333760142326355
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,160,64,0.007996799796819687
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,160,64,0.025097599625587462
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,160,32,0.005817599967122078
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,160,32,0.01613759994506836
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,160,16,0.004764800146222115
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,160,16,0.0118367999792099
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,160,8,0.004527999833226204
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,160,8,0.009676799923181535
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,160,4,0.004163200035691261
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,160,4,0.00843840017914772
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,160,2,0.004134399816393852
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,160,2,0.007727999985218048
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,160,1,0.003932800143957138
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,160,1,0.007478400319814682
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,192,128,0.016390399634838106
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,192,128,0.05271999835968018
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,192,64,0.008057600259780884
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,192,64,0.028355199098587035
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,192,32,0.006115199998021126
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,192,32,0.018003199994564057
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,192,16,0.005004800111055374
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,192,16,0.012585599720478059
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,192,8,0.004572800174355507
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,192,8,0.010060799866914749
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,192,4,0.004134399816393852
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,192,4,0.008515200018882752
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,192,2,0.0039264000952243805
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,192,2,0.007923199981451034
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,192,1,0.003923200070858002
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,192,1,0.007603199779987335
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,256,128,0.021516799926757812
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,256,128,0.06842240095138549
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,256,64,0.009747199714183807
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,256,64,0.03537600040435791
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,256,32,0.006963200122117996
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,256,32,0.021481600403785706
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,256,16,0.005353600159287453
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,256,16,0.014422400295734406
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,256,8,0.004575999826192856
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,256,8,0.010761599987745285
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,256,4,0.005008000135421753
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,256,4,0.008822400122880936
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,256,2,0.00398080013692379
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,256,2,0.008022399991750718
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,256,1,0.003984000161290168
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,256,1,0.008035200089216233
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,320,128,0.023660799860954283
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,320,128,0.08259199857711792
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,320,64,0.011884800344705581
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,320,64,0.04489279985427856
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,320,32,0.007199999690055847
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,320,32,0.024585600197315215
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,320,16,0.005721599981188774
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,320,16,0.015945599973201753
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,320,8,0.0049183998256921765
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,320,8,0.011878400295972823
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,320,4,0.0044511999934911724
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,320,4,0.009532800316810608
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,320,2,0.00416640006005764
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,320,2,0.008086399734020233
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,320,1,0.003942399844527244
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,320,1,0.007599999755620956
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,384,128,0.027526399493217467
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,384,128,0.09940479993820191
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,384,64,0.014556799829006196
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,384,64,0.05339840054512024
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,384,32,0.008444800227880477
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,384,32,0.028703999519348145
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,384,16,0.0061471998691558834
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,384,16,0.017948800325393678
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,384,8,0.004982399940490723
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,384,8,0.01233920007944107
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,384,4,0.004380799829959869
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,384,4,0.009804800152778625
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,384,2,0.004153599962592125
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,384,2,0.008479999750852585
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,384,1,0.003903999924659729
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,384,1,0.008019199967384339
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,512,128,0.03336640000343323
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,512,128,0.12453440427780152
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,512,64,0.01910399943590164
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,512,64,0.0673919975757599
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,512,32,0.00952960029244423
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,512,32,0.03500159978866577
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,512,16,0.006831999868154526
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,512,16,0.021510399878025055
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,512,8,0.005401600152254105
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,512,8,0.014572800695896148
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,512,4,0.004678399860858917
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,512,4,0.011078400164842605
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,512,2,0.004383999854326248
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,512,2,0.009040000289678574
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,512,1,0.004383999854326248
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,512,1,0.008003199845552445
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,768,128,0.044803199172019956
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,768,128,0.18290879726409912
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,768,64,0.025523200631141663
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,768,64,0.09895679950714112
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,768,32,0.013689599931240082
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,768,32,0.05180479884147644
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,768,16,0.008422400057315826
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,768,16,0.02868799865245819
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,768,8,0.006083200126886368
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,768,8,0.017955200374126436
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,768,4,0.004924799874424934
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,768,4,0.012348800152540206
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,768,2,0.004396799951791763
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,768,2,0.009824000298976898
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,768,1,0.0042559999972581865
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,768,1,0.00888959988951683
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1024,128,0.05589759945869446
