framework,version,device,op_name,kernel_source,bmm_dtype,num_tokens,num_heads,latency
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1,128,0.006060799956321717
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1,64,0.0047807998955249785
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1,128,0.008848000317811966
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1,32,0.007606399804353714
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1,16,0.00411520004272461
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1,16,0.007599999755620956
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1,8,0.004003199934959412
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1,8,0.007651200145483017
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1,4,0.0038975998759269714
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1,4,0.007433599978685379
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1,64,0.00793280005455017
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1,2,0.004208000004291534
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1,2,0.007340800017118454
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1,1,0.007315199822187424
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1,32,0.004383999854326248
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1,1,0.003852799907326698
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2,128,0.005958399921655655
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2,64,0.004800000041723251
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2,32,0.004227200150489807
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2,64,0.009027200192213059
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2,32,0.008044800162315369
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2,16,0.004108799993991852
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2,16,0.007897599786520004
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2,128,0.008828800171613693
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2,8,0.004195199906826019
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2,4,0.007449600100517273
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2,2,0.004268800094723702
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2,2,0.007836800068616867
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4,128,0.005907199904322624
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4,64,0.004912000149488449
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4,64,0.008700799942016602
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4,32,0.004259200021624565
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4,32,0.0084927998483181
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4,16,0.0043680001050233844
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4,16,0.007971200346946716
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2,4,0.0039680000394582745
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2,1,0.003932800143957138
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4,8,0.004249599948525429
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4,8,0.00761599987745285
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4,4,0.0039935998618602754
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4,2,0.004204799979925155
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4,2,0.007926400005817413
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4,1,0.00419199988245964
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4,1,0.0074592001736164095
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8,128,0.005795200169086456
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8,128,0.010476800054311753
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8,64,0.004822399839758873
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8,64,0.009151999652385712
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8,32,0.0042559999972581865
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8,32,0.00825600028038025
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8,16,0.004220800101757049
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8,16,0.008243200182914735
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8,8,0.003891199827194214
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8,8,0.0077344000339508055
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8,4,0.004073600098490715
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8,4,0.00756480023264885
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8,2,0.003872000053524971
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8,2,0.007651200145483017
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8,1,0.004137599840760231
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8,1,0.007599999755620956
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2,8,0.007203199714422226
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,16,128,0.006025600060820579
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,16,64,0.004886399954557419
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,16,16,0.004182400181889534
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2,1,0.007020799815654755
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,16,128,0.012303999811410903
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,16,64,0.009852799773216247
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4,128,0.009251199662685394
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,16,8,0.00820159986615181
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,16,16,0.008019199967384339
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,16,4,0.003932800143957138
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,16,32,0.005033599957823753
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,16,4,0.00716480016708374
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,16,2,0.00419199988245964
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4,4,0.007046400010585785
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,16,2,0.0076672002673149105
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,16,8,0.004150399938225746
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,16,1,0.003929600119590759
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,16,1,0.0070271998643875126
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,32,128,0.015423999726772308
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,32,64,0.005078399926424027
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,32,16,0.005183999985456466
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,32,64,0.01191679984331131
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,32,32,0.004358400031924248
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,32,32,0.009657599776983262
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,32,128,0.006150399893522262
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,32,16,0.008720000088214875
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,32,8,0.004112000018358231
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,32,8,0.007791999727487564
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,32,4,0.0042304001748561856
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,32,4,0.008115199953317642
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,32,2,0.004067200049757957
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,32,2,0.00772159993648529
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,32,1,0.004214400053024292
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,32,1,0.007711999863386154
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,48,128,0.0068512000143527985
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,48,128,0.018713599443435668
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,48,64,0.00535999983549118
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,48,64,0.013676799833774567
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,48,8,0.0040640000253915785
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,48,32,0.0046431999653577805
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,48,32,0.010979200154542923
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,48,16,0.004428799822926521
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,48,16,0.009331200271844864
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,48,8,0.008265600353479386
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,48,4,0.003964800015091896
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,16,32,0.00867839977145195
