framework,version,device,op_name,kernel_source,mla_dtype,kv_cache_dtype,num_heads,batch_size,isl,tp_size,step,latency
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,1,0.32315733035405475
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,1,0.30324800809224445
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,1,0.3399786551793416
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,1,0.3030719955762227
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,1,0.310805340607961
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,1,0.3052320082982381
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,1,0.3359413146972656
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,1,0.3035999933878581
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,1,0.30030399560928345
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,1,0.3203679919242859
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,1,0.3158133427302043
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,1,0.3081546624501546
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,1,0.336298664410909
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,1,0.3272426724433899
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,1,0.31139200925827026
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,1,0.31909332672754925
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,1,0.3061013420422872
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,1,0.31334400177001953
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,1,0.33403201897939044
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,1,0.3270240028699239
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,1,0.3120479981104533
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,1,0.31109867493311566
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,1,0.3356746832529704
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,1,0.5392906665802002
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,1,0.3378293514251709
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,1,0.33897598584493
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,1,0.3465119997660319
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,1,0.3412853479385376
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,1,0.3476426601409912
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,1,0.45523734887441
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,1,0.34063998858133954
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,1,0.3155786593755086
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,1,0.3433813254038493
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,1,0.31152000029881793
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,1,0.6856106917063395
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,1,0.3396960099538167
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,1,0.37059199810028076
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,1,0.3135253389676412
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,1,0.3436319828033447
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,1,0.36079998811086017
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,1,0.34964799880981445
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,1,0.3431306680043538
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,1,0.34578665097554523
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,1,0.31835732857386273
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,1,0.38519465923309326
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,1,0.33461864789326984
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,1,0.3428906599680583
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,1,0.3404586712519328
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,1,0.3466346661249797
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,1,0.3172053297360738
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,1,0.3401920000712077
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,1,0.3221279978752136
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,1,0.3450239896774292
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,1,0.3120479981104533
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,1,0.30990399916966754
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,1,0.3240373333295186
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,1,0.33795734246571857
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,1,0.3439519802729289
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,1,0.3470880190531413
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,1,1.3093173503875732
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,1,0.31639466683069867
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,1,0.3259626626968384
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,1,0.3113386631011963
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,1,0.3497973283131917
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,1,0.3473333517710368
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,1,1.334287961324056
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,1,0.3176640073458354
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,1,0.35146133104960126
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,1,0.31303999821345013
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,1,0.30661332607269287
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,1,0.30587200323740643
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,1,3.1602773666381836
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,1,0.6952640215555826
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,1,0.6384053230285645
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,1,0.32707200447718304
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,1,1.6210346221923828
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,1,0.3070346713066101
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,1,0.33986131350199383
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,1,0.31298667192459106
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,1,0.30343466997146606
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,1,1.627295970916748
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,1,0.6624746719996134
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,1,0.3431679805119832
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,3,0.3408586581548055
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,1,3.5139999389648438
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,3,0.34167468547821045
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,3,0.3409279982248942
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,1,0.3482666810353597
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,1,0.3487146695454915
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,3,1.3431413968404133
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,3,0.3452746470769246
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,3,0.3402719895044963
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,3,0.3388213316599528
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,1,0.562064011891683
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,3,0.3457119862238566
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,3,0.3415093421936035
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,3,0.37852799892425537
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,3,0.35394132137298584
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,3,0.3470453421274821
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,3,0.33291200796763104
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,3,0.37007999420166016
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,3,0.34436798095703125
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,3,0.3454720179239909
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,3,0.34849599997202557
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,1,0.3423839807510376
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,3,0.3444693485895793
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,3,0.32329599062601727
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,3,0.3441280126571655
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,3,0.3484319845835368
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,3,0.3437013228734334
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,3,0.345792015393575
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,3,0.3925333420435588
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,3,0.31648000081380206
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,3,0.3123146692911784
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,3,0.3426506519317627
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,3,0.3525386651357015
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,3,0.3504800001780192
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,3,0.337775985399882
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,3,0.6870293617248535
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,3,0.34557334582010907
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,3,0.3454506794611613
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,3,0.34578665097554523
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,3,0.34266666571299237
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,3,0.3481920162836711
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,3,0.3437653382619222
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,3,0.3493013381958008
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,3,0.3683946530024211
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,3,0.3461600144704183
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,3,0.36152533690134686
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,3,0.35174401601155597
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,3,0.355951984723409
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,3,0.3470240036646525
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,3,0.3345173199971517
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,3,0.3439146677652995
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,3,0.35394132137298584
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,3,0.3418293396631877
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,3,0.3447680075963338
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,3,0.34195733070373535
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,3,0.35017601648966473
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,3,0.342522660891215
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,3,0.34036799271901447
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,3,0.39311468601226807
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,3,0.34145065148671466
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,3,0.31307733058929443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,3,0.49697601795196533
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,3,0.8137226899464926
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,3,0.3492000102996826
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,3,0.35330132643381756
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,3,0.35339200496673584
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,3,0.32222400108973187
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,3,0.34570133686065674
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,3,0.31807466348012287
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,3,0.35209067662556964
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,3,0.4654666582743327
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,3,0.3391520182291667
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,3,1.3066879908243816
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,3,0.3479573329289754
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,3,0.6381653149922689
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,3,0.34459733963012695
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,3,0.30346665779749554
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,3,0.5158400138219198
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,3,0.3448266585667928
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,3,0.3482026656468709
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,3,1.2773706912994385
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,3,0.46107200781504315
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,3,2.833973248799642
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,3,0.3065066734949748
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,3,0.3346133232116699
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,3,2.974442799886068
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,3,0.342901349067688
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,3,0.4535520076751709
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,3,0.3235306739807129
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,3,2.8106559117635093
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,7,0.3349173466364543
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,7,0.3345653216044108
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,7,0.337936004002889
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,3,5.6353333791097
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,7,0.34539198875427246
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,7,0.34034132957458496
