framework,version,device,op_name,kernel_source,batch_size,isl,num_heads,num_key_value_heads,head_dim,beam_width,attn_dtype,kv_cache_dtype,step,latency
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,1,0.011055999745925268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,31,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,31,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,63,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,63,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,127,0.009754666437705358
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,127,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,255,0.009706666693091393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,255,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,511,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,511,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,1023,0.012890666723251343
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,1023,0.012906666845083237
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,2047,0.029648000995318096
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,1,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,2047,0.016970666746298473
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,4095,0.03149333347876867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,4095,0.04752533137798309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,1,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,1,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,3,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,3,0.010778666784365972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,7,0.010768000036478043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,7,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,3,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,15,0.010837333897749582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,15,0.01101333275437355
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,7,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,15,0.010869332899649939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,float16,15,0.010874666273593903
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,31,0.011194666226704916
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,31,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,63,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,63,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,4,128,1,float16,fp8,7,0.010901333143313726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,127,0.011120000233252844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,255,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,255,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,511,0.012821332861979803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,1023,0.029333333174387615
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,511,0.01313599944114685
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,1023,0.015024000157912573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,2047,0.04775999983151754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,2047,0.031498665610949196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,1,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,3,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,4095,0.04993600149949392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,3,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,float16,4095,0.08457066615422566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,7,0.009290666629870733
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,7,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,15,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,15,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,31,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,31,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,63,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,127,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,3,0.010901333143313726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,255,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,255,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,511,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,511,0.010661333799362183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,1023,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,1023,0.01071999967098236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,2047,0.010842667271693548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,2047,0.010640000303586325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,4095,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,4095,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,8191,0.015119999647140503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,8191,0.015040000279744467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,16383,0.01728533332546552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,16383,0.017184000462293625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,1,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,1,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,3,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,3,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,7,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,7,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,15,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,15,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,31,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,63,0.008661333471536636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,64,8,128,1,float16,fp8,127,0.011183999478816986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,127,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,255,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,255,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,float16,1,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,511,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,1023,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,1023,0.011125333607196808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,63,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,4,128,1,float16,fp8,127,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,2047,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,2047,0.009706666693091393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,4095,0.012928000340859095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,8191,0.014885333677132925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,8191,0.01470400020480156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,16383,0.01687466725707054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,16383,0.015205333630243937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,1,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,3,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,3,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,7,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,7,0.009429333110650381
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,31,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,63,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,15,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,15,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,127,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,31,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,31,0.009306666751702627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,63,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,63,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,127,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,255,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,255,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,511,0.010677333921194077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,511,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,1023,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,1023,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,2047,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,4095,0.01320533330241839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,4095,0.01301866645614306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,8191,0.01509333277742068
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,fp8,4095,0.012800000607967377
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,16383,0.017269333203633625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,float16,1,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,1,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,3,0.008639999975760778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,3,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,7,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,7,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,15,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,31,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,127,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,31,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,64,8,128,1,float16,float16,511,0.00921066664159298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,63,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,63,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,2047,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,127,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,255,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,127,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,255,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,511,0.009408000235756239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,8191,0.015135999768972397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,511,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,1023,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,1023,0.009984000275532404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,4,128,1,float16,fp8,16383,0.01695466662446658
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,2047,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,2047,0.01110400011142095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,8191,0.016586666305859882
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,8191,0.014869333555301031
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,4095,0.013162666310866674
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,1,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,float16,16383,0.03125333289305369
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,16383,0.017184000462293625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,1,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,1,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,3,0.01098666712641716
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,3,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,7,0.012495999534924826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,15,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,31,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,15,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,31,0.011488000551859537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,63,0.012538666526476542
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,63,0.012117333710193634
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,127,0.011429333438475927
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,127,0.011440000186363855
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,255,0.012709333250919977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,255,0.012096000214417776
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,511,0.014917333920796713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,511,0.0143306665122509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,1023,0.03046933313210805
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,1023,0.01720533271630605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,float16,2047,0.04924799998601278
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,2047,0.03181333343187968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,1,0.013445333888133367
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,3,0.012896000097195307
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,7,0.01301866645614306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,3,0.01452800010641416
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,7,0.012805332740147909