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,48,4,0.007878399640321731
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,48,2,0.004239999875426293
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,48,2,0.007964800298213958
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,48,1,0.0048351999372243885
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,48,1,0.007644800096750259
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,64,128,0.007545600086450577
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,64,128,0.022438399493694305
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,64,64,0.005811199918389321
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,64,64,0.015263999998569488
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,64,32,0.004742399975657463
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,64,32,0.011622399836778641
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,64,16,0.004316800087690353
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,64,2,0.004998400062322617
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,64,16,0.009612800180912017
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,64,8,0.003958399966359138
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,64,8,0.008531200140714646
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,64,4,0.004188799858093261
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,64,4,0.007952000200748443
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,64,2,0.008111999928951263
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,64,1,0.004793599992990494
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,64,1,0.007766400277614593
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,80,128,0.007891199737787246
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,80,128,0.025788798928260803
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,80,64,0.006041600182652473
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,80,64,0.017155200242996216
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,80,32,0.004966399818658829
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,80,32,0.01265919953584671
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,80,16,0.0048640001565217975
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,80,16,0.010047999769449234
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,80,8,0.004112000018358231
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,80,8,0.008700799942016602
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,80,1,0.007836800068616867
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,80,4,0.00398080013692379
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,80,4,0.008463999629020691
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,80,2,0.0039264000952243805
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,80,2,0.007872000336647034
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,80,1,0.0048767998814582825
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,96,128,0.008463999629020691
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,96,128,0.029267200827598573
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,96,64,0.006207999959588051
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,96,64,0.018438400328159334
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,96,32,0.005183999985456466
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,96,32,0.013174399733543396
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,96,16,0.004844800010323525
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,96,16,0.010486400127410889
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,96,8,0.004047999903559685
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,96,8,0.00899839997291565
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,96,4,0.004732799902558327
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,96,4,0.008406399935483932
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,96,2,0.004233599826693535
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,96,2,0.007983999699354172
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,96,1,0.004086399823427201
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,96,1,0.007836800068616867
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,128,128,0.010441599786281586
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,128,128,0.03654400110244751
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,128,64,0.00727040022611618
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,128,64,0.02229759991168976
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,128,32,0.005497600138187409
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,128,32,0.015091200172901154
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,128,4,0.008627200126647949
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,128,16,0.004543999955058098
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,128,16,0.011503999680280685
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,128,8,0.004239999875426293
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,128,8,0.009491200000047684
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,128,4,0.004150399938225746
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,128,2,0.004243199899792671
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,128,2,0.008111999928951263
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,128,1,0.003824000060558319
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,128,1,0.008191999793052674
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,160,128,0.012080000340938568
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,160,128,0.044284799695014955
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,160,64,0.007929600030183791
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,160,64,0.025699201226234435
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,160,32,0.005903999879956245
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,160,32,0.016892799735069276
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,160,16,0.0047680001705884935
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,160,16,0.012451200187206269
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,160,8,0.005033599957823753
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,160,8,0.010118400305509567
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,160,4,0.004076800122857094
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,160,4,0.008924800157546996
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,160,2,0.0047775998711586
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,160,2,0.008284799754619598
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,160,1,0.0038463998585939406
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,160,1,0.00782719999551773
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,192,128,0.016540800034999848
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,192,128,0.05314559936523437
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,192,64,0.008006399869918824
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,192,64,0.029017600417137145
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,192,32,0.006128000095486641
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,192,32,0.018361599743366243
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,192,16,0.0047520000487566
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,192,16,0.013209599256515502
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,192,8,0.0048096001148223875
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,192,8,0.010700800269842149
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,192,4,0.0040352001786232
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,192,4,0.009132800251245498
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,192,2,0.004147199913859367
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,192,2,0.008268799632787704
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,192,1,0.003779200091958046
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,192,1,0.008131200075149536
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,256,128,0.021689599752426146
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,256,128,0.06822400093078614
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,256,64,0.00952640026807785
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,256,64,0.03579840064048767
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,256,32,0.006838399916887283
