framework,version,device,op_name,kernel_source,mla_dtype,kv_cache_dtype,num_heads,batch_size,isl,tp_size,step,latency
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,1,1,0,0.33211733897527057
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,1,1,0,0.31726932525634766
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,1,16,0,0.33604268232981366
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,1,32,0,0.3423573176066081
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,1,4,0,0.32631466786066693
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,1,2,0,0.33109333117802936
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,1,16,0,0.3186346689860026
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,1,32,0,0.33604268232981366
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,1,8,0,0.32614399989446
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,1,1,0,0.32290132840474445
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,1,2,0,0.31675734122594196
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,1,4,0,0.31522132953008014
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,1,8,0,0.33450667063395184
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,1,16,0,0.31675734122594196
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,1,32,0,0.3170986572901408
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,1,2,0,0.3251199920972188
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,1,64,0,0.3444053332010905
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,1,1,0,0.3258026639620463
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,1,2,0,0.32255999247233075
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,1,4,0,0.32477867603302
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,1,4,0,0.33740798632303876
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,1,16,0,0.6114986737569174
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,1,8,0,0.320853332678477
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,1,64,0,0.32392533620198566
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,1,32,0,0.3280213276545207
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,1,8,0,0.3300693432490031
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,1,1,0,0.32307199637095135
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,1,2,0,0.3421866496404012
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,1,4,0,0.32733867565790814
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,1,8,0,0.35396265983581543
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,1,64,0,0.3399680058161418
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,1,16,0,0.32392533620198566
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,1,64,0,0.3415040175120036
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,1,32,0,0.3426986535390218
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,32,1,1,0,0.34065067768096924
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,32,1,2,0,0.32341333230336505
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,32,1,4,0,0.3275093237559001
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,32,1,8,0,0.34508800506591797
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,32,1,16,0,0.31778132915496826
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,32,1,32,0,0.3391146659851074
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,32,1,64,0,0.3188053369522095
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,64,1,1,0,0.314026673634847
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,1,64,0,0.3218773404757182
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,64,1,2,0,0.328874667485555
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,64,1,4,0,0.3394560019175212
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,64,1,16,0,0.33553067843119305
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,64,1,8,0,0.3275093237559001
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,64,1,32,0,0.33672531445821124
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,64,1,64,0,0.3380906581878662
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,128,1,1,0,0.32255999247233075
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,128,1,2,0,0.33058132727940875
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,128,1,4,0,0.3672746817270915
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,128,1,8,0,0.33587201436360675
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,128,1,16,0,0.3351893424987793
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,128,1,32,0,0.34201598167419434
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,128,1,64,0,0.3357013463973999
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,1,128,0,0.31675734122594196
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,128,1,128,0,0.31487999359766644
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,256,1,2,0,0.33297065893809
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,256,1,1,0,0.5707093477249146
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,256,1,4,0,0.33399466673533124
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,256,1,8,0,0.31829333305358887
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,256,1,16,0,0.33536001046498615
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,256,1,32,0,0.31914667288462323
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,256,1,64,0,0.36061867078145343
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,256,1,128,0,0.3380906581878662
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,16,1,0,0.2629973292350769
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,16,4,0,0.27101866404215497
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,16,2,0,0.26692267258961994
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,16,16,0,0.26316799720128375
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,16,32,0,0.27477333943049115
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,16,8,0,0.2606079975763957
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,16,64,0,0.27460267146428424
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,16,128,0,0.25565866629282635
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,16,1,0,0.25565866629282635
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,16,2,0,0.26282666126887005
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,16,8,0,0.2693120042483012
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,16,4,0,0.27426133553187054
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,16,16,0,0.2698240081469218
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,16,64,0,0.26897066831588745
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,16,32,0,0.27357866366704303
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,1,128,0,0.35140268007914227
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,16,128,0,0.26658133665720624
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,16,1,0,0.26948267221450806
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,16,4,0,0.26419200499852497
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,16,2,0,0.2730666597684224
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,16,16,0,0.2720426718393962
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,16,8,0,0.289792001247406
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,16,32,0,0.9376426537831625
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,16,64,0,0.2537813385327657
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,16,128,0,0.2486613392829895
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,16,2,0,0.24849067131678262
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,16,1,0,0.32392533620198566
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,16,4,0,0.27477333943049115
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,16,8,0,0.2688000003496806
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,16,16,0,0.2841599980990092
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,16,32,0,0.27153066794077557
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,16,64,0,0.25617067019144696
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,16,128,0,0.2577066620190938
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,16,1,0,0.2995199958483378
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,16,2,0,0.2706773281097412
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,16,4,0,0.2696533401807149
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,16,8,0,0.2681173284848531
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,16,16,0,0.26521599292755127
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,16,32,0,0.27084799607594806
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,16,64,0,0.2788693308830261
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,16,128,0,0.24780799945195517
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,32,16,2,0,0.2979840040206909
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,32,16,1,0,0.5355519851048788
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,32,16,4,0,0.2826240062713623
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,32,16,8,0,0.27477333943049115