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1024,128,0.23958079814910888
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1024,64,0.031302401423454286
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1024,64,0.123580801486969
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1024,32,0.01855359971523285
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1024,32,0.06716799736022949
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1024,16,0.009510400146245957
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1024,16,0.03511680066585541
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1024,8,0.0069311998784542085
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1024,8,0.021414400637149812
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1024,4,0.005350400134921074
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1024,4,0.014630399644374847
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1024,2,0.004665600135922432
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1024,2,0.01098880022764206
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1024,1,0.004303999990224838
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1024,1,0.009065599739551544
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1536,128,0.0786080002784729
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1536,128,0.35450561046600343
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1536,64,0.0430976003408432
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1536,64,0.1820639967918396
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1536,32,0.024675199389457704
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1536,32,0.0989471971988678
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1536,16,0.013545599579811097
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1536,16,0.052425599098205565
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1536,8,0.008329600095748901
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1536,8,0.02863680124282837
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1536,4,0.006067200005054474
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1536,4,0.017961600422859193
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1536,2,0.004953600093722343
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1536,2,0.012444800138473511
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,1536,1,0.005177599936723709
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,1536,1,0.00982080027461052
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2048,128,0.10111360549926758
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2048,128,0.47000961303710936
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2048,64,0.054339200258255005
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2048,64,0.23955199718475342
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2048,32,0.03046720027923584
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2048,32,0.12358399629592895
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2048,16,0.01733119934797287
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2048,16,0.06647359728813171
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2048,8,0.009628800302743911
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2048,8,0.035392001271247864
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2048,4,0.00695360004901886
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2048,4,0.02152000069618225
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2048,2,0.005283199995756149
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2048,2,0.014239999651908874
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,2048,1,0.004575999826192856
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,2048,1,0.010732799768447876
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,3072,128,0.14776320457458497
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,3072,128,0.7055520057678223
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,3072,64,0.07613120079040528
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,3072,64,0.35377600193023684
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,3072,32,0.04215039908885956
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,3072,32,0.18260799646377562
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,3072,16,0.024128000438213348
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,3072,16,0.09894400238990783
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,3072,8,0.013670399785041809
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,3072,8,0.051769602298736575
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,3072,4,0.00822720006108284
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,3072,4,0.028598400950431823
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,3072,2,0.006150399893522262
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,3072,2,0.017888000607490538
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,3072,1,0.004975999891757965
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,3072,1,0.012579199671745301
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4096,128,0.18980159759521484
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4096,128,0.9436320304870606
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4096,64,0.09809920191764832
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4096,64,0.4683839797973633
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4096,32,0.052611202001571655
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4096,32,0.23939199447631837
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4096,16,0.030057600140571593
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4096,16,0.12352960109710694
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4096,8,0.018969599902629853
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4096,8,0.066975998878479
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4096,4,0.009587199985980987
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4096,4,0.03538239896297455
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4096,2,0.0070592001080513
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4096,2,0.02157440036535263
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,4096,1,0.005363199859857559
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,4096,1,0.01454080045223236
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,6144,128,0.2762752056121826
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,6144,128,1.428102397918701
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,6144,64,0.1534656047821045
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,6144,64,0.7023231983184814
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,6144,32,0.07499840259552001
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,6144,32,0.35355520248413086
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,6144,16,0.04159039855003357
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,6144,16,0.18182400465011597
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,6144,8,0.023984000086784363
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,6144,8,0.09862719774246216
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,6144,4,0.01369599997997284
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,6144,4,0.05153599977493286
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,6144,2,0.008169600367546081
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,6144,2,0.0283488005399704
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,6144,1,0.0059519998729228975
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,6144,1,0.01775040030479431
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8192,128,0.3637984037399292
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8192,128,1.9071647644042968
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8192,64,0.20967040061950684
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8192,64,0.9435327529907227
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8192,32,0.0970687985420227
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8192,32,0.46420798301696775
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8192,16,0.05247039794921875
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8192,16,0.23936960697174073
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8192,8,0.030294400453567506
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8192,8,0.12364480495452881
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8192,4,0.016678400337696075
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8192,4,0.06629440188407898
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8192,2,0.009536000341176987
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8192,2,0.035139200091362
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,float16,8192,1,0.0068351998925209045
SGLang,0.5.10,NVIDIA H200,mla_gen_pre,default,fp8,8192,1,0.021510399878025055