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,256,16,0.005356800183653832
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,256,32,0.022012799978256226
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,256,16,0.015199999511241912
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,256,8,0.004374400153756142
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,256,8,0.01138240024447441
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,256,4,0.004399999976158142
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,256,4,0.0095040000975132
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,256,2,0.0049183998256921765
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,256,2,0.00843520015478134
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,256,1,0.003996799886226654
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,256,1,0.008060800284147263
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,320,128,0.023603199422359465
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,320,128,0.08298559784889221
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,320,64,0.01165120005607605
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,320,64,0.04525440037250519
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,320,32,0.0069760002195835115
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,320,32,0.024966399371623992
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,320,16,0.005872000008821487
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,320,16,0.016704000532627106
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,320,8,0.004774399846792221
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,320,8,0.01225920021533966
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,320,4,0.004294399917125702
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,320,4,0.01014079973101616
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,320,2,0.004009599983692169
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,320,2,0.008700799942016602
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,320,1,0.0040224000811576845
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,320,1,0.008284799754619598
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,384,128,0.02725439965724945
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,384,128,0.10015679597854614
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,384,64,0.01387840062379837
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,384,32,0.008217599987983704
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,384,64,0.053395199775695804
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,384,32,0.02945919930934906
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,384,16,0.006143999844789505
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,384,16,0.018403199315071107
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,384,8,0.004755200073122978
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,384,8,0.013033600151538849
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,384,4,0.004444799944758415
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,384,4,0.010204800218343735
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,384,2,0.004595199972391129
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,384,2,0.009318400174379349
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,384,1,0.004083200171589851
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,384,1,0.008268799632787704
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,512,128,0.032943999767303465
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,512,128,0.1248095989227295
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,512,64,0.018694399297237395
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,512,64,0.06697919964790344
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,512,32,0.009532800316810608
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,512,32,0.0358240008354187
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,512,16,0.006905599683523178
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,512,16,0.02197439968585968
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,512,8,0.005248000100255013
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,512,8,0.015065599977970124
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,512,4,0.004588799923658371
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,512,4,0.011395200341939925
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,512,2,0.004179200157523155
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,512,2,0.009728000313043595
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,512,1,0.004073600098490715
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,512,1,0.008700799942016602
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,768,128,0.044854399561882016
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,768,128,0.18314239978790284
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,768,64,0.02524479925632477
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,768,64,0.09927039742469787
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,768,32,0.013583999872207642
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,768,4,0.013212800025939941
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,768,32,0.051910400390625
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,768,16,0.008182399719953538
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,768,16,0.02910720109939575
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,768,8,0.005888000130653381
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,768,8,0.018492799997329713
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1024,128,0.05565440058708191
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,768,4,0.004755200073122978
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,768,2,0.0044064000248909
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,768,2,0.010416000336408614
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,768,1,0.004812800139188766
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,768,1,0.009167999774217606
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1024,128,0.2397696018218994
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1024,64,0.031651198863983154
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1024,64,0.12378879785537719
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1024,32,0.019196799397468566
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1024,32,0.06732800006866455
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1024,16,0.009542399644851684
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1024,16,0.0358815997838974
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1024,8,0.006892800331115723
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1024,8,0.021993599832057953
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1024,4,0.005363199859857559
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1024,4,0.014934399724006652
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1024,2,0.004694399982690811
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1024,2,0.011392000317573547
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1024,1,0.004211200028657913
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1024,1,0.009743999689817429
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1536,32,0.024425600469112397
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1536,128,0.07817919850349427
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1536,128,0.35653119087219237
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1536,64,0.0427839994430542
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1536,64,0.18281279802322387
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1536,32,0.09899200201034546
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1536,16,0.013731199502944946
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1536,16,0.05294399857521057
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1536,8,0.008271999657154083
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1536,8,0.02905600070953369
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1536,4,0.006108799949288368
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1536,4,0.01831679940223694
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1536,2,0.004899200052022934
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1536,2,0.013279999792575835