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,7,0.8927679856618246
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,7,0.3320586681365967
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,7,0.34042131900787354
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,7,0.3507466713587443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,7,0.34166399637858075
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,7,0.3521173397699992
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,7,1.3025013605753581
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,7,1.240399996439616
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,7,0.34415467580159503
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,3,0.34590399265289307
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,7,0.34697067737579346
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,7,0.3394613265991211
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,7,1.413599967956543
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,7,0.34346667925516766
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,7,0.3365279833475749
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,7,0.3444960117340088
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,7,0.3146773378054301
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,7,0.3118293285369873
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,7,0.30852800607681274
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,7,0.3367413282394409
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,7,0.3452320098876953
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,7,0.3494400183359782
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,7,1.2332853476206462
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,7,0.32890133062998456
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,7,0.3216853340466817
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,7,1.3426613807678223
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,7,0.34005868434906006
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,7,0.3509813149770101
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,7,0.34141866366068524
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,7,0.3174453377723694
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,7,0.3059840003649394
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,7,1.0144906838734944
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,7,0.3471573193868001
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,7,0.3521813154220581
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,7,0.3497973283131917
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,7,0.3428906599680583
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,7,0.3133920033772786
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,7,0.35166935125986737
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,7,0.36959465344746906
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,7,0.3365866740544637
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,3,1.1558559735616047
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,7,0.35461334387461346
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,7,0.35224000612894696
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,7,0.34833065668741864
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,7,0.3092426657676697
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,7,0.35170666376749676
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,7,0.35132265090942383
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,7,0.349237322807312
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,7,0.3497120141983032
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,7,0.3322400053342183
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,7,0.3540906508763631
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,7,0.3227893312772115
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,7,0.3126773238182068
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,7,0.3498400052388509
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,7,0.3468426863352458
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,7,0.3489439884821574
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,7,0.6213440100351969
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,7,0.35655999183654785
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,7,0.3508960008621216
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,7,0.32595733801523846
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,7,0.3090826670328776
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,7,0.34962133566538495
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,7,0.6507200002670288
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,7,0.3495413462320964
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,7,0.34837865829467773
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,7,0.35098667939503986
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,7,1.5710612932840984
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,7,0.3101653258005778
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,7,0.33442668120066327
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,7,1.5496959686279297
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,7,0.3505760033925374
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,7,0.3412160078684489
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,7,0.5979146560033163
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,7,3.3686561584472656
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,7,0.3436853488286336
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,7,0.3399786551793416
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,7,0.3113173246383667
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,7,1.5480213165283203
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,7,3.3320480982462564
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,15,0.33852267265319824
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,7,0.5915679931640625
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,7,0.3484640121459961
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,7,6.7152055104573565
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,15,0.33267199993133545
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,15,1.7540480295817058
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,15,0.34409066041310626
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,15,0.34277331829071045
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,7,0.37995731830596924
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,15,0.3421279986699422
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,15,0.3354666630427043
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,15,0.38204801082611084
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,7,0.3543359835942586
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,15,0.3447573184967041
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,15,0.34840532143910724
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,15,0.34082667032877606
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,15,0.7885013421376547
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,15,0.3478826681772868
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,15,0.34512531757354736
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,15,0.34834667046864826
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,7,0.35983999570210773
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,15,0.5185279846191406
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,15,0.3404853343963623
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,15,3.021216074625651
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,15,0.3182133237520854
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,15,0.34409066041310626
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,15,0.35157867272694904
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,15,0.3462933301925659
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,15,0.3664906819661458
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,15,0.38017598787943524
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,15,0.3489333391189575
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,15,2.0381600062052407
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,15,0.35336001714070636
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,15,1.3965333302815754
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,15,0.35228268305460614
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,15,0.3512800137201945
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,15,0.3142399986584981
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,15,0.3518773317337036
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,15,0.3457653522491455
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,15,0.35091201464335126
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,15,0.3458240032196045
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,15,0.3510986566543579
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,15,0.3516853253046672
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,15,1.4213706652323406
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,15,0.31990933418273926
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,15,0.3203146656354268
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,15,0.3596693277359009
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,15,1.3595412572224934
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,15,0.35020267963409424
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,15,0.45047465960184735
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,15,0.35472532113393146
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,15,1.5155733426411946
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,15,0.3434026638666789
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,15,0.31623466809590656
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,15,0.34417064984639484
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,15,0.36017600695292157
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,15,0.34911465644836426
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,15,1.4030133883158367
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,15,0.3521813154220581
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,15,0.3537866671880086
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,15,0.3147146701812744
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,15,0.34033600489298504
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,15,0.35228268305460614
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,15,0.34882664680480957
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,15,0.3288639982541402
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,15,0.35925865173339844
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,15,0.6231520175933838
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,15,0.3118613362312317
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,15,0.34971733887990314
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,15,0.33985066413879395
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,15,0.3456053336461385
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,15,0.35073065757751465
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,15,0.6089280049006144
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,15,0.34411199887593585
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,15,0.3625386555989583
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,15,1.5813493728637695
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,15,0.34749333063761395
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,15,0.31003199021021527
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,15,0.3480159838994344
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,15,0.3461493253707886
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,15,0.6057226657867432
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,15,1.564458688100179
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,15,3.3621066411336265
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,15,0.5507786671320597
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,15,0.347109317779541
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,15,0.31575467189153034
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,15,0.34574933846791583
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,15,0.35204267501831055
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,15,0.3500373363494873
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,15,0.6148159901301066
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,15,1.573962688446045
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,31,0.30846933523813885
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,15,3.340677261352539
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,31,0.34146666526794434
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,15,0.35304534435272217
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,31,0.3428639968236287
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,31,0.3402506510416667
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,31,0.34229334195454914
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,15,6.72758420308431
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,31,0.3829600016276042
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,31,0.34095998605092365