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1,128,0.005395200103521347
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1,128,0.008012799918651581
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1,64,0.004774399846792221
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1,64,0.00823040008544922
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1,32,0.004620800167322159
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1,32,0.007983999699354172
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1,16,0.004281599819660187
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1,16,0.007811199873685837
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1,8,0.006006399914622307
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1,8,0.008172799646854401
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1,4,0.004233599826693535
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1,4,0.0072512000799179075
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1,2,0.004172800108790398
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1,2,0.007068800181150437
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1,1,0.0042015999555587765
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1,1,0.007011199742555619
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2,128,0.005552000179886818
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2,128,0.0088639996945858
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2,64,0.0049632001668214795
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2,64,0.008393599838018417
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2,32,0.004560000076889992
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2,32,0.00811839997768402
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2,16,0.004572800174355507
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2,16,0.007583999633789062
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2,8,0.005750399827957153
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2,8,0.008185599744319916
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2,4,0.0043680001050233844
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2,4,0.007753600180149078
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2,2,0.00419199988245964
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2,2,0.007360000163316727
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2,1,0.004128000140190125
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2,1,0.007113599777221679
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4,128,0.005481600016355515
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4,128,0.00878399983048439
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4,64,0.004851200059056282
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4,64,0.008243200182914735
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4,32,0.004614400118589402
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4,32,0.00827839970588684
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4,16,0.004268800094723702
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4,16,0.007936000078916549
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4,8,0.0059264000505208966
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4,8,0.008425600081682205
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4,4,0.004239999875426293
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4,4,0.00748480036854744
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4,2,0.006207999959588051
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4,2,0.007327999919652939
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4,1,0.004112000018358231
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4,1,0.007168000191450119
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8,128,0.005606399849057198
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8,128,0.009839999675750732
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8,64,0.004662400111556053
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8,64,0.008579199761152267
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8,32,0.004339199885725975
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8,32,0.007955200225114822
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8,16,0.004208000004291534
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8,16,0.00809599980711937
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8,8,0.004284799844026565
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8,8,0.00782719999551773
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8,4,0.004371200129389763
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8,4,0.007843200117349625
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8,2,0.004208000004291534
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8,2,0.0074432000517845156
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8,1,0.0041184000670909885
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8,1,0.007305599749088287
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16,128,0.00570559985935688
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16,128,0.011088000237941742
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16,64,0.00461760014295578
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16,64,0.009180799871683121
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16,32,0.00451200008392334
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16,32,0.009148799628019334
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16,16,0.004307200014591217
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16,16,0.008284799754619598
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16,8,0.004291199892759323
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16,8,0.007846400141716003
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16,4,0.004182400181889534
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16,4,0.007491199672222138
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16,2,0.004080000147223472
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16,2,0.007657600194215774
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16,1,0.00419199988245964
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16,1,0.007619199901819229
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,32,128,0.006032000109553337
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,32,128,0.014188799262046813
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,32,64,0.00485760010778904
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,32,64,0.010655999928712846
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,32,32,0.004454400017857551
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,32,32,0.009017600119113922
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,32,16,0.004428799822926521
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,32,16,0.008633600175380706
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,32,8,0.004332799836993218
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,32,8,0.008297599852085114
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,32,4,0.0042975999414920805
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,32,4,0.00806720033288002
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,32,2,0.004281599819660187
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,32,2,0.007884799689054488
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,32,1,0.004137599840760231
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,32,1,0.007295999675989151
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,48,128,0.006451199948787689
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,48,128,0.0174687996506691
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,48,64,0.0050976000726222995
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,48,64,0.01250240057706833
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,48,32,0.004659200087189674
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,48,32,0.01016639992594719
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,48,16,0.00456320010125637
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,48,16,0.00888959988951683
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,48,8,0.00445760004222393
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,48,8,0.008345600217580795
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,48,4,0.004150399938225746
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,48,4,0.007772800326347351
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,48,2,0.004169600084424019
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,48,2,0.007711999863386154
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,48,1,0.004374400153756142
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,48,1,0.007526399940252304
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,64,128,0.0068800002336502075
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,64,128,0.020927999913692475