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,32,16,16,0,0.27084799607594806
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,32,16,32,0,0.27460267146428424
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,32,16,128,0,0.25446399052937824
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,32,16,64,0,0.2696533401807149
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,64,16,2,0,0.5502293507258097
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,64,16,1,0,1.0159786542256672
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,64,16,4,0,0.30105600754419964
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,1,128,0,0.3242666721343994
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,64,16,8,0,0.26709334055582684
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,64,16,16,0,0.2532693346341451
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,64,16,128,0,0.2585600018501282
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,64,16,32,0,0.27767467498779297
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,64,16,64,0,0.2681173284848531
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,128,16,4,0,0.544426679611206
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,128,16,1,0,1.9805866877237956
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,128,16,8,0,0.30702932675679523
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,128,16,2,0,1.0190506776173909
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,128,16,16,0,0.27938133478164673
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,128,16,32,0,0.2720426718393962
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,128,16,64,0,0.27852799495061237
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,128,16,128,0,0.25514666239420575
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,256,16,8,0,0.5471573273340861
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,256,16,4,0,1.0308266480763753
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,256,16,2,0,1.9964586893717449
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,256,16,1,0,3.8910293579101562
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,256,16,16,0,0.31214932600657147
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,256,16,64,0,0.2775040070215861
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,256,16,32,0,0.2653866608937581
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,32,1,0,0.276309331258138
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,32,2,0,0.2778453429539998
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,32,4,0,0.26658133665720624
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,32,8,0,0.25514666239420575
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,32,32,0,0.27101866404215497
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,32,64,0,0.2717013359069824
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,32,16,0,0.2752853234608968
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,256,16,128,0,0.24388267596562704
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,32,128,0,0.8724479675292969
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,1,128,0,0.34986666838328045
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,32,2,0,0.2764799992243449
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,32,4,0,0.27409066756566364
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,32,1,0,0.26999467611312866
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,32,8,0,0.2727253238360087
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,32,64,0,0.2764799992243449
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,32,16,0,0.2744320034980774
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,32,32,0,0.2773333390553792
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,32,128,0,0.2841599980990092
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,32,2,0,0.27818665901819867
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,32,1,0,0.2717013359069824
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,32,4,0,0.3176106611887614
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,32,8,0,0.27562665939331055
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,1,128,0,0.3283626635869344
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,32,64,0,0.2744320034980774
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,32,16,0,0.6937599976857504
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,32,32,0,0.2797226707140605
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,32,128,0,0.2583893338839213
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,32,1,0,0.34628268082936603
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,32,2,0,0.27665066719055176
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,32,8,0,0.33399466673533124
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,32,32,0,0.31675734122594196
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,32,4,0,0.2698240081469218
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,32,16,0,0.2805759906768799
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,32,128,0,0.25514666239420575
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,32,64,0,0.8782506783803304
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,32,1,0,0.4288853406906128
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,32,4,0,0.27187200387318927
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,32,16,0,0.2788693308830261
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,32,8,0,0.2786986629168193
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,32,32,0,0.2792106668154399
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,32,2,0,0.28228267033894855
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,32,128,0,0.24576000372568765
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,32,64,0,0.2635093331336975
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,32,32,1,0,0.831829309463501
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,32,32,8,0,0.2728959918022156
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,32,32,2,0,0.4288853406906128
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,32,32,4,0,1.1951786677042644
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,32,32,32,0,0.2802346746126811
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,32,32,16,0,0.27084799607594806
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,32,1,128,0,0.3252906600634257
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,32,32,64,0,0.27409066756566364
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,32,32,128,0,0.2469546596209208
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,64,32,4,0,0.43212799231211346
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,64,32,8,0,0.8198826313018799
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,64,32,32,0,0.2727253238360087
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,64,32,2,0,0.7978666623433431
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,64,32,1,0,1.5312213897705078
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,64,32,16,0,0.27938133478164673
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,64,32,64,0,0.2773333390553792
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,64,32,128,0,0.24593067169189453
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,128,32,4,0,0.8063999811808268
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,128,32,16,0,0.2826240062713623
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,128,32,64,0,0.2701653242111206
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,128,32,32,0,0.27613866329193115
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,128,32,1,0,2.9924694697062173
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,128,32,8,0,0.44066135088602704
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,128,32,2,0,1.5329279899597168
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,128,32,128,0,0.24661332368850708
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,64,1,128,0,0.3454293409983317
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,256,32,4,0,1.5493119557698567
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,256,32,8,0,0.8261973063151041
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,256,32,16,0,0.45875199635823566
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,64,1,0,0.2757973273595174
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,256,32,32,0,0.2826240062713623
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,256,32,1,0,5.918890635172526
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,64,2,0,0.26265599330266315
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,256,32,2,0,3.0064640045166016
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,64,8,0,0.27511467536290485
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,64,4,0,0.2711893320083618
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,64,32,0,0.27050666014353436
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,64,16,0,0.2810879945755005
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,64,64,0,0.2788693308830261