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,31,0.34441065788269043
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,31,0.3517920176188151
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,31,0.36670398712158203
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,31,0.3404639959335327
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,31,0.3106879989306132
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,31,0.3539573351542155
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,31,0.345792015393575
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,31,0.3493013381958008
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,31,1.5400746663411458
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,31,0.33853332201639813
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,31,0.3505760033925374
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,31,0.3096800049146016
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,31,0.8674560387929281
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,31,1.4000266393025715
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,31,0.3415840069452922
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,31,0.34794668356577557
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,31,0.3502613306045532
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,31,0.3405760129292806
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,31,0.41360000769297284
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,31,0.3190666635831197
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,31,0.3442933162053426
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,31,0.3418773412704468
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,31,0.34727998574574787
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,31,0.3766186634699504
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,31,0.373855988184611
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,31,0.5015573501586914
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,31,0.34856534004211426
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,31,0.3421866496404012
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,31,0.3482133150100708
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,31,0.3470826546351115
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,31,0.3508266607920329
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,31,0.3603359858194987
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,31,0.5863519906997681
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,31,0.6132586797078451
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,31,0.34992531935373944
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,31,0.3468266725540161
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,31,0.3543039957682292
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,31,0.3435893456141154
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,31,0.31782933076222736
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,31,0.34277331829071045
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,31,0.34859732786814374
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,31,0.3722613255182902
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,31,0.35283732414245605
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,31,0.3471519947052002
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,31,0.31969600915908813
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,31,0.3465119997660319
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,31,0.31215999523798627
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,31,0.38763733704884845
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,31,0.31016000111897785
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,31,0.33511467774709064
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,31,0.6188960075378418
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,31,0.345578670501709
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,31,0.35304534435272217
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,31,0.34621334075927734
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,31,0.3402986526489258
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,31,0.33959468205769855
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,31,0.3235359986623128
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,31,0.6119466622670492
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,31,0.3463093439737956
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,31,0.3484426736831665
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,31,0.4419573148091634
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,31,1.5910986264546711
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,31,0.33845333258310956
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,31,0.33196266492207843
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,31,0.33402132987976074
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,31,1.580901304880778
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,31,0.34435733159383136
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,31,0.6240053176879883
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,31,0.321397324403127
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,31,0.3059786756833394
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,31,3.3731892903645835
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,31,0.34080533186594647
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,31,0.31111466884613037
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,31,0.35717864831288654
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,31,1.6359306971232097
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,31,1.401045322418213
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,31,0.34221867720286053
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,31,3.3812745412190757
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,31,6.769642512003581
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,63,0.34350399176279706
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,31,0.34998401006062824
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,63,0.3437066475550334
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,31,0.3531839847564697
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,63,0.34623467922210693
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,63,0.37145066261291504
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,63,0.33349335193634033
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,63,0.34618667761484784
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,63,0.35100801785786945
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,63,0.7494666576385498
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,63,0.3475946585337321
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,63,0.3197493354479472
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,63,0.31757332881291706
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,63,0.344490647315979
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,63,0.3404639959335327
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,63,0.3489813407262166
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,63,0.34514665603637695
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,63,1.2803626855214436
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,63,0.3748106559117635
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,63,0.3736746708552043
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,63,0.3125493327776591
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,63,0.3477706511815389
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,63,0.3566186825434367
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,63,0.350053350130717
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,63,0.5323626597722372
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,63,0.35149868329366046
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,63,0.3415840069452922
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,63,0.34910400708516437
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,63,0.3667840162913005
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,63,0.35153599580128986
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,63,1.4550293286641438
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,63,0.32155199845631915
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,63,0.6890880266825358
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,63,0.3488853375116984
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,63,0.3500746488571167
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,63,0.3447200059890747
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,63,1.3964160283406575
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,63,1.4341386159261067
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,63,0.3419253428777059
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,63,0.35652267932891846
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,63,0.34967466195424396
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,63,0.3487253189086914
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,63,0.34700798988342285
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,63,0.34562134742736816
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,63,0.3468480110168457
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,63,0.34434668223063153
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,63,0.3497120141983032
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,63,0.3500106732050578
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,63,0.4582506815592448
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,63,0.30692799886067706
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,63,0.35149868329366046
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,63,1.395151933034261
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,63,0.3542240063349406
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,63,0.33104532957077026
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,63,0.3241386612256368
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,63,0.3380906581878662
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,63,0.31224532922108966
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,63,0.3461546500523885
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,63,0.35231467088063556
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,63,0.36374398072560626
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,63,0.4469013214111328
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,63,0.3552159865697225
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,63,0.6346666812896729
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,63,0.3500959873199463
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,63,0.5224159955978394
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,63,0.31967467069625854
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,63,0.369429349899292
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,63,0.6362346808115641
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,63,0.32291199763615924
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,63,1.6138399442036946
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,63,0.32949866851170856
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,63,0.34513068199157715
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,63,0.33504533767700195
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,63,0.30903466542561847
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,63,0.37883198261260986
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,63,0.3325813412666321
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,63,0.6649440129597982
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,63,1.636122703552246
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,63,3.406773249308268
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,63,0.31249600648880005
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,63,0.34055999914805096
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,63,0.3275253375371297
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,63,0.7300639947255453
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,63,1.7477439244588215
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,63,0.3507680098215739
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,63,0.36033066113789874
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,63,3.5211893717447915
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,63,0.34457067648569745
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,63,6.992085138956706
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,63,0.32416532437006634
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,127,0.4930186669031779
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,127,0.7125813166300455
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,127,0.5043413241704305
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,127,0.7865066528320312
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,127,0.49872533480326336
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,127,0.7804959615071615
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,127,1.9295040766398113
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,127,0.3460479974746704
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,127,3.1467625300089517