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,64,64,0.005369599908590317
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,64,64,0.014230400323867798
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,64,32,0.004636799916625023
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,64,32,0.011084800213575363
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,64,16,0.00456320010125637
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,64,16,0.008816000074148178
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,64,8,0.004390399903059006
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,64,8,0.008268799632787704
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,64,4,0.004681599885225296
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,64,4,0.007791999727487564
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,64,2,0.00411520004272461
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,64,2,0.007945600152015685
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,64,1,0.0042720001190900804
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,64,1,0.007852800190448761
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,80,128,0.007238399982452392
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,80,128,0.023763200640678404
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,80,64,0.005503999814391136
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,80,64,0.015670399367809295
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,80,32,0.004774399846792221
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,80,32,0.011910399794578553
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,80,16,0.0051231998950243
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,80,16,0.009692800045013428
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,80,8,0.0045056000351905824
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,80,8,0.008473599702119828
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,80,4,0.004214400053024292
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,80,4,0.007945600152015685
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,80,2,0.004185599833726883
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,80,2,0.007568000257015229
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,80,1,0.00416640006005764
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,80,1,0.0074592001736164095
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,96,128,0.007609599828720092
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,96,128,0.02725439965724945
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,96,64,0.005721599981188774
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,96,64,0.017369599640369417
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,96,32,0.004825599864125252
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,96,32,0.012531200051307678
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,96,16,0.004623999819159508
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,96,16,0.009788800030946732
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,96,8,0.0043519999831914905
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,96,8,0.00880960002541542
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,96,4,0.004329600185155868
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,96,4,0.008457600325345992
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,96,2,0.004278400167822838
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,96,2,0.007916799932718276
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,96,1,0.004172800108790398
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,96,1,0.0075935997068881985
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,128,128,0.00827839970588684
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,128,128,0.03484799861907959
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,128,64,0.006108799949288368
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,128,64,0.020448000729084016
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,128,32,0.00498879998922348
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,128,32,0.014156800508499146
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,128,16,0.004735999926924706
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,128,16,0.011072000116109848
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,128,8,0.004646399989724159
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,128,8,0.009180799871683121
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,128,4,0.0044064000248909
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,128,4,0.008303999900817871
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,128,2,0.00419199988245964
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,128,2,0.007811199873685837
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,128,1,0.0046431999653577805
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,128,1,0.0077632002532482145
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,160,128,0.009718400239944459
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,160,128,0.043670400977134705
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,160,64,0.006592000275850296
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,160,64,0.023571200668811798
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,160,32,0.005142400041222572
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,160,32,0.015414400398731232
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,160,16,0.004707200080156326
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,160,16,0.011840000003576278
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,160,8,0.004659200087189674
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,160,8,0.009737599641084671
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,160,4,0.004390399903059006
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,160,4,0.008803199976682663
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,160,2,0.00432640016078949
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,160,2,0.007993599772453308
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,160,1,0.004156799986958504
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,160,1,0.0077855996787548065
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,192,128,0.012742400169372559
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,192,128,0.050988799333572386
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,192,64,0.006822399795055389
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,192,64,0.0268095999956131
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,192,32,0.005436800047755241
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,192,32,0.017385600507259368
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,192,16,0.004806400090456009
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,192,16,0.012601600587368011
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,192,8,0.0045471999794244765
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,192,8,0.009942399710416794
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,192,4,0.004281599819660187
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,192,4,0.008534400165081025
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,192,2,0.004249599948525429
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,192,2,0.008185599744319916
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,192,1,0.004281599819660187
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,192,1,0.007686399668455124
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,256,128,0.01900479942560196
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,256,128,0.06552640199661255
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,256,64,0.00782719999551773
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,256,64,0.034220799803733826
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,256,32,0.005833600088953972
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,256,32,0.02056960016489029
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,256,16,0.005036799982190132
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,256,16,0.013884800672531127
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,256,8,0.004748800024390221
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,256,8,0.010755199939012527
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,256,4,0.004524800181388855
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,256,4,0.009027200192213059
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,256,2,0.004502400010824204
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,256,2,0.008636800199747085
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,256,1,0.0044064000248909
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,256,1,0.008316799998283386
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,320,128,0.02181120067834854
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,320,128,0.08021439909934998