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,1536,1,0.004483199864625931
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,1536,1,0.01032319962978363
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2048,128,0.10092159509658813
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2048,128,0.4725759983062744
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2048,64,0.05406079888343811
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2048,64,0.23944640159606934
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2048,32,0.030873599648475646
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2048,16,0.017740799486637114
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2048,32,0.12360320091247559
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2048,16,0.06686080098152161
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2048,8,0.009622400254011154
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2048,8,0.03607679903507233
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2048,4,0.006924799829721451
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2048,4,0.02215999960899353
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2048,2,0.005225599929690361
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2048,2,0.015059199929237366
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,2048,1,0.004668800160288811
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,2048,1,0.011462400108575821
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,3072,128,0.14543039798736573
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,3072,64,0.0762336015701294
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,3072,128,0.7094016075134277
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,3072,64,0.3543008089065552
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,3072,32,0.042294400930404666
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,3072,32,0.18228160142898558
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,3072,16,0.024307200312614442
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,3072,16,0.09899200201034546
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,3072,8,0.013935999572277069
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,3072,8,0.05143359899520874
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,3072,4,0.008275199681520462
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,3072,4,0.029084798693656922
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,3072,2,0.005958399921655655
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,3072,2,0.01858240067958832
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,3072,1,0.004879999905824661
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,3072,1,0.01316480040550232
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4096,128,0.18825279474258422
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4096,64,0.09796479940414429
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4096,128,0.9442879676818847
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4096,32,0.053302401304244997
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4096,64,0.46964478492736816
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4096,32,0.23989439010620117
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4096,16,0.030089598894119263
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4096,16,0.1232640027999878
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4096,8,0.017561599612236023
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4096,8,0.06658880114555359
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4096,4,0.009545599669218063
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4096,4,0.03591679930686951
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4096,2,0.006815999746322632
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4096,2,0.02205760031938553
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,4096,1,0.00535999983549118
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,4096,1,0.015017600357532501
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,6144,128,0.2766144037246704
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,6144,64,0.1417888045310974
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,6144,32,0.0750656008720398
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,6144,64,0.7037856101989746
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,6144,128,1.4339743614196778
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,6144,32,0.35484158992767334
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,6144,16,0.04063040018081665
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,6144,16,0.18215680122375488
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,6144,8,0.02425280064344406
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,6144,8,0.09899200201034546
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,6144,4,0.013920000195503235
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,6144,4,0.05138239860534668
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,6144,2,0.008140800148248672
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,6144,2,0.029129600524902342
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,6144,1,0.0060639999806880954
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,6144,1,0.018287999927997588
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8192,64,0.943836784362793
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8192,128,0.3630847930908203
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8192,64,0.186080002784729
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8192,32,0.0969376027584076
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8192,128,1.906540870666504
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8192,16,0.052127999067306516
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8192,32,0.4664127826690674
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8192,16,0.23961279392242432
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8192,8,0.029788801074028017
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8192,8,0.1236575961112976
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8192,4,0.01686079949140549
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8192,4,0.06653439998626709
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8192,2,0.009571199864149093
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8192,2,0.03585279881954193
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,float16,8192,1,0.006726399809122085
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_pre,default,fp8,8192,1,0.02197439968585968
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1,128,0.0055263999849557875
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1,64,0.0047136001288890835
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1,128,0.008563199639320373
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1,32,0.004473600164055824
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1,64,0.007779199630022049
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1,32,0.007526399940252304
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1,16,0.004287999868392944
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1,16,0.007942400127649307
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1,8,0.006025600060820579
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1,4,0.004425600171089172
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1,8,0.008671999722719193
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1,4,0.007744000107049942
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1,2,0.007673600316047668
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1,2,0.004176000133156776
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1,1,0.007836800068616867
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1,1,0.0042720001190900804
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2,128,0.0056928001344203946
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2,64,0.005023999884724617
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2,128,0.008832000195980072
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2,32,0.0046431999653577805
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2,64,0.008777599781751633
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2,16,0.0043680001050233844
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2,32,0.00823040008544922
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2,16,0.00806720033288002