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,127,0.346346656481425
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,127,0.3483733336130778
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,127,0.34353601932525635
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,127,0.34699734052022296
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,127,0.37057065963745117
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,127,0.3277386625607808
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,127,0.3241386612256368
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,127,0.4540799856185913
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,127,0.31541333595911664
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,127,0.33979201316833496
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,127,0.34093332290649414
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,127,0.341045339902242
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,127,0.35388799508412677
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,127,0.8046773274739584
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,127,0.3295466701189677
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,127,0.31301865975062054
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,127,0.34020264943440753
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,127,0.342303991317749
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,127,0.34508800506591797
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,127,0.34511999289194745
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,127,0.35239466031392414
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,127,0.31065066655476886
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,127,0.35126932462056476
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,127,0.34098132451375324
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,127,0.34613335132598877
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,127,0.34959999720255536
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,127,0.30776000022888184
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,127,0.31303467353185016
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,127,0.339194655418396
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,127,0.343392014503479
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,127,0.340554674466451
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,127,0.34013867378234863
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,127,0.3348533312479655
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,127,0.3104853431383769
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,127,0.3219573299090068
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,127,0.3343840042750041
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,127,0.33212266365687054
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,127,0.3331306576728821
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,127,0.336298664410909
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,127,0.337007999420166
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,127,0.6516746679941813
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,127,0.31197333335876465
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,127,0.3333333333333333
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,127,0.33209067583084106
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,127,0.3361599842707316
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,127,0.3353866736094157
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,127,0.33035733302434284
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,127,0.6695626576741537
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,127,1.670789400736491
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,127,0.6775786876678467
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,127,0.3174560070037842
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,127,0.29954665899276733
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,127,0.30077866713205975
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,127,0.29602666695912677
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,127,0.30551467339197796
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,127,0.3023680051167806
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,127,0.29681599140167236
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,127,0.29736000299453735
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,127,3.5042025248209634
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,127,0.30128000179926556
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,127,1.778783957163493
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,127,0.7571787039438883
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,127,0.3619733254114787
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,127,0.2998453378677368
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,127,0.343392014503479
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,127,0.33397332827250165
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,127,0.3301653265953064
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,127,0.33829331398010254
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,127,0.3290133277575175
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,127,0.30169065793355304
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,127,0.8809866905212402
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,127,1.9736053148905437
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,127,0.5086186726888021
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,127,3.918229420979818
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,127,2.0208586057027182
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,255,0.3555999994277954
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,255,0.3305013378461202
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,255,0.3129013379414876
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,127,7.7500050862630205
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,255,0.33482666810353595
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,255,1.663258711496989
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,255,3.064154624938965
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,255,0.40774933497111004
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,255,0.35049064954121906
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,255,0.3114560047785441
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,255,0.3499893347422282
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,255,0.33657066027323407
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,255,0.3389493227005005
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,255,0.34382398923238117
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,255,0.4477866490681966
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,255,0.3001226584116618
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,255,0.35468268394470215
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,255,0.3455573320388794
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,127,0.3786986668904622
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,255,0.35581334431966144
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,255,0.3514133294423421
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,255,0.3489546775817871
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,255,0.7588106791178385
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,255,0.3450560172398885
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,255,0.35313598314921063
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,255,0.5202560027440389
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,255,0.34537601470947266
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,255,0.3455520073572795
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,255,0.3091413378715515
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,255,0.3370240132013957
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,127,0.3947306474049886
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,255,0.33637332916259766
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,255,0.3508426745732625
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,255,2.4182507197062173
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,255,0.3428639968236287
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,255,0.3089546759923299
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,255,0.3062933286031087
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,255,0.34723734855651855
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,255,0.3405439853668213
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,255,0.3456000089645386
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,255,0.3384373188018799
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,255,0.3100159962972005
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,255,0.3138879934946696
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,127,0.3139359951019287
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,255,1.250864028930664
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,255,0.3115466634432475
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,255,0.34378667672475177
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,255,0.33958399295806885
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,255,0.35631998380025226
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,255,0.35396798451741535
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,255,0.3507466713587443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,255,0.3115946650505066
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,255,0.35204799969991046
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,255,0.34364267190297443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,255,0.31305599212646484
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,255,0.3481493393580119
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,255,0.333621342976888
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,255,0.3255839943885803
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,255,0.3640906810760498
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,255,0.3075679937998454
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,255,0.34541865189870197
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,255,0.35042667388916016
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,255,0.3714880148569743
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,255,0.3471839825312297
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,255,0.32823999722798664
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,255,0.6871946652730306
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,255,0.32290132840474445
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,255,0.3376053174336751
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,255,1.3243146737416585
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,255,0.3465706507364909
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,255,0.5197546482086182
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,255,0.4031840165456136
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,255,0.7656213442484537
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,255,1.7467786471048992
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,255,0.3478826681772868
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,255,0.3084426720937093
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,255,0.5104480187098185
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,255,0.9883039792378744
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,255,0.3669653336207072
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,255,2.000784079233805
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,255,0.36640532811482746
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,255,0.3548479874928792
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,255,3.89140256245931
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,255,0.3572479883829753
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,255,4.78224531809489
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,255,1.1399199962615967
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,255,0.7457173665364584
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,255,0.5632693370183309
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,255,0.6096106767654419
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,255,2.4056480725606284
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,511,0.9832800229390463
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,511,0.3096959988276164
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,511,0.32422399520874023
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,511,0.32064000765482586
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,511,0.3076053261756897
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,511,0.34725332260131836
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,255,9.55959447224935
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,511,0.3400053183237712
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,511,0.3450666666030884
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,511,0.35628267129262287
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,511,0.3511413335800171
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,511,0.9122347036997477