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,320,64,0.008416000008583068
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,320,64,0.04192320108413696
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,320,32,0.006572800129652024
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,320,32,0.023558400571346283
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,320,16,0.0051807999610900875
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,320,16,0.015830400586128234
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,320,8,0.004960000142455101
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,320,8,0.011935999989509583
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,320,4,0.004623999819159508
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,320,4,0.009465599805116654
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,320,2,0.004377600178122521
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,320,2,0.008544000238180161
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,320,1,0.0042720001190900804
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,320,1,0.008102399855852127
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,384,128,0.02505599856376648
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,384,128,0.09631360173225403
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,384,64,0.009318400174379349
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,384,64,0.050012797117233276
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,384,32,0.0065760001540184024
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,384,32,0.02704319953918457
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,384,16,0.005385600030422211
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,384,16,0.017107200622558594
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,384,8,0.00493439994752407
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,384,8,0.01231039986014366
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,384,4,0.0046016000211238865
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,384,4,0.009913600236177444
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,384,2,0.004403200000524521
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,384,2,0.008857599645853042
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,384,1,0.004428799822926521
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,384,1,0.00836160033941269
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,512,128,0.030451199412345885
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,512,128,0.12567039728164672
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,512,64,0.014633600413799287
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,512,64,0.0652064025402069
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,512,32,0.007724799960851669
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,512,32,0.03367680013179779
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,512,16,0.005756799876689911
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,512,16,0.020563200116157532
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,512,8,0.0050592001527547835
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,512,8,0.014179199934005737
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,512,4,0.004735999926924706
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,512,4,0.010847999900579452
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,512,2,0.004508800059556961
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,512,2,0.009110400080680847
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,512,1,0.0045056000351905824
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,512,1,0.00827839970588684
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,768,128,0.04188160002231598
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,768,128,0.18478720188140868
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,768,64,0.022700800001621245
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,768,64,0.09546560049057007
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,768,32,0.009433600306510925
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,768,32,0.04953919947147369
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,768,16,0.006620799750089645
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,768,16,0.026761600375175477
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,768,8,0.005404800176620483
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,768,8,0.01703680008649826
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,768,4,0.004915200173854828
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,768,4,0.012694400548934937
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,768,2,0.004687999933958053
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,768,2,0.01003199964761734
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,768,1,0.004428799822926521
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,768,1,0.00880960002541542
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1024,128,0.053487998247146604
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1024,128,0.24211521148681642
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1024,64,0.028995200991630554
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1024,64,0.12416960000991821
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1024,32,0.012870399653911591
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1024,32,0.06506239771842956
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1024,16,0.007894399762153625
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1024,16,0.033769598603248595
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1024,8,0.005708799883723259
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1024,8,0.020505599677562714
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1024,4,0.005040000006556511
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1024,4,0.013737599551677703
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1024,2,0.004806400090456009
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1024,2,0.010937599837779999
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1024,1,0.004515200108289719
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1024,1,0.009030400216579438
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1536,128,0.07533119916915894
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1536,128,0.3593696117401123
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1536,64,0.040633600950241086
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1536,64,0.18434879779815674
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1536,32,0.021740800142288207
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1536,32,0.0950976014137268
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1536,16,0.00923840031027794
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1536,16,0.04930559992790222
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1536,8,0.006467200070619583
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1536,8,0.026576000452041625
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1536,4,0.005305600166320801
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1536,4,0.01703680008649826
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1536,2,0.004793599992990494
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1536,2,0.012355200201272964
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,1536,1,0.00456320010125637
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,1536,1,0.010227199643850327
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2048,128,0.09730240106582641
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2048,128,0.48931522369384767
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2048,64,0.05136640071868896
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2048,64,0.24127678871154784
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2048,32,0.028182399272918702
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2048,32,0.12424319982528687
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2048,16,0.011292800307273865
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2048,16,0.06461439728736877
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2048,8,0.007772800326347351
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2048,8,0.03339200019836426
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2048,4,0.005552000179886818
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2048,4,0.02019200026988983
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2048,2,0.005078399926424027
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2048,2,0.014169600605964661
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,2048,1,0.004816000163555145
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,2048,1,0.011046399921178817
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,3072,128,0.14608639478683472