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,64,128,0,0.26146133740743
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,64,2,0,0.2720426718393962
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,64,1,0,0.28279467423756915
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,64,4,0,0.28091732660929364
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,256,32,128,0,0.2527573307355245
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,64,16,0,0.2797226707140605
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,256,32,64,0,0.27801599105199176
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,64,8,0,0.899072011311849
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,64,32,0,0.2728959918022156
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,64,64,0,0.2768213351567586
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,64,128,0,0.28040534257888794
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,64,4,0,0.4935679833094279
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,64,2,0,0.2759679953257243
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,64,1,0,0.28091732660929364
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,64,8,0,0.27665066719055176
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,64,64,0,0.2797226707140605
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,64,32,0,0.272213339805603
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,64,16,0,0.27357866366704303
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,64,4,0,0.9200639724731445
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,64,8,0,1.2801706790924072
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,64,32,0,0.2720426718393962
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,64,16,0,1.5095465977986653
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,64,128,0,2.09715207417806
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,64,2,0,0.2727253238360087
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,64,1,0,0.39082666238149005
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,64,64,0,2.191871960957845
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,64,128,0,0.26606933275858563
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,64,2,0,0.44185598691304523
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,64,1,0,0.7292586962381998
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,64,4,0,0.2845013340314229
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,64,16,0,0.2730666597684224
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,64,8,0,0.6145706574122111
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,64,64,0,0.2775040070215861
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,64,32,0,0.279039998849233
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,64,128,0,0.26385066906611127
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,32,64,2,0,0.7014400164286295
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,32,64,4,0,0.3968000014623006
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,32,64,16,0,0.27477333943049115
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,32,64,8,0,0.6123520135879517
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,32,64,1,0,1.338197390238444
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,32,64,64,0,0.29627732435862225
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,32,64,32,0,0.28142933050791424
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,32,64,128,0,0.2600959936777751
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,64,64,8,0,0.38980265458424884
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,64,64,2,0,1.357141335805257
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,64,64,1,0,2.5842347145080566
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,64,64,64,0,0.27665066719055176
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,64,64,4,0,0.7157759666442871
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,64,64,32,0,0.2855253418286641
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,64,64,16,0,0.27562665939331055
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,64,64,128,0,0.32716800769170123
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,128,64,64,0,0.2788693308830261
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,128,64,32,0,1.6168959935506184
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,128,64,8,0,0.7432533105214437
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,128,64,16,0,0.41335467497507733
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,128,64,128,0,0.2573653260866801
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,128,64,4,0,1.3498026529947917
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,128,64,2,0,2.9315414428710938
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,128,64,1,0,5.105663935343425
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,256,64,16,0,0.7640746434529623
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,256,64,32,0,0.45260798931121826
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,256,64,64,0,0.29422932863235474
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,256,64,128,0,0.32153600454330444
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,256,64,8,0,1.5189332962036133
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,128,4,0,0.2734079957008362
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,128,2,0,0.279039998849233
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,256,64,2,0,5.113002777099609
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,128,1,0,0.27938133478164673
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,256,64,4,0,2.618879954020182
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,128,8,0,0.26794666051864624
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,128,16,0,0.27562665939331055
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,128,64,0,0.26948267221450806
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,128,32,0,1.6129706700642903
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,128,128,0,0.258730669816335
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,128,1,0,0.5474986632664999
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,128,4,0,0.2677759925524394
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,256,64,1,0,10.167637507120768
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,128,2,0,0.28330665826797485
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,128,8,0,0.9456640084584554
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,128,32,0,0.2744320034980774
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,128,16,0,0.2759679953257243
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,128,128,0,0.26077866554260254
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,128,2,0,0.28091732660929364
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,128,64,0,0.26999467611312866
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,128,1,0,0.37700267632802326
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,128,4,0,0.2805759906768799
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,128,16,0,0.2752853234608968
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,128,8,0,0.9437867005666097
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,128,128,0,1.0990933577219646
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,128,64,0,0.2810879945755005
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,128,32,0,0.274944007396698
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,128,2,0,0.37802668412526447
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,128,1,0,0.7272106806437174
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,128,4,0,0.9845759868621826
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,128,32,0,0.3476479848225911
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,128,64,0,0.9480533599853516
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,128,16,0,0.8171520233154297
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,128,128,0,0.26180267333984375
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,128,8,0,0.6106453339258829
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,128,2,0,0.6981973648071289
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,128,4,0,0.6307839949925741
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,128,16,0,0.27426133553187054
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,128,32,0,0.2795520027478536
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,128,1,0,1.3271040121714275
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,128,8,0,0.8108373483022054
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,128,64,0,0.27767467498779297
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,128,128,0,0.24627200762430826
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,32,128,16,0,0.27562665939331055
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,32,128,2,0,1.3218133449554443
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,32,128,4,0,0.7051946322123209
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,32,128,8,0,0.7871146996816