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2,8,0.008867199718952178
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2,8,0.005987200140953064
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2,4,0.004396799951791763
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2,4,0.007657600194215774
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2,2,0.007680000364780426
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2,2,0.004239999875426293
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2,1,0.00445760004222393
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2,1,0.007795199751853943
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4,128,0.005686400085687637
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4,64,0.004953600093722343
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4,128,0.009356799721717834
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4,64,0.00873280018568039
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4,32,0.0046847999095916745
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4,32,0.008528000116348267
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4,16,0.004454400017857551
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4,16,0.008038400113582611
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4,8,0.006095999851822853
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4,8,0.008444800227880477
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4,4,0.007769600301980972
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4,2,0.004281599819660187
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4,1,0.004211200028657913
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4,1,0.007660800218582153
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8,128,0.00565119981765747
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8,64,0.004729599878191948
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8,16,0.004492799937725067
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8,16,0.008457600325345992
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8,8,0.004383999854326248
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8,8,0.0080704003572464
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8,4,0.004332799836993218
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8,4,0.008131200075149536
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8,2,0.004348799958825112
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8,2,0.007753600180149078
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8,1,0.00432640016078949
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8,1,0.00769599974155426
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16,128,0.005798399820923806
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4,4,0.004383999854326248
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16,64,0.004704000055789947
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16,32,0.004467200115323066
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8,32,0.005222399905323982
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16,32,0.008995199948549271
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16,16,0.005190400034189224
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16,128,0.012188799679279327
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16,16,0.008271999657154083
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16,8,0.005145600065588951
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16,8,0.008179199695587159
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16,4,0.00424639992415905
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16,2,0.0043680001050233844
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16,2,0.007974400371313094
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16,1,0.00424639992415905
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16,1,0.007568000257015229
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,32,128,0.006143999844789505
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,32,128,0.015267199277877808
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,32,64,0.0049183998256921765
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,32,64,0.011584000289440155
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,32,32,0.004595199972391129
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,32,32,0.009593600034713745
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,32,16,0.004633599892258644
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,32,16,0.008806400001049042
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,32,8,0.004428799822926521
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,32,8,0.008620800077915191
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,32,4,0.004396799951791763
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,32,4,0.008419200032949447
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,32,2,0.004425600171089172
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,32,2,0.007919999957084655
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,32,1,0.004374400153756142
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,32,1,0.008076799660921096
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,48,128,0.006697600334882736
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4,2,0.007100799679756164
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,48,64,0.005190400034189224
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,48,128,0.018243199586868285
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,48,32,0.004694399982690811
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,48,64,0.012761600315570831
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,48,32,0.010745599865913391
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,48,16,0.004515200108289719
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,48,8,0.004441599920392036
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,48,16,0.008816000074148178
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,48,8,0.008374399691820144
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16,64,0.009436800330877303
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,48,4,0.004265600070357323
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,48,4,0.008099199831485748
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8,64,0.008531200140714646
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,48,2,0.0043935999274253845
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8,128,0.009862399846315383
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8,32,0.007897599786520004
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,48,2,0.008105599880218506
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,48,1,0.004275200143456459
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,48,1,0.007478400319814682
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,64,128,0.006870400160551071
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,64,128,0.021030400693416596
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16,4,0.007993599772453308
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,64,64,0.014867199957370758
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,64,32,0.004566400125622749
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,64,32,0.011142399907112122
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,64,16,0.004691199958324432
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,64,16,0.009279999881982803
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,64,8,0.005116799846291542
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,64,8,0.008931200206279754
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,64,4,0.004284799844026565
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,64,64,0.005273599922657013
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,64,4,0.008284799754619598
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,64,2,0.0042720001190900804
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,64,2,0.008271999657154083
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,64,1,0.005049600079655647
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,64,1,0.008009599894285202
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,80,128,0.007254400104284286
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,80,16,0.005302400141954422
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,80,128,0.0243136003613472
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,80,64,0.016371199488639833