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,3072,128,0.7609600067138672
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,3072,64,0.07177600264549255
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,3072,64,0.3578880071640015
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,3072,32,0.04078400135040283
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,3072,32,0.18418560028076172
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,3072,16,0.02130240052938461
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,3072,16,0.09488639831542969
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,3072,8,0.009251199662685394
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,3072,8,0.04921599924564361
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,3072,4,0.006415999680757523
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,3072,4,0.026582399010658266
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,3072,2,0.005289600044488907
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,3072,2,0.017001600563526155
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,3072,1,0.0047807998955249785
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,3072,1,0.012294399738311767
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4096,128,0.1952191948890686
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4096,128,1.0344703674316407
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4096,64,0.0982815980911255
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4096,64,0.4830175876617432
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4096,32,0.050457602739334105
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4096,32,0.24073920249938965
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4096,16,0.02789120078086853
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4096,16,0.12403520345687866
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4096,8,0.011673600226640702
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4096,8,0.06479039788246155
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4096,4,0.007670400291681289
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4096,4,0.03355840146541596
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4096,2,0.005673599988222122
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4096,2,0.020470400154590607
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,4096,1,0.005027199909090996
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,4096,1,0.013827200233936309
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,6144,128,0.28078720569610593
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,6144,128,1.5227168083190918
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,6144,64,0.13288960456848145
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,6144,64,0.7397632122039794
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,6144,32,0.07100800275802613
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,6144,32,0.3596544027328491
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,6144,16,0.03956159949302673
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,6144,16,0.183843195438385
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,6144,8,0.021180799603462218
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,6144,8,0.09484480023384094
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,6144,4,0.009302400052547455
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,6144,4,0.04923200011253357
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,6144,2,0.006374400109052658
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,6144,2,0.026719999313354493
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,6144,1,0.005318399891257286
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,6144,1,0.017059199512004852
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8192,128,0.3780672073364258
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8192,128,2.0363840103149413
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8192,64,0.1772447943687439
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8192,64,0.9975839614868164
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8192,32,0.09253759980201721
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8192,32,0.4876255989074707
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8192,16,0.04946880042552948
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8192,16,0.24150400161743163
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8192,8,0.027190399169921876
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8192,8,0.12361600399017333
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8192,4,0.011449600011110306
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8192,4,0.06446400284767151
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8192,2,0.007545600086450577
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8192,2,0.03377279937267304
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,8192,1,0.005744000151753426
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,8192,1,0.02051199972629547
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,12288,128,0.5710847854614258
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,12288,128,3.067500877380371
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,12288,64,0.26314239501953124
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,12288,64,1.5162591934204102
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,12288,32,0.13167680501937867
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,12288,32,0.7527743816375733
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,12288,16,0.07060800194740295
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,12288,16,0.3599967956542969
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,12288,8,0.038806399703025816
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,12288,8,0.18477120399475097
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,12288,4,0.020921599864959717
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,12288,4,0.09506239891052246
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,12288,2,0.009193599969148637
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,12288,2,0.049481600522994995
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,12288,1,0.006521599739789963
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,12288,1,0.027014398574829103
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16384,128,0.7479072093963623
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16384,128,4.2343902587890625
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16384,64,0.36776959896087646
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16384,64,2.0410816192626955
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16384,32,0.17412480115890502
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16384,32,1.0311488151550292
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16384,16,0.09070079922676086
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16384,16,0.4853792190551758
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16384,8,0.04954560101032257
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16384,8,0.2415776014328003
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16384,4,0.02690559923648834
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16384,4,0.12380160093307495
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16384,2,0.011472000181674958
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16384,2,0.06474239826202392
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,16384,1,0.007715199887752533
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,16384,1,0.03374719917774201
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,20480,128,0.9431167602539062
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,20480,128,5.2944801330566404
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,20480,64,0.4561024188995361
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,20480,64,2.562601661682129
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,20480,32,0.21757121086120607
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,20480,32,1.2993887901306151
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,20480,16,0.11114879846572875
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,20480,16,0.6216959953308105
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,20480,8,0.05967680215835571
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,20480,8,0.3001568078994751
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,20480,4,0.0319680005311966
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,20480,4,0.1541632056236267
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,20480,2,0.016927999258041383
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,20480,2,0.07855679988861083
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,float16,20480,1,0.008316799998283386
SGLang,0.5.10,NVIDIA H200,mla_gen_post,default,fp8,20480,1,0.04166719913482666