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,64,8,128,1,float16,fp8,4095,0.01309866706530253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,15,0.014576000471909841
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,15,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,31,0.013914667069911957
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,7,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,63,0.012885333349307379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,63,0.014085333794355392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,4,128,1,float16,fp8,15,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,127,0.014864000181357065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,127,0.012874666601419449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,255,0.012837332983811697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,255,0.013898666948080063
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,1023,0.04923733572165171
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,511,0.016789333273967106
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,511,0.029680001238981884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,1023,0.03141866624355316
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,1,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,3,0.00973866693675518
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,7,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,float16,2047,0.0857973297437032
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,1,0.009642666826645533
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,3,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,7,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,15,0.009653333574533463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,31,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,63,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,63,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,1,0.012789333860079447
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,127,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,127,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,255,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,255,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,511,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,511,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,1023,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,1023,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,2047,0.012944000462690989
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,4095,0.014837333311637243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,4095,0.014906667172908783
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,31,0.013167999684810638
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,8191,0.017269333203633625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,8191,0.01700266698996226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,16383,0.03200000027815501
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,fp8,16383,0.02027200038234393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,1,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,1,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,3,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,7,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,3,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,7,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,15,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,31,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,63,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,31,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,127,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,127,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,255,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,255,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,64,8,128,1,float16,fp8,2047,0.05064533154169718
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,511,0.010629333555698395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,511,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,1023,0.010656000425418219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,1023,0.010645333677530289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,15,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,2047,0.012746666868527731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,4095,0.012800000607967377
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,8191,0.014831999937693277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,16383,0.04776533444722494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,16383,0.031082667410373688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,4,128,1,float16,float16,2047,0.012752000242471695
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,1,0.015184000134468079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,1,0.014874666929244995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,3,0.015114666273196539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,3,0.014831999937693277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,7,0.014917333920796713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,7,0.015029333531856537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,15,0.015103999525308609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,15,0.01481066644191742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,31,0.014885333677132925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,31,0.014762666076421738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,63,0.015114666273196539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,63,0.014885333677132925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,127,0.015087999403476715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,127,0.015013333410024643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,fp8,2047,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,255,0.015013333410024643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,4095,0.013717333475748697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,8191,0.029290666182835896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,255,0.014864000181357065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,511,0.031082667410373688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,511,0.018730666488409042
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,1,0.01893866683046023
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,fp8,1023,0.03305600086847941
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,3,0.019007999449968338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,4,128,1,float16,float16,1023,0.04957333207130432
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,3,0.01725333308180173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,7,0.01700266698996226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,7,0.019152000546455383
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,15,0.016832000265518825
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,31,0.016976000120242436
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,64,8,128,1,float16,float16,63,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,63,0.01897066707412402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,15,0.019205333044131596
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,31,0.018826667219400406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,127,0.01893866683046023
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,127,0.017082666357358296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,255,0.01717866708834966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,255,0.03014400104681651
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,1,0.017557332913080852
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,511,0.03146133323510488
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,1023,0.05128533144791921
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,511,0.04934933284918467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,float16,1,0.02309333284695943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,float16,3,0.022730665902296703
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,float16,7,0.023007998863856
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,fp8,3,0.01921066641807556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,fp8,63,0.01684800038735072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,fp8,1,0.019610666980346043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,float16,15,0.022874665757020313
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,fp8,15,0.01933866615096728
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,float16,31,0.021210665504137676
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,fp8,7,0.01964266722400983
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,fp8,31,0.019754666835069656
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,fp8,63,0.01931200052301089
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,64,8,128,1,float16,float16,1023,0.08520000179608662
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,float16,63,0.022330666581789654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,fp8,127,0.019029332945744198
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,fp8,255,0.019120000302791595
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,float16,1,0.029653333127498627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,fp8,1,0.025290665527184803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,float16,255,0.03183999905983607
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,fp8,3,0.025472000241279602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,fp8,7,0.02515200028816859
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,float16,15,0.029114666084448498
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,fp8,15,0.025487999121348064
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,fp8,31,0.025242666403452556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,float16,31,0.029311999678611755
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,float16,63,0.029909332593282063
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,fp8,63,0.02537599951028824
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,4,128,1,float16,float16,127,0.02313599983851115
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,float16,127,0.03311999887228012
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,fp8,127,0.0252960001428922
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,float16,255,0.04957866668701172
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,fp8,255,0.0329066663980484
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,float16,1,0.03736533224582672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,float16,3,0.029680001238981884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,fp8,1,0.03151999910672506
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,float16,3,0.03770666569471359
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,fp8,3,0.03145066648721695
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,fp8,7,0.03139200061559677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,float16,15,0.03616533428430557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,float16,31,0.0372533326347669
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,float16,7,0.03793599953254064
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,fp8,31,0.031498665610949196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,float16,63,0.03721066564321518
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,fp8,15,0.03186666717131933
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,64,8,128,1,float16,float16,7,0.029520000020662945
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,fp8,63,0.0315733328461647
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,float16,127,0.03941333293914795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,4,128,1,float16,fp8,127,0.03126933425664902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,fp8,1,0.04388799766699473
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,float16,1,0.05223466455936432
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,float16,3,0.052933335304260254
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,float16,7,0.05193066596984863
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,fp8,3,0.04473066826661428
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,fp8,7,0.04496533175309499