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,32,128,64,0,0.2797226707140605
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,32,128,32,0,0.276309331258138
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,32,128,128,0,0.26368000109990436
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,32,128,1,0,2.569728056589762
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,64,128,16,0,0.5237760146458944
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,64,128,2,0,2.5774079958597818
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,64,128,4,0,1.3400746981302898
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,64,128,8,0,0.7205546696980795
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,64,128,32,0,0.2764799992243449
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,64,128,64,0,0.27033599217732746
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,64,128,128,0,0.25753599405288696
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,64,128,1,0,5.070677439371745
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,128,128,16,0,0.7621973355611166
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,128,128,4,0,2.6129066149393716
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,128,128,8,0,1.3767679532368977
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,128,128,32,0,0.44646398226420086
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,128,128,64,0,0.32204800844192505
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,128,128,128,0,0.2640213370323181
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,128,128,2,0,5.081941286722819
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,128,128,1,0,10.13811175028483
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,256,128,32,0,0.839680035909017
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,256,128,16,0,1.4527146021525066
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,256,128,64,0,0.5307733217875162
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,256,128,8,0,2.675882657368978
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,256,128,4,0,5.147136052449544
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,256,1,0,0.27135999997456867
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,256,2,0,0.2788693308830261
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,256,4,0,0.2768213351567586
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,256,8,0,0.2769920031229655
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,256,16,0,0.27426133553187054
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,256,32,0,0.27426133553187054
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,256,128,2,0,10.122751871744791
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,256,64,0,0.9618773460388184
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,256,128,128,0,0.9485653241475424
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,256,128,0,0.2578773299853007
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,256,4,0,0.2725546757380168
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,256,1,0,0.3872426748275757
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,256,2,0,0.35020800431569415
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,256,16,0,0.27409066756566364
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,256,8,0,0.25548799832661945
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,256,128,0,0.254805326461792
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,256,32,0,0.27323732773462933
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,256,64,0,0.27938133478164673
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,256,2,0,0.38656000296274823
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,256,4,0,0.2855253418286641
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,256,8,0,0.2728959918022156
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,256,1,0,0.7046826680501302
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,256,128,0,0.2573653260866801
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,256,16,0,0.27801599105199176
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,256,64,0,0.27716267108917236
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,256,128,1,0,20.18457539876302
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,256,32,0,0.27357866366704303
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,256,2,0,0.7386453151702881
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,256,4,0,0.7268693447113037
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,256,16,0,0.27511467536290485
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,256,8,0,0.28040534257888794
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,256,32,0,0.5331626733144125
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,256,128,0,0.2629973292350769
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,256,64,0,0.2720426718393962
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,256,1,0,1.3341013590494792
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,256,2,0,1.3366613388061523
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,256,8,0,0.3954346577326457
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,256,4,0,0.7123626867930094
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,256,16,0,0.7070720195770264
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,256,1,0,2.580138683319092
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,256,64,0,0.2802346746126811
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,256,32,0,0.2727253238360087
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,256,128,0,0.25463465849558514
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,32,256,16,0,0.5910186767578125
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,32,256,8,0,0.754688024520874
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,32,256,32,0,0.2851840058962504
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,32,256,2,0,2.5929387410481772
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,32,256,64,0,0.30856533845265705
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,32,256,4,0,1.3528745969136555
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,32,256,128,0,0.24012800057729086
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,32,256,1,0,5.103103955586751
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,64,256,8,0,1.3849600156148274
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,64,256,16,0,0.7709013621012369
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,64,256,64,0,0.3015679915746053
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,64,256,4,0,2.6272427241007485
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,64,256,32,0,0.45550934473673504
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,64,256,2,0,5.10481071472168
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,64,256,128,0,0.2585600018501282
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,128,256,8,0,2.7054080963134766
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,128,256,16,0,1.4709760348002117
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,128,256,32,0,0.8528213500976562
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,64,256,1,0,10.13862419128418
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,128,256,4,0,5.190143903096517
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,128,256,64,0,0.5396480162938436
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,128,256,128,0,0.37802668412526447
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,128,256,2,0,10.19050661722819
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,128,256,1,0,20.246016184488933
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,256,256,32,0,1.641983985900879
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,256,256,8,0,5.339306513468425
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,256,256,16,0,2.874709447224935
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,256,256,64,0,1.0463573137919109
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,512,2,0,0.2872320016225179
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,512,1,0,0.4230826695760091
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,256,256,4,0,10.319189071655273
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,512,4,0,0.2797226707140605
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,512,8,0,0.27511467536290485
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,512,16,0,0.2728959918022156
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,512,32,0,0.2677759925524394
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,512,64,0,0.2775040070215861
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,512,128,0,0.2515626748402913
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,256,256,128,0,0.6980266571044922
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,512,8,0,0.9449813365936279
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,512,16,0,0.2764799992243449
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,512,4,0,0.2836479942003886