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,80,32,0.012361600250005721
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,80,16,0.010380800068378448
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,80,8,0.004633599892258644
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,80,8,0.008950400352478027
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,80,4,0.004329600185155868
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,80,4,0.008531200140714646
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,80,64,0.005660799890756607
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,80,2,0.008454400300979614
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,80,32,0.004774399846792221
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,80,1,0.004403200000524521
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,96,128,0.008009599894285202
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,80,1,0.00822720006108284
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,96,64,0.005711999908089638
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,96,64,0.0179967999458313
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,96,128,0.027875199913978577
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,96,32,0.004732799902558327
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,80,2,0.004211200028657913
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,96,32,0.013152000308036805
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,96,16,0.004758400097489357
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,96,16,0.01080000028014183
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,96,8,0.004720000177621841
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,96,8,0.009254399687051773
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,96,4,0.0044351998716592785
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,96,4,0.008291199803352356
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,96,2,0.0042015999555587765
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,96,2,0.008364800363779068
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,96,1,0.004307200014591217
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,96,1,0.00828159973025322
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,128,128,0.00835840031504631
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,128,128,0.03543039858341217
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,128,64,0.006246399879455566
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,128,64,0.021328000724315642
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,128,32,0.00485760010778904
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,128,32,0.014793600142002105
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,128,16,0.004608000069856644
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,128,16,0.011219199746847153
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,128,8,0.00461760014295578
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,128,8,0.009836799651384353
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,128,4,0.004416000097990036
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,128,4,0.00896959975361824
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,128,2,0.004233599826693535
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,128,2,0.008367999643087386
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,128,1,0.004278400167822838
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,128,1,0.00836160033941269
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,160,128,0.009404800087213516
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,160,128,0.044207999110221864
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,160,64,0.006646399945020675
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,160,64,0.024233600497245787
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,160,32,0.005238400027155876
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,160,32,0.016467200219631196
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,160,4,0.004419200122356415
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,160,16,0.004656000062823296
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,160,8,0.004838399961590767
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,160,16,0.012374400347471236
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,160,8,0.010320000350475311
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,160,4,0.008979199826717377
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,160,2,0.004239999875426293
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,160,2,0.008659200370311737
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,160,1,0.004335999861359597
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,160,1,0.008083199709653854
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,192,128,0.012943999469280243
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,192,128,0.05170239806175232
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,192,64,0.006995200365781784
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,192,32,0.005443200096487999
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,192,64,0.027728000283241273
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,192,16,0.004704000055789947
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,192,32,0.018035200238227845
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,192,8,0.004668800160288811
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,192,16,0.01321599930524826
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,192,4,0.004483199864625931
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,192,8,0.010860799998044967
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,192,1,0.008303999900817871
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,192,4,0.009507200121879578
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,192,2,0.008556800335645676
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,192,2,0.004387199878692627
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,192,1,0.004377600178122521
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,256,128,0.01895039975643158
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,256,128,0.06590399742126465
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,256,64,0.007599999755620956
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,256,32,0.005644800141453743
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,256,64,0.034995201230049136
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,256,32,0.02099519968032837
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,256,16,0.004966399818658829
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,256,16,0.014684799313545226
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,256,8,0.004975999891757965
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,256,8,0.011507199704647064
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,256,4,0.004454400017857551
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,256,2,0.004524800181388855
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,256,4,0.009753599762916565
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,256,1,0.004361600056290627
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,256,2,0.009283199906349182
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,256,1,0.0086496002972126
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,320,128,0.022038400173187256
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,320,128,0.0803167998790741
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,320,64,0.008473599702119828
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,320,64,0.042985600233078
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,320,32,0.006681600213050842
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,320,32,0.024646399915218352
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,320,16,0.00527999997138977
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,320,16,0.016143999993801117
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,320,8,0.0047136001288890835
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,320,4,0.004550400003790855
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,320,2,0.004540799930691719
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,320,8,0.01226240023970604
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,320,4,0.01064639985561371
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,320,2,0.009187199920415879