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,511,0.35153067111968994
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,511,0.341103990872701
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,511,0.35734931627909344
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,511,0.3131093382835388
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,511,0.34410667419433594
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,511,0.410863995552063
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,511,0.349509318669637
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,511,0.35977598031361896
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,511,0.34722665945688885
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,511,0.31884799400965375
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,511,0.35282667477925617
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,511,0.3489546775817871
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,511,0.3444799979527791
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,511,0.35626665751139325
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,511,0.3408426841100057
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,511,0.34904531637827557
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,511,0.9891040325164795
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,511,0.34750401973724365
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,511,0.35492801666259766
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,511,0.3567519982655843
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,511,0.30854399998982746
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,511,0.34491201241811115
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,511,0.3728959957758586
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,511,0.34386134147644043
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,511,0.34490664800008136
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,511,0.3184373378753662
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,511,0.31748799482981366
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,511,0.34348801771799725
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,511,0.34725332260131836
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,511,0.31940799951553345
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,511,0.34295467535654706
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,511,0.349018653233846
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,511,0.3403253157933553
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,255,0.5374186833699545
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,511,0.3476373354593913
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,511,0.4774880011876424
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,511,0.35732801755269367
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,511,0.3437653382619222
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,511,0.37989334265391034
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,511,0.38020801544189453
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,511,0.3213599920272827
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,511,0.3474773168563843
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,511,0.3441760142644246
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,511,0.34461331367492676
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,511,0.36821333567301434
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,511,0.3447786569595337
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,511,0.8536586761474609
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,511,0.3513173262278239
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,511,0.34540800253550213
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,511,0.34090133508046466
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,511,0.4593120018641154
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,511,0.3490826686223348
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,511,0.3174720009167989
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,511,0.33827733993530273
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,511,0.3715360164642334
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,511,0.43933331966400146
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,511,0.9915680090586344
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,511,0.5490773518880209
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,511,0.33355732758839923
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,511,1.9370293617248535
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,511,0.3497653404871623
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,511,0.3612693150838216
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,511,0.7408373355865479
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,511,2.7379093170166016
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,511,0.6022400061289469
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,511,1.4235572814941406
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,511,0.5410133202870687
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,511,0.5053386688232422
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,511,5.380912144978841
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,511,0.5496160189310709
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,511,3.1537278493245444
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,511,1.1937386989593506
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,511,1.0469813346862793
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,511,1.5552372932434082
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,511,6.295541127522786
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,511,0.9677226543426514
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,511,12.6551513671875
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,511,1.01146666208903
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,1023,0.2985919912656148
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,1023,0.3130826751391093
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,1023,0.2994133234024048
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,1023,0.3144746621449788
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,1023,0.30634133021036786
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,1023,0.3171786665916443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,1023,0.3349333206812541
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,1023,0.3296639919281006
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,1023,0.33692800998687744
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,1023,0.3259893258412679
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,1023,0.3454506794611613
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,1023,0.3137493332227071
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,1023,0.3147040009498596
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,1023,0.3341546853383382
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,1023,0.34090133508046466
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,1023,0.3391679922739665
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,1023,0.343120018641154
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,1023,0.3350293238957723
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,1023,0.2991466720898946
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,1023,0.3130506674448649
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,1023,0.3299306631088257
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,1023,0.33270400762557983
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,1023,0.33299734195073444
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,1023,0.3327999909718831
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,1023,0.32940266529719037
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,1023,0.31033066908518475
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,1023,0.31758934259414673
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,1023,0.30849599838256836
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,1023,0.326090673605601
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,1023,0.32984532912572223
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,1023,0.3301653265953064
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,1023,0.3309920032819112
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,1023,0.3095466693242391
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,1023,0.3693866729736328
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,1023,0.3195786674817403
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,1023,0.3100159962972005
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,1023,0.33826665083567303
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,1023,0.346010684967041
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,1023,0.3296746611595154
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,1023,0.3360053300857544
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,1023,0.6111199855804443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,1023,0.31033599376678467
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,1023,0.34276266892751056
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,1023,0.5715680122375488
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,1023,0.3311466574668884
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,1023,0.3323413332303365
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,1023,0.30405332644780475
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,1023,0.336847980817159
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,1023,0.3240320086479187
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,1023,1.2482826709747314
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,1023,0.6724906762441
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,1023,0.40137068430582684
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,1023,0.326255997021993
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,1023,0.30530667304992676
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,1023,0.3200586636861165
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,1023,0.34620265165964764
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,1023,0.34563199679056805
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,1023,0.3561600049336751
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,1023,2.9358558654785156
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,1023,1.5810933113098145
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,1023,0.9188053607940674
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,1023,0.6031146844228109
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,1023,0.5410879850387573
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,1023,0.5054399967193604
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,1023,0.4924693504969279
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,1023,0.5423786640167236
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,1023,9.511130650838217
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,1023,4.729226748148601
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,1023,0.30543466409047443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,1023,0.3086346586545308
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,1023,2.367664019266764
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,1023,1.1898612976074219
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,1023,1.0437706311543782
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,1023,0.965002695719401
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,1023,0.30001600583394367
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,1023,0.9254506429036459
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,1023,1.0058720111846924
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,1023,4.734266599019368
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,1023,2.0426294008890786
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,1023,19.01422373453776
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,1023,2.3727572758992515
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,1023,0.3006666700045268
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,1023,9.458330790201822
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,2047,0.33898667494455975
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,2047,0.3632213274637858
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,2047,0.33697064717610675
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,2047,0.46458665529886883
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,2047,0.34433599313100177
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,2047,0.33079467217127484
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,2047,0.3396426836649577
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,2047,0.34695998827616376
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,2047,0.3439306815465291
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,2047,0.33645331859588623
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,2047,0.3476320107777913
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,2047,0.8378400007883707
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,2047,0.3412533203760783
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,2047,0.35123201211293537