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,512,1,0,1.009493350982666
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,512,2,0,0.423253337542216
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,256,256,2,0,20.480511983235676
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,512,32,0,0.27613866329193115
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,512,64,0,0.28040534257888794
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,512,128,0,0.2850133379300435
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,512,2,0,0.7690239747365316
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,512,4,0,0.7575893402099609
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,512,8,0,0.28279467423756915
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,512,16,0,0.27767467498779297
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,512,1,0,1.4522026379903157
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,512,64,0,0.2711893320083618
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,512,32,0,0.2723840077718099
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,512,128,0,0.26333866516749066
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,512,8,0,0.4312746524810791
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,512,4,0,0.7796053091684977
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,512,2,0,1.4602239926656086
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,512,32,0,0.2788693308830261
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,512,16,0,0.2821120023727417
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,512,1,0,2.8339198430379233
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,512,64,0,0.2778453429539998
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,512,128,0,0.9285973707834879
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,512,16,0,0.452949325243632
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,512,32,0,0.27801599105199176
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,512,64,0,0.27818665901819867
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,512,4,0,1.4709760348002117
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,512,8,0,1.6269653638203938
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,512,2,0,2.8342612584431968
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,512,128,0,0.25924267371495563
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,256,256,1,0,40.55978647867838
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,512,1,0,5.573973337809245
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,32,512,32,0,0.4935679833094279
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,32,512,16,0,0.841386636098226
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,32,512,4,0,2.8683945337931314
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,32,512,8,0,1.5100587209065754
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,32,512,64,0,1.0103466510772705
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,32,512,128,0,0.26180267333984375
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,32,512,2,0,5.619712193806966
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,64,512,32,0,0.9186986287434896
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,64,512,64,0,0.5782186587651571
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,64,512,16,0,1.6027305920918782
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,64,512,8,0,2.943488121032715
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,32,512,1,0,11.095381418863932
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,64,512,4,0,5.681664148966472
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,64,512,128,0,0.39816534519195557
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,64,512,2,0,11.185834248860678
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,128,512,16,0,3.1180801391601562
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,128,512,32,0,1.764693260192871
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,128,512,8,0,5.8485762278238935
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,1024,2,0,0.5029546817143759
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,1024,1,0,0.9052159786224365
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,128,512,64,0,1.0859519640604656
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,1024,4,0,0.2974720001220703
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,128,512,128,0,0.7352320353190104
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,1024,8,0,0.28296534220377606
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,1024,16,0,0.27477333943049115
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,1024,32,0,0.2764799992243449
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,128,512,4,0,11.267925262451172
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,64,512,1,0,22.09996795654297
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,1024,64,0,0.285866657892863
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,1024,128,0,0.7987199624379476
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,1024,2,0,0.9101653099060059
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,1024,4,0,0.5039786497751871
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,1024,1,0,1.7121280034383137
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,1024,8,0,0.3012266755104065
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,1024,16,0,0.28381866216659546
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,1024,32,0,0.26692267258961994
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,1024,64,0,0.29098665714263916
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,1024,128,0,0.27187200387318927
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,1024,4,0,0.917674700419108
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,1024,2,0,1.7151999473571777
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,1024,8,0,0.5072213411331177
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,1024,16,0,0.30958932638168335
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,128,512,2,0,22.398635864257812
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,1024,32,0,0.274944007396698
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,1024,64,0,0.2863786617914836
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,1024,1,0,3.3262933095296225
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,1024,128,0,0.26129066944122314
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,1024,8,0,0.9323519865671793
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,1024,4,0,1.7353386878967285
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,1024,16,0,0.529578685760498
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,1024,32,0,0.32290132840474445
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,1024,64,0,0.279039998849233
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,1024,2,0,3.3464320500691733
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,1024,128,0,0.2513920068740845
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,1024,1,0,6.550698598225911
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,1024,16,0,0.9772373040517172
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,1024,8,0,1.7675946553548176
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,1024,32,0,0.571733315785726
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,1024,4,0,3.3704961140950522
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,1024,64,0,0.3667626778284709
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,1024,128,0,0.26385066906611127
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,1024,2,0,6.574933369954427
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,32,1024,16,0,1.8583893775939941
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,32,1024,8,0,3.4595839182535806
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,32,1024,32,0,1.0560853481292725
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,1024,1,0,13.05855941772461
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,32,1024,4,0,6.689109166463216
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,32,1024,64,0,0.6568959951400757
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,128,512,1,0,44.31189473470052
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,32,1024,128,0,0.4493653376897176
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,32,1024,2,0,13.071360270182291
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,64,1024,16,0,3.623594601949056
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,64,1024,64,0,1.2340906461079915
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,64,1024,8,0,6.8435624440511065
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,64,1024,128,0,0.8137386639912924
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,64,1024,32,0,2.0706987380981445
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,1536,2,0,0.8193706671396891
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,1536,4,0,0.463701327641805