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,float16,15,0.052005335688591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,fp8,15,0.045552000403404236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,fp8,31,0.04379733403523763
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,1,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,1,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,3,0.009925333162148794
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,3,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,7,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,fp8,63,0.04447466631730398
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,7,0.009354666496316591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,15,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,15,0.009648000200589498
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,31,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,63,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,63,0.00914666677514712
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,31,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,127,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,127,0.009306666751702627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,float16,127,0.05736533304055532
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,255,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,511,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,511,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,1023,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,2047,0.012794667234023413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,fp8,127,0.046480000019073486
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,2047,0.012736000120639801
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,4095,0.01421333352724711
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,1023,0.010911999891201654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,8191,0.017317333569129307
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,16383,0.048165331284205117
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,fp8,16383,0.031231999397277832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,1,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,8191,0.03001066545645396
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,1,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,3,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,3,0.010415999839703241
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,7,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,float16,31,0.05197866757710775
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,15,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,15,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,31,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,31,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,64,8,128,1,float16,float16,63,0.05287999908129374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,63,0.00860799973209699
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,127,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,255,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,255,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,511,0.010869332899649939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,511,0.010821333775917688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,255,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,1023,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,1023,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,2047,0.012831999609867731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,2047,0.012661332885424295
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,4095,0.02940266579389572
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,4,128,1,float16,float16,4095,0.014896000425020853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,4095,0.015135999768972397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,8191,0.0489279975493749
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,8191,0.031189332405726116
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,16383,0.04845866560935974
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,16383,0.08425066868464152
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,7,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,fp8,63,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,64,8,128,1,float16,float16,127,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,float16,1,0.06622933348019917
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,fp8,1,0.055770665407180786
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,fp8,3,0.056287998954455055
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,float16,3,0.06776000062624614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,float16,7,0.06635199983914693
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,fp8,7,0.05436266462008158
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,float16,15,0.06772799789905548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,fp8,15,0.05428266525268555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,float16,31,0.06619200110435486
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,fp8,31,0.0558240016301473
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,float16,63,0.06603200236956279
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,4,128,1,float16,fp8,63,0.05596800148487091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,float16,1,0.09293333689371745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,fp8,1,0.08092799782752991
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,float16,3,0.09355733791987102
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,fp8,3,0.0812906672557195
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,fp8,7,0.08273066580295563
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,float16,7,0.09329600135485332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,float16,15,0.09418666362762451
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,fp8,15,0.0820906658967336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,float16,31,0.09327999750773112
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,fp8,31,0.08059733112653096
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,float16,63,0.09329600135485332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,64,8,128,1,float16,fp8,63,0.08161599934101105
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,4,128,1,float16,fp8,1,0.10102400183677673
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,4,128,1,float16,float16,3,0.12122133374214172
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,4,128,1,float16,fp8,3,0.10123200217882793
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,4,128,1,float16,fp8,7,0.10225600004196167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,4,128,1,float16,float16,7,0.12184000015258789
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,4,128,1,float16,float16,1,0.12334932883580525
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,4,128,1,float16,float16,15,0.12402666608492534
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,4,128,1,float16,fp8,15,0.1020960013071696
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,4,128,1,float16,float16,31,0.12158399820327759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,4,128,1,float16,fp8,31,0.10129066308339436
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,8,128,1,float16,fp8,1,0.15189866224924722
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,8,128,1,float16,float16,3,0.17375467220942178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,8,128,1,float16,float16,7,0.173418660958608
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,8,128,1,float16,fp8,7,0.15197867155075073
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,1,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,1,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,8,128,1,float16,float16,1,0.17339734236399332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,3,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,3,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,7,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,15,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,7,0.009770666559537252
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,15,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,31,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,31,0.010181333248813948
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,63,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,8,128,1,float16,fp8,3,0.15120533108711243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,63,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,127,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,127,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,255,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,255,0.010602666685978571
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,511,0.01080000028014183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,511,0.010778666784365972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,1023,0.012479999413092932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,1023,0.012778667112191519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,2047,0.01481066644191742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,2047,0.013818666338920593
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,8191,0.048613334695498146
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,float16,4095,0.029733332494894665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,8191,0.03120533376932144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,1,0.009557333464423815
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,1,0.009568000212311745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,3,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,3,0.009392000113924345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,7,0.00926399976015091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,15,0.009445333232482275
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,15,0.009743999689817429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,31,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,63,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,63,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,8,128,1,float16,float16,15,0.1737920045852661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,8,128,1,float16,fp8,31,0.1516639987627665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,127,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,127,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,4,128,1,float16,fp8,4095,0.017221332838137943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,511,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,511,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,255,0.010048000141978264
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,7,0.00926399976015091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,1023,0.012773333738247553
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,1023,0.01292266696691513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,255,0.010266666611035665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,2047,0.029343999922275543
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,2047,0.014767999450365702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,8,128,1,float16,fp8,15,0.1514026621977488
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,4095,0.029520000020662945
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,4095,0.04717866579691569
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,31,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,fp8,8191,0.04782933493455251
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,1,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,1,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,64,8,128,1,float16,float16,31,0.17378666003545126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,7,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,15,0.009765333185593287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,3,0.00933333362142245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,15,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,31,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,63,0.010794666906197866
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,63,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,127,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,31,0.010768000036478043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,255,0.010672000547250112