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,2047,0.35095465183258057
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,2047,0.34118398030598956
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,2047,0.3452213207880656
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,2047,0.34005868434906006
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,2047,0.3195893367131551
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,2047,0.31016000111897785
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,2047,0.3755626678466797
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,2047,0.3508586486180623
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,2047,0.34594134489695233
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,2047,0.3062826593716939
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,2047,0.39928531646728516
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,2047,0.3118293285369873
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,1023,0.32577600081761676
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,1023,2.947392145792643
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,2047,0.3457653522491455
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,2047,0.3569653431574504
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,2047,0.3458293279012044
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,1023,1.8944212595621746
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,1023,1.7967146237691243
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,2047,0.3485066493352254
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,2047,0.33583998680114746
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,2047,0.34185067812601727
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,2047,0.37015998363494873
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,2047,0.35488001505533856
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,2047,0.30475733677546185
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,2047,0.34559468428293866
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,2047,0.3458933432896932
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,2047,0.35411731402079266
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,2047,0.3542986710866292
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,2047,0.36392001310984295
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,2047,0.36799466609954834
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,2047,0.48343467712402344
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,2047,0.3567626476287842
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,2047,0.3484426736831665
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,2047,0.3556053241093953
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,2047,0.35731200377146405
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,2047,0.3399733304977417
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,2047,0.3516480127970378
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,2047,0.3722826639811198
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,2047,0.5221386750539144
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,2047,0.35860268274943036
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,2047,0.36289600531260174
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,2047,0.37098666032155353
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,2047,0.9260479609171549
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,1023,0.3167733351389567
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,2047,0.3628053267796834
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,2047,0.3511626720428467
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,2047,0.5054719845453898
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,2047,0.5377973318099976
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,2047,0.7287306785583496
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,2047,2.0636107126871743
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,2047,1.188538630803426
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,2047,0.4772160053253174
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,2047,0.48578667640686035
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,2047,0.5407146612803141
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,2047,0.9777759710947672
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,2047,3.9408321380615234
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,2047,1.0584426720937092
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,2047,1.996986707051595
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,2047,0.9370453357696533
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,2047,0.913487990697225
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,2047,7.997882843017578
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,2047,1.0191946824391682
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,2047,2.042261282602946
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,2047,1.8809013366699219
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,2047,4.047941207885742
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,2047,8.093072255452475
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,2047,1.9022186597188313
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,2047,1.8000532786051433
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,2047,16.19706090291341
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,2047,1.9100319544474285
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,2047,4.045498530069987
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,2047,8.068623860677084
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,2047,3.697781244913737
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,2047,3.523674647013346
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,2047,16.16326395670573
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,2047,3.4209601084391275
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,4095,0.3502720197041829
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,4095,0.3431946833928426
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,4095,0.35875733693440753
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,4095,0.34239999453226727
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,4095,0.8782026767730713
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,4095,0.3707626660664876
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,4095,0.34029332796732586
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,4095,0.3483039935429891
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,4095,0.36374398072560626
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,4095,0.35653332869211835
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,4095,0.5604639848073324
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,4095,0.3452639977137248
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,4095,0.35014931360880536
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,4095,0.3530293305714925
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,4095,0.3535999854405721
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,4095,0.34599467118581134
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,4095,0.3520853519439697
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,4095,0.3283360004425049
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,4095,0.3518986701965332
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,2047,32.39128112792969
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,4095,0.3680853446324666
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,4095,0.3514293432235718
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,4095,0.3522026538848877
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,4095,0.3663626511891683
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,4095,0.36425065994262695
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,4095,1.1468799908955891
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,4095,0.3487039804458618
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,4095,0.3467573324839274
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,4095,0.5297173261642456
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,4095,0.3606719970703125
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,4095,0.3381439844767253
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,4095,0.33406933148701984
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,4095,0.37166933218638104
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,4095,0.35879464944203693
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,4095,0.495743989944458
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,4095,0.3297119935353597
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,4095,0.894981304804484
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,4095,0.47704533735911053
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,4095,0.35441601276397705
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,4095,0.34384532769521076
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,4095,0.47382934888203937
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,4095,0.4102826515833537
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,4095,0.33212266365687054
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,4095,0.8531839847564697
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,4095,0.3636266787846883
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,4095,1.0233439604441326
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,4095,1.4061546325683594
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,4095,0.4076426823933919
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,4095,0.3651039997736613
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,2047,3.6308587392171225
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,4095,1.069439967473348
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,4095,1.753434658050537
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,4095,0.6409759918848673
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,4095,0.5135840177536011
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,4095,0.49000000953674316
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,4095,0.47645334402720135
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,4095,0.534005324045817
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,4095,0.4734453360239665
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,4095,0.9608373641967773
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,4095,1.797770659128825
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,4095,0.9176479975382487
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,4095,0.8968160152435303
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,4095,0.8873333136240641
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,4095,3.424895922342936
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,4095,6.759082794189453
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,4095,0.9858293533325195
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,4095,1.9061066309611003
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,4095,3.70356814066569
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,4095,1.8121760686238606
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,4095,1.7595200538635254
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,4095,1.7358345985412598
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,4095,7.3318131764729815
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,4095,1.8880693117777507
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,4095,14.738335927327475
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,4095,29.36523691813151
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,4095,3.715733210245768
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,4095,3.518853187561035
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,4095,7.335952123006185
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,4095,3.7147251764933267
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,4095,14.661691029866537
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,4095,3.4192374547322593
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,4095,3.362911860148112
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,4095,7.326437632242839
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,4095,6.954458872477214
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,4095,14.643781026204428
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,4095,6.626298904418945
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,4095,6.750133514404297
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,8191,0.35155733426411945
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,8191,0.3511413335800171
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,8191,0.3339466651280721
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,4095,29.321428934733074
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,8191,0.35728001594543457
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,8191,0.3483946720759074
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,8191,0.568399985631307
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,8191,0.34414398670196533
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,8191,0.4872959852218628
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,8191,0.3568586508433024