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,1536,1,0,1.5162026087443035
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,1536,16,0,0.27187200387318927
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,1536,32,0,0.2706773281097412
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,1536,8,0,0.29337600866953534
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,1536,64,0,0.2769920031229655
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,64,1024,4,0,13.261824289957682
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,1536,128,0,0.2688000003496806
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,32,1024,1,0,26.03656514485677
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,1536,4,0,0.8229546546936035
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,1536,8,0,0.4700160026550293
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,1536,16,0,0.2949120004971822
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,1536,2,0,1.5202986399332683
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,1536,1,0,2.9023574193318686
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,1536,32,0,0.28484266996383667
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,1536,64,0,0.27613866329193115
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,1536,128,0,0.25702399015426636
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,1536,16,0,0.4800853331883748
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,1536,8,0,0.8321706453959147
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,1536,32,0,0.3089066743850708
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,1536,4,0,1.5245653788248699
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,1536,64,0,0.2792106668154399
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,1536,128,0,0.26146133740743
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,1536,2,0,2.9197654724121094
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,64,1024,2,0,26.186241149902344
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,1536,1,0,5.687466939290364
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,1536,32,0,0.5113173325856527
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,1536,8,0,1.5489706993103027
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,1536,16,0,0.8669866720835367
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,1536,4,0,2.932053248087565
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,1536,64,0,0.3331413269042969
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,1536,128,0,0.26845866441726685
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,1536,2,0,5.714944203694661
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,1536,32,0,0.9267199834187826
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,1536,16,0,1.6165547370910645
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,1536,8,0,2.9989547729492188
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,1536,64,0,0.5746346712112427
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,1536,4,0,5.761706670125325
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,1536,128,0,0.3911679983139038
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,1536,1,0,11.30239995320638
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,32,1536,16,0,3.1310507456461587
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,1536,2,0,11.35308837890625
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,32,1536,8,0,5.874858856201172
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,32,1536,32,0,1.747114658355713
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,32,1536,64,0,1.0574506918589275
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,32,1536,128,0,0.6929066975911459
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,32,1536,4,0,11.510784149169922
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,2048,2,0,1.1878399848937988
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,2048,1,0,2.2398293813069663
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,2048,4,0,0.6609919865926107
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,64,1024,1,0,52.086100260416664
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,2048,16,0,0.2764799992243449
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,2048,8,0,0.398848017056783
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,2048,32,0,0.274944007396698
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,2048,64,0,0.28040534257888794
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,2048,128,0,0.2686293323834737
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,1536,1,0,22.49676767985026
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,2048,16,0,0.40396801630655926
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,2048,4,0,1.195520003636678
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,2048,8,0,0.6652586857477824
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,2048,2,0,2.2568960189819336
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,2048,32,0,0.9545386632283529
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,2048,64,0,0.2783573269844055
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,2048,128,0,0.3002026677131653
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,32,1536,2,0,22.649002075195312
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,2048,1,0,4.339029312133789
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,2048,32,0,0.4198400179545085
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,2048,8,0,1.2096853256225586
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,2048,4,0,2.2615040143330893
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,2048,16,0,0.6857386430104574
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,2048,2,0,4.344661394755046
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,2048,64,0,0.2916693290074666
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,2048,128,0,0.2658986647923787
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,2048,1,0,8.538112004597982
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,2048,16,0,1.2506453196207683
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,2048,32,0,0.7273813088734945
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,2048,8,0,2.2910292943318686
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,2048,64,0,0.4621653159459432
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,2048,4,0,4.39517879486084
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,2048,128,0,0.32102400064468384
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,2048,2,0,8.604330698649088
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,2048,8,0,4.4508161544799805
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,2048,16,0,2.383530616760254
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,2048,64,0,0.8096426328023275
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,2048,32,0,1.33461332321167
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,32,1536,1,0,45.082112630208336
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,2048,4,0,8.650581359863281
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,2048,1,0,17.000106811523438
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,2048,128,0,0.5416959921518961
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,32,2048,32,0,2.5425920486450195
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,2048,2,0,17.00881067911784
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,32,2048,16,0,4.643840154012044
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,32,2048,64,0,1.4993066787719727
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,32,2048,8,0,8.801962534586588
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,32,2048,128,0,0.9675093491872152
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,3072,2,0,2.1084159215291343
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,3072,4,0,1.179477294286092
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,3072,8,0,0.6541653474171957
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,3072,1,0,4.070912043253581
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,3072,16,0,0.4150613149007161
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,3072,64,0,0.28808534145355225
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,3072,32,0,0.2993493278821309
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,3072,128,0,0.2759679953257243
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,3072,8,0,1.1617279847462971
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,3072,4,0,2.120533307393392
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,32,2048,4,0,17.191424051920574
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,3072,16,0,0.6702079772949219
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,3072,32,0,0.4288853406906128
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,3072,2,0,4.053162574768066
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,3072,64,0,0.2887679934501648
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,3072,128,0,0.27153066794077557