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,320,1,0.004243199899792671
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,320,1,0.0086496002972126
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,384,128,0.024748800694942473
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,384,128,0.09743040204048156
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,384,64,0.009587199985980987
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,384,64,0.050518399477005003
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,384,32,0.00658240020275116
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,384,32,0.027584001421928406
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,384,16,0.005215999856591225
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,384,16,0.017926399409770966
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,384,8,0.004806400090456009
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,384,8,0.013193599879741669
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,384,4,0.005327999964356423
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,384,2,0.004422400146722794
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,512,128,0.12608959674835205
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,384,4,0.010454399883747101
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,384,2,0.009340800344944
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,384,1,0.004454400017857551
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,384,1,0.008662399649620057
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,512,128,0.03073279857635498
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,512,64,0.015564799308776855
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,512,64,0.06615039706230164
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,512,32,0.007753600180149078
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,512,32,0.034176000952720643
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,512,16,0.005612799897789955
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,512,16,0.021135999262332915
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,512,8,0.004902400076389313
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,512,8,0.014735999703407287
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,512,4,0.005084799975156784
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,512,4,0.011580800265073776
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,512,2,0.004598399996757508
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,512,2,0.009679999947547913
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,512,1,0.0049727998673915865
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,512,1,0.008726400136947633
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,768,128,0.04288960099220276
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,768,128,0.18437119722366332
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,768,64,0.022793599963188173
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,768,64,0.09617599844932556
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,768,32,0.009225600212812424
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,768,32,0.05004159808158874
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,768,16,0.006364800035953522
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,768,16,0.027344000339508057
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,768,8,0.005286400020122528
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,768,8,0.017849600315093993
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,768,4,0.004985599964857102
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,768,4,0.01308799982070923
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,768,2,0.0046016000211238865
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,768,2,0.010556799918413162
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,768,1,0.0052767999470233916
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,768,1,0.009375999867916106
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1024,128,0.05330560207366943
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1024,64,0.028960001468658448
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1024,128,0.24242560863494872
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1024,64,0.1250432014465332
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1024,32,0.012185599654912949
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1024,32,0.06561279892921448
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1024,16,0.007702399790287018
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1024,16,0.034281599521636966
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1024,8,0.0055263999849557875
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1024,8,0.021238400042057036
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1024,4,0.005008000135421753
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1024,4,0.014655999839305878
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1024,2,0.004745600000023842
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1024,2,0.011606399714946748
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1024,1,0.005094400048255921
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1024,1,0.009926400333642959
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1536,128,0.07446399927139283
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1536,64,0.040380799770355226
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1536,128,0.3617791891098022
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1536,64,0.18447999954223632
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1536,32,0.021792000532150267
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1536,32,0.09598720073699951
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1536,16,0.009139200299978256
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1536,16,0.04990079998970032
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1536,8,0.006460800021886826
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1536,2,0.01321599930524826
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1536,8,0.027616000175476073
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1536,4,0.005135999992489815
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1536,4,0.01800000071525574
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1536,2,0.005094400048255921
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,1536,1,0.005113599821925163
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,1536,1,0.010652799904346467
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2048,128,0.09747200012207032
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2048,64,0.051900798082351686
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2048,128,0.48905601501464846
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2048,64,0.24181759357452393
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2048,32,0.028044798970222475
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2048,32,0.12437119483947753
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2048,16,0.011587200313806533
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2048,16,0.0649407982826233
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2048,8,0.0077344000339508055
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2048,8,0.03461759984493255
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2048,4,0.005558399856090546
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,3072,128,0.14210879802703857
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2048,4,0.020956799387931824
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2048,2,0.004841599985957146
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2048,2,0.014761599898338317
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,2048,1,0.0054143998771905896
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,2048,1,0.011436799913644791
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,3072,64,0.07174720168113709
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,3072,128,0.7545504093170166
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,3072,32,0.041126400232315063
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,3072,8,0.050121599435806276
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,3072,4,0.006550399959087372
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,3072,64,0.36046719551086426