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,8191,0.35970131556193036
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,8191,1.4452373186747234
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,8191,0.3736906846364339
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,8191,0.3285599946975708
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,8191,0.3755253156026204
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,8191,0.35493866602579754
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,8191,0.36504534880320233
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,8191,0.4864480098088582
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,8191,0.3385759989420573
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,8191,0.35968534151713055
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,8191,0.3627359867095947
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,8191,0.35919467608133954
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,8191,0.3487733205159505
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,8191,0.3505280017852783
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,4095,58.79247538248698
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,8191,0.9300586382548014
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,8191,0.36555198828379315
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,8191,0.35280001163482666
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,8191,0.33695467313130695
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,8191,0.46078399817148846
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,8191,0.39792533715566
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,8191,0.36512001355489093
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,8191,0.3808480103810628
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,8191,0.4883413314819336
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,8191,0.4580106735229492
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,8191,0.829749345779419
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,8191,1.3117706775665283
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,8191,0.3792959849039714
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,8191,2.1648693084716797
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,8191,0.3709919850031535
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,8191,0.7125493685404459
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,8191,0.3397173484166463
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,8191,1.607850710550944
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,8191,0.5554986794789633
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,4095,7.112927754720052
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,8191,0.8691893418629965
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,8191,0.47304534912109375
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,8191,0.47020800908406574
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,8191,0.4773706595102946
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,8191,0.481663982073466
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,8191,0.7417973677317301
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,8191,1.235749324162801
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,8191,2.256864070892334
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,8191,0.9158666928609213
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,8191,0.8936533133188883
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,8191,0.8847200075785319
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,8191,3.776010513305664
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,8191,0.875877300898234
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,8191,0.9737226963043213
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,8191,1.7264960606892903
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,8191,3.4247093200683594
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,8191,1.794314702351888
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,8191,6.649274826049805
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,8191,1.7013120651245117
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,8191,2.467754681905111
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,8191,1.8630986213684082
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,8191,13.280442555745443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,8191,27.61908213297526
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,8191,13.77688471476237
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,8191,6.959893544514974
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,8191,3.5742241541544595
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,8191,3.553333282470703
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,8191,3.4084959030151367
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,8191,3.375125249226888
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,8191,3.664053281148275
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,8191,13.825088500976562
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,8191,6.744864145914714
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,8191,6.947120030721028
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,8191,27.70110829671224
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,8191,6.639343897501628
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,8191,7.312261581420898
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,8191,6.580485026041667
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,8191,55.33973185221354
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1024,1,4,8191,27.519808451334637
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1024,1,8,8191,13.734725952148438
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1024,1,16,8191,13.343498229980469
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,16383,0.3768693208694458
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,16383,0.36987733840942383
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,16383,0.402890682220459
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,16383,0.3861386775970459
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,16383,0.3299520015716553
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,16383,0.3733760118484497
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,16383,0.3805600007375081
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,16383,0.9125653107961019
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,16383,0.36046401659647626
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,16383,0.3768586715062459
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1024,1,64,8191,12.995674133300781
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,16383,0.38359999656677246
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,16383,0.35736000537872314
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,16383,0.37674665451049805
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,16383,0.3694399992624919
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,16383,0.3757813374201457
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1024,1,128,8191,14.011887868245443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1024,1,2,8191,55.170817057291664
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,16383,0.9174239635467529
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,16383,0.4031786521275838
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,16383,0.4596746762593587
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1024,1,32,8191,13.12453842163086
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,16383,1.120848019917806
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,16383,0.3775999943415324
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,16383,0.3860479990641276
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,16383,0.375983993212382
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,16383,0.36681067943573
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,16383,0.9769813219706217
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,16383,0.4156959851582845
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,16383,0.38815466562906903
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,16383,0.4057919979095459
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,16383,0.4549013376235962
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,16383,0.8148852984110514
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,16383,0.3808053334554036
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,16383,0.3791840076446533
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,16383,1.3364213307698567
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,16383,0.8426132996877035
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,16383,0.48766934871673584
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,16383,1.5755839347839355
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,16383,0.4761973222096761
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,16383,0.4706133206685384
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,16383,0.4660160144170125
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,16383,0.4658240079879761
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,16383,1.3535946210225422
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,16383,0.8896000385284424
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,16383,0.8796532948811849
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,16383,1.6175999641418457
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,16383,0.9961600303649902
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,16383,0.8769013086954752
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,16383,3.1455307006835938
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,16383,0.874229351679484
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,16383,1.4057920773824055
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,16383,1.7263092994689941
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,16383,1.7041120529174805
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,16383,2.3526132901509604
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,16383,1.691109339396159
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,16383,4.433157285054524
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,16383,1.6868693033854167
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,16383,7.2127736409505205
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,16383,1.8570079803466797
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1024,1,1,8191,110.75770060221355
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,16383,6.73908805847168
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,16383,3.3719733556111655
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,16383,3.3258771896362305
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,16383,3.304976145426432
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,16383,13.219098409016928
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,16383,3.289408047993978
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,16383,3.616666793823242
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,16383,26.34900156656901
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,16383,6.862138748168945
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,256,1,16,16383,6.73252805074056
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,256,1,4,16383,13.412090301513672
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,256,1,32,16383,6.67678960164388
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,16383,6.651519775390625
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,256,1,2,16383,26.60692850748698
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,16383,7.202842712402344
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,16383,53.33959452311198
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,512,1,8,16383,13.383168538411459
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,512,1,64,16383,12.977242787679037
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,512,1,16,16383,13.303333282470703
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,512,1,4,16383,26.545305887858074
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,512,1,32,16383,13.043999989827475
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,512,1,128,16383,14.453765869140625
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,512,1,2,16383,53.16519673665365
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,32767,0.6733973026275635
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,32767,0.6547786792119344
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,32767,0.6499573389689127
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,32767,0.6520746548970541
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,512,1,1,16383,106.56178792317708
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,32767,0.5872799952824911
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,32767,0.5769866704940796
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,32767,0.5785546700159708
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,32767,1.7725599606831868
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,32767,0.7357707023620605
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,32767,0.6617546478907267
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,32767,0.6578880151112875