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,2048,1,0,33.918975830078125
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,3072,1,0,7.926272074381511
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,3072,8,0,2.154325326283773
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,3072,16,0,1.1866453488667805
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,3072,32,0,0.6964906851450602
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,3072,64,0,0.49083733558654785
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,3072,4,0,4.0762027104695635
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,3072,128,0,0.30498133103052777
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,3072,2,0,7.98139762878418
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,3072,8,0,4.138154665629069
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,3072,16,0,2.2125226656595864
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,3072,32,0,1.2497920195261638
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,3072,64,0,0.7594666481018066
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,3072,128,0,0.5102933247884115
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,3072,4,0,8.023040135701498
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,3072,1,0,15.700138092041016
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,32,2048,2,0,34.152618408203125
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,3072,2,0,15.828992207845053
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,3072,16,0,4.279637336730957
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,3072,64,0,1.3810346921284993
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,3072,32,0,2.3471786181131997
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,3072,8,0,8.1167360941569
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,3072,128,0,0.8837119738260905
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,4096,4,0,1.747114658355713
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,3072,4,0,15.91876220703125
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,4096,2,0,3.2766294479370117
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,4096,1,0,6.372181574503581
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,4096,8,0,0.9753599961598715
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,4096,16,0,0.5937493244806925
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,4096,32,0,0.38843735059102374
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,4096,64,0,0.32716800769170123
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,3072,1,0,31.263743082682293
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,4096,128,0,0.2990079919497172
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,4096,8,0,1.7638400395711262
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,4096,16,0,0.9963520367940267
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,4096,32,0,0.6055253346761068
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,4096,64,0,0.40908801555633545
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,4096,2,0,6.3556264241536455
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,4096,128,0,0.3304106593132019
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,4096,4,0,3.298645337422689
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,32,2048,1,0,67.69851684570312
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,3072,2,0,31.481343587239582
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,4096,8,0,3.3268054326375327
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,4096,1,0,12.53171157836914
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,4096,16,0,1.8027520179748535
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,4096,4,0,6.391808191935222
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,4096,32,0,1.035264015197754
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,4096,64,0,0.649727980295817
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,4096,128,0,0.4365653196970622
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,4096,2,0,12.518741607666016
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,4096,8,0,6.473728179931641
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,4096,16,0,3.4536107381184897
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,4096,32,0,1.8868907292683919
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,4096,64,0,1.1182080109914143
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,4096,128,0,0.7287466526031494
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,4096,4,0,12.591445922851562
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,4096,1,0,24.91357930501302
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,3072,1,0,62.546600341796875
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,16,4096,16,0,6.6658986409505205
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,4096,2,0,24.93286387125651
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,16,4096,64,0,2.0592640240987143
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,16,4096,128,0,1.2864853541056316
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,16,4096,32,0,3.5822931925455728
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,16,4096,8,0,12.77883783976237
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,6144,4,0,3.312981287638346
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,6144,8,0,1.8030932744344075
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,6144,16,0,1.0376533667246501
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,6144,2,0,6.332927703857422
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,6144,32,0,0.6638933420181274
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,6144,64,0,0.44236799081166583
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,6144,128,0,0.41198933124542236
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,16,4096,4,0,25.076222737630207
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,6144,1,0,12.463104248046875
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,6144,8,0,3.3349971771240234
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,6144,16,0,1.836202621459961
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,6144,32,0,1.071786642074585
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,6144,64,0,0.6877866586049398
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,6144,128,0,0.46353065967559814
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,6144,4,0,6.363818486531575
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,4096,1,0,49.4906005859375
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,6144,2,0,12.426069895426432
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,6144,16,0,3.4034347534179688
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,6144,8,0,6.4192854563395185
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,6144,64,0,1.1306666533152263
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,6144,32,0,1.8918399810791016
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,6144,4,0,12.481365203857422
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,6144,128,0,0.7458133697509766
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,16,4096,2,0,49.801900227864586
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,6144,1,0,24.539647420247395
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,6144,2,0,24.622421264648438
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,6144,16,0,6.550528208414714
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,6144,64,0,2.0164267222086587
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,6144,8,0,12.61636225382487
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,6144,32,0,3.528191884358724
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,6144,128,0,1.24945068359375
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,8192,4,0,5.357738494873047
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,6144,4,0,24.68420155843099
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,8192,8,0,2.8549121220906577
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,8192,2,0,10.37226676940918
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,8192,16,0,1.6035839716593425
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,8192,32,0,0.9774080117543539
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,8192,64,0,0.6446079810460409
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,8192,128,0,0.537770668665568
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,6144,1,0,49.17230733235677
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,8192,1,0,20.46907679239909
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,8192,8,0,5.420373280843099
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,8192,16,0,2.904575983683268
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,8192,4,0,10.410666783650717
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,8192,32,0,1.651370684305827
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,8192,64,0,1.020074685414632
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,8192,128,0,0.6734506289164225