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,3072,32,0.1842527985572815
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,3072,1,0.004812800139188766
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,3072,16,0.021129600703716278
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,3072,16,0.09520000219345093
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,3072,8,0.009139200299978256
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,3072,4,0.027619200944900512
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,3072,2,0.005248000100255013
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4096,64,0.4874239921569824
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,3072,2,0.017977599799633027
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4096,32,0.24127678871154784
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,3072,1,0.012982399761676788
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4096,128,0.19621440172195434
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4096,64,0.09189760088920593
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4096,128,1.0232352256774901
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4096,32,0.050444799661636355
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4096,16,0.027590399980545043
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4096,16,0.12433279752731323
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4096,8,0.011539199948310852
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4096,8,0.06548479795455933
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4096,4,0.007756800204515457
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4096,4,0.034169599413871765
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4096,2,0.005462399870157242
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4096,2,0.021129600703716278
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,4096,1,0.004927999898791313
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,4096,1,0.014838400483131408
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,6144,128,0.2863424062728882
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,6144,64,0.13373440504074097
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,6144,32,0.07065600156784058
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,6144,64,0.7492671966552734
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,6144,8,0.0951807975769043
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,6144,128,1.5630175590515136
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,6144,32,0.35622398853302
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,6144,16,0.039315199851989745
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,6144,16,0.18353919982910155
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,6144,8,0.02280319929122925
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,6144,4,0.009113600105047226
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,6144,4,0.0496832013130188
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,6144,2,0.006294400244951248
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,6144,2,0.027744001150131224
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,6144,1,0.005321599915623665
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,6144,1,0.01769919991493225
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8192,128,0.37655680179595946
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8192,64,0.18139840364456178
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8192,32,0.09178239703178406
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8192,8,0.0292959988117218
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8192,64,1.016476821899414
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8192,32,0.4792960166931152
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8192,16,0.05083199739456177
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8192,128,2.10196475982666
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8192,1,0.005552000179886818
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8192,16,0.24012160301208496
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8192,8,0.12445119619369507
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8192,4,0.011443199962377549
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8192,4,0.06534079909324646
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,8192,2,0.007574400305747986
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8192,2,0.034108799695968625
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,8192,1,0.021222400665283202
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,12288,128,0.5731359958648682
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,12288,64,0.2554816007614136
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,12288,32,0.13213440179824829
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,12288,16,0.07037439942359924
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,12288,32,0.7349408149719239
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,12288,64,1.513209629058838
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,12288,8,0.041529598832130435
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,12288,128,3.168441581726074
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,12288,16,0.36588799953460693
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,12288,8,0.1844383955001831
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,12288,4,0.020598399639129638
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,12288,4,0.095551997423172
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,12288,2,0.009091199934482574
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,12288,2,0.04976640045642853
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,12288,1,0.006415999680757523
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,12288,1,0.027622398734092713
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16384,128,0.7178944110870361
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16384,64,0.3371903896331787
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16384,32,0.17240959405899048
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16384,16,0.09219520092010498
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16384,32,0.9968064308166504
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16384,64,2.0284351348876952
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16384,8,0.05278720259666443
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16384,4,0.12391999959945679
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16384,16,0.48204479217529295
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16384,8,0.2411423921585083
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16384,4,0.027251198887825012
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16384,128,4.224863815307617
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16384,2,0.011219199746847153
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16384,2,0.06525120139122009
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,16384,1,0.007552000135183335
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,16384,1,0.03482879996299744
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,20480,64,0.4195871829986572
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,20480,128,0.879248046875
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,20480,32,0.2122431993484497
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,20480,16,0.1104159951210022
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,20480,32,1.2562368392944336
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,20480,16,0.6047200202941895
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,20480,8,0.06403840184211732
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,20480,64,2.6351999282836913
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,20480,8,0.3019423961639404
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,20480,4,0.0349151998758316
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,20480,2,0.017788800597190856
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,20480,4,0.1541440010070801
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,20480,128,5.095593643188477
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,20480,2,0.07918400168418885
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,float16,20480,1,0.00836160033941269
SGLang,0.5.6.post2,NVIDIA H200,mla_gen_post,default,fp8,20480,1,0.04256640076637268