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,32767,0.659717321395874
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,32767,0.584330677986145
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,32767,0.5810346603393555
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,32767,0.5811839898427328
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,32767,1.7761386235555012
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,32767,0.7957226435343424
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,32767,0.729306697845459
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,32767,0.7165813446044922
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,32767,0.7157333691914877
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,32767,0.6343946854273478
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,32767,0.6258773406346639
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,32767,0.6299733320871989
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,32767,2.5965280532836914
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,32767,1.5137759844462078
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,32767,0.818015972773234
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,32767,0.7198560237884521
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,32767,0.7163946628570557
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,32767,0.6363893349965414
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,32767,0.6311999956766764
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,32767,0.6350826819737753
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,32767,2.606640021006266
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,32767,3.009690602620443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,32767,1.576250712076823
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,32767,0.8827679951985677
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,32767,0.8671627044677734
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,32767,0.8664533297220866
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,32767,0.8607146739959717
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,32767,0.8609653313954672
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,32767,2.641002655029297
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,32767,1.69870392481486
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,32767,1.8847840627034504
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,32767,3.081829388936361
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,32767,6.055482864379883
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,32767,1.6879839897155762
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,32767,1.6817439397176106
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,32767,1.6793546676635742
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,32767,2.740816116333008
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,64,1,1,32767,15.036052703857422
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,32767,3.3592907587687173
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,32767,3.311392148335775
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,32767,4.886405309041341
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,32767,8.965487798055014
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,32767,3.338245391845703
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,32767,3.3211841583251953
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,32767,3.625610669453939
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,32767,13.328960418701172
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,32767,6.5245920817057295
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,32767,6.5644105275472
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,32767,7.12278938293457
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,32767,6.49946657816569
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,32767,6.489877065022786
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,128,1,2,32767,26.11712646484375
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,256,1,8,32767,13.429264068603516
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,128,1,1,32767,52.010294596354164
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,256,1,128,32767,14.262016296386719
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,256,1,64,32767,13.17242685953776
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,256,1,1,32767,104.4238789876302
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,65535,1.2782666683197021
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,65535,1.278218666712443
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,65535,1.2486240069071453
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,65535,1.1117226282755535
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,65535,1.2549813588460286
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,65535,1.099503993988037
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,65535,1.1061066786448162
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,65535,1.3811945915222168
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,65535,1.3636639912923176
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,65535,3.4956159591674805
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,65535,1.3612640698750813
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,65535,1.361189365386963
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,65535,1.1968533198038738
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,65535,1.1928053696950276
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,65535,1.1983199914296467
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,65535,1.4680639902750652
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,65535,1.3680373827616374
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,65535,5.127813339233398
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,65535,1.3586400349934895
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,65535,1.3641600608825684
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,65535,1.201632022857666
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,65535,1.1959413687388103
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,65535,1.1971413294474285
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,65535,1.5471733411153157
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,65535,5.14575990041097
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,65535,2.8949333826700845
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,65535,1.374608039855957
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,65535,1.3759946823120117
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,65535,1.2128960291544597
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,65535,1.2062880198160808
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,65535,1.2061599890391033
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,65535,5.155690511067708
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,65535,1.687562624613444
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,65535,3.046698570251465
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,65535,5.87336540222168
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,65535,1.6605173746744792
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,65535,1.6571893692016602
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,65535,1.652016003926595
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,65535,1.6529332796732585
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,65535,5.211034774780273
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,65535,6.0392106374104815
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,65535,3.6743946075439453
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,65535,11.943375905354818
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,65535,3.2987359364827475
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,65535,3.3131306966145835
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,65535,3.2906452814737954
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,65535,3.2828585306803384
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,65535,5.404975891113281
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,64,1,8,65535,6.601861317952474
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,64,1,4,65535,9.64791488647461
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,64,1,16,65535,6.536511739095052
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,64,1,32,65535,6.534021377563477
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,64,1,64,65535,6.547648111979167
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,64,1,128,65535,7.133904139200847
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,64,1,2,65535,20.593072255452473
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,128,1,4,65535,26.251449584960938
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,128,1,8,65535,12.922879536946615
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,128,1,16,65535,12.894922892252604
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,128,1,32,65535,12.868794759114584
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,128,1,64,65535,12.8832155863444
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,128,1,128,65535,14.124640146891275
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,1,1,1,131071,2.653536001841227
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,1,1,2,131071,2.643184026082357
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,1,1,4,131071,2.642826716105143
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,1,1,8,131071,2.6546452840169272
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,1,1,16,131071,2.329045295715332
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,1,1,32,131071,2.321450710296631
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,1,1,64,131071,2.3083093961079917
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,1,1,128,131071,10.151472091674805
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,2,1,1,131071,2.656522591908773
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,2,1,2,131071,2.632533391316732
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,2,1,4,131071,2.63210662206014
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,2,1,8,131071,2.651066621144613
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,2,1,16,131071,2.3532427151997886
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,2,1,32,131071,2.3189013799031577
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,2,1,64,131071,2.317216078440348
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,2,1,128,131071,10.201599756876627
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,4,1,1,131071,2.855423927307129
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,4,1,2,131071,2.663637320200602
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,4,1,4,131071,2.6556480725606284
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,4,1,8,131071,2.6722081502278647
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,4,1,16,131071,2.3517120679219565
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,4,1,32,131071,2.3399359385172525
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,4,1,64,131071,2.338832060496012
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,4,1,128,131071,10.499173482259115
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,8,1,1,131071,5.7229868570963545
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,8,1,2,131071,3.0330241521199546
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,8,1,4,131071,2.7054932912190757
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,8,1,8,131071,2.692954699198405
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,8,1,16,131071,2.3581226666768393
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,8,1,32,131071,2.345989386240641
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,8,1,64,131071,2.345194657643636
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,8,1,128,131071,10.25588289896647
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,16,1,1,131071,11.690308888753256
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,16,1,2,131071,6.010016123453776
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,16,1,4,131071,3.332570711771647
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,16,1,8,131071,3.233925183614095
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,16,1,16,131071,3.2379465103149414
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,16,1,32,131071,3.233557383219401
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,16,1,64,131071,3.231477419535319
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,16,1,128,131071,10.371557235717773
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,128,32,1,1,131071,31.04889678955078
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,64,32,1,2,131071,12.321253458658854
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,32,32,1,4,131071,8.042847951253256
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,16,32,1,8,131071,6.502634684244792
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,8,32,1,16,131071,6.489072163899739
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,4,32,1,32,131071,6.472543716430664
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,2,32,1,64,131071,6.472960154215495
VLLM,0.14.0,NVIDIA L40S,generation_mla,vllm_triton_mla,float16,float16,1,32,1,128,131071,10.73083241780599