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,6144,2,0,49.07520039876302
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,16,4096,1,0,98.95526123046875
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,8192,2,0,20.45422871907552
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,8192,32,0,2.982741355895996
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,8192,16,0,5.488639831542969
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,8192,64,0,1.7283412615458171
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,8192,128,0,1.09550937016805
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,8192,8,0,10.487637201944986
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,8192,4,0,20.528128306070965
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,8192,1,0,40.5580800374349
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,8,8192,16,0,10.657450358072916
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,8192,2,0,40.62719980875651
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,8,8192,8,0,20.678826649983723
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,8,8192,64,0,3.146239916483561
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,8,8192,128,0,1.8834773699442546
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,8,8192,32,0,5.6531626383463545
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,6144,1,0,97.74234008789062
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,10240,4,0,7.960917154947917
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,10240,8,0,4.18781852722168
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,10240,16,0,2.2912000020345054
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,10240,32,0,1.342634677886963
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,10240,64,0,0.8676693439483643
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,8,8192,4,0,40.83541361490885
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,10240,2,0,15.417855580647787
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,10240,128,0,0.6526293357213339
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,10240,8,0,7.952725092569987
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,10240,1,0,30.453590393066406
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,10240,16,0,4.218197186787923
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,10240,32,0,2.3420586585998535
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,8192,1,0,81.16958109537761
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,10240,4,0,15.441919962565104
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,10240,64,0,1.3969066937764485
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,10240,128,0,0.9152853488922119
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,10240,2,0,30.43891143798828
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,8,8192,2,0,81.06752014160156
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,10240,16,0,8.059221267700195
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,10240,8,0,15.534591674804688
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,10240,32,0,4.320597330729167
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,10240,128,0,1.5209813117980957
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,10240,64,0,2.459989388783773
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,10240,4,0,30.68433125813802
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,12288,8,0,5.726037343343099
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,10240,1,0,60.39159647623698
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,12288,32,0,1.7875626881917317
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,12288,16,0,3.1112534205118814
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,12288,4,0,10.981375376383463
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,12288,64,0,1.1340800126393635
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,12288,128,0,0.7772160371144613
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,12288,2,0,21.377365112304688
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,10240,2,0,60.49177551269531
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,12288,8,0,10.984619140625
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,12288,4,0,21.516459147135418
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,12288,32,0,3.1636479695638022
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,12288,16,0,5.765120188395183
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,12288,128,0,1.192618687947591
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,12288,64,0,1.8619732856750488
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,12288,1,0,42.28727467854818
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,12288,2,0,42.3200429280599
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,8,8192,1,0,161.40424601236978
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,12288,8,0,21.54376475016276
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,12288,32,0,5.895338694254558
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,12288,64,0,3.289087931315104
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,12288,16,0,11.106133778889975
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,12288,128,0,1.969322681427002
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,12288,4,0,42.66496022542318
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,10240,1,0,120.4648946126302
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,16384,4,0,18.43029276529948
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,16384,16,0,5.076138814290364
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,16384,8,0,9.569962819417318
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,16384,64,0,1.7271466255187988
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,16384,32,0,2.8487679163614907
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,16384,128,0,1.1538773377736409
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,12288,1,0,84.24993896484375
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,16384,2,0,36.24004364013672
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,16384,16,0,9.60972785949707
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,16384,8,0,18.49395243326823
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,16384,32,0,5.159594535827637
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,12288,2,0,84.35489908854167
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,16384,64,0,2.926762580871582
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,16384,128,0,1.8036053975423176
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,16384,4,0,36.50799051920573
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,16384,1,0,71.98344421386719
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,4,16384,16,0,18.69107182820638
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,16384,2,0,71.97440083821614
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,4,16384,32,0,9.782954533894857
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,4,16384,8,0,36.66261291503906
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,4,16384,128,0,3.108352025349935
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,4,16384,64,0,5.335552215576172
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,4,16384,4,0,72.22749837239583
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,12288,1,0,167.8726806640625
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,1,32768,16,0,17.79541269938151
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,1,32768,8,0,34.49463399251302
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,1,32768,32,0,9.439744313557943
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,1,32768,128,0,3.191808064778646
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,1,32768,64,0,5.252095858256022
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,16384,1,0,143.31340535481772
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,1,32768,4,0,67.94820149739583
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,4,16384,2,0,143.4760538736979
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,8,2,32768,16,0,34.664449055989586
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,1,32768,2,0,134.90756225585938
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,16,2,32768,8,0,68.27622477213542
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,2,2,32768,64,0,9.608874638875326
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,1,2,32768,128,0,5.409621556599935
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,4,2,32768,32,0,18.011136372884113
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,32,2,32768,4,0,134.98794555664062
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,4,16384,1,0,286.54591878255206
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,1,32768,1,0,270.0631103515625
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,64,2,32768,2,0,268.77695719401044
VLLM,0.12.0,NVIDIA A100-SXM4-80GB,context_mla,vllm_triton_mla,float16,float16,128,2,32768,1,0,536.1470947265625