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,255,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,511,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,511,0.010821333775917688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,1023,0.012869333227475485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,1023,0.012981332838535309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,2047,0.029152000943819683
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,2047,0.017130666722853977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,4095,0.04756799836953481
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,1,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,1,0.010703999549150467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,3,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,3,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,64,8,128,1,float16,float16,8191,0.08326933284600575
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,7,0.011050666371981302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,7,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,7,0.010656000425418219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,15,0.011141333729028702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,15,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,31,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,63,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,63,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,127,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,127,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,float16,127,0.009568000212311745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,255,0.011050666371981302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,255,0.010768000036478043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,511,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,511,0.012671999633312225
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,1023,0.029370665550231934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,1023,0.015103999525308609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,2047,0.04780266682306925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,1,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,1,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,4095,0.04950400193532308
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,3,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,4095,0.0849120020866394
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,3,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,7,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,7,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,4095,0.031157332162062328
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,15,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,31,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,31,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,63,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,63,0.008623999853928884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,127,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,127,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,255,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,255,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,511,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,511,0.0106133334338665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,1023,0.009674666449427605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,1023,0.010821333775917688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,2047,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,2047,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,4095,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,4095,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,8191,0.015263999501864115
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,8191,0.014949332922697067
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,16383,0.01716800034046173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,fp8,16383,0.017258666455745697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,float16,31,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,1,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,1,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,3,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,7,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,7,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,15,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,15,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,31,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,31,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,63,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,63,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,127,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,127,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,255,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,255,0.008602666358153025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,8,128,1,float16,fp8,2047,0.03236266722281774
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,511,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,1023,0.00943999985853831
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,1023,0.009226666763424873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,2047,0.010058666889866194
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,2047,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,4,128,1,float16,float16,15,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,4095,0.013023999830087027
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,4095,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,8191,0.015024000157912573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,8191,0.014736000448465347
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,16383,0.015157333264748255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,float16,16383,0.017210666090250015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,1,0.00914666677514712
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,1,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,3,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,3,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,7,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,7,0.009279999881982803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,31,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,31,0.009984000275532404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,63,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,63,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,127,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,127,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,255,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,255,0.009733333562811216
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,511,0.010501333822806677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,3,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,511,0.010565333068370819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,1023,0.010159999753038088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,1023,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,2047,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,2047,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,4095,0.013007999708255133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,4095,0.013082666943470636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,8191,0.014896000425020853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,8191,0.015125333021084467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,float16,16383,0.01708799973130226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,16383,0.017125333348910015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,1,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,3,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,48,8,128,1,float16,fp8,511,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,3,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,7,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,15,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,15,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,31,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,31,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,63,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,127,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,127,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,255,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,255,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,511,0.009205333267649015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,511,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,1023,0.009861333295702934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,1023,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,48,4,128,1,float16,fp8,3,0.010618666807810465
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,4,128,1,float16,fp8,15,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,2047,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,2047,0.011546666423479715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,4095,0.012847999731699625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,4095,0.013151999562978745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,8191,0.017184000462293625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,8191,0.015072000523408255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,16383,0.031157332162062328
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,16383,0.016869333883126576
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,1,0.011685332904259363
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,1,0.011509332805871964
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,3,0.011242666592200598
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,3,0.011178666104873022
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,7,0.01116266722480456
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,7,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,15,0.011781333635250727
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,15,0.011349332829316458
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,31,0.01116266722480456
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,1,0.0086666668454806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,31,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,63,0.011178666104873022
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,float16,7,0.008650666723648706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,63,0.012096000214417776
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,127,0.01118933285276095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,255,0.011333333949247995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,255,0.011109333485364914
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,511,0.014794666320085526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,48,8,128,1,float16,fp8,63,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,511,0.01312000056107839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,1023,0.029311999678611755
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,2047,0.04879466692606608
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,1,0.012874666601419449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,1,0.01310933381319046
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,3,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,7,0.012970666090647379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,3,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,15,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,31,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,31,0.013023999830087027
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,63,0.012906666845083237
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,63,0.012757333616415659
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,127,0.014970666418472925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,127,0.012960000584522883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,float16,127,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,255,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,255,0.013130666067202887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,511,0.029482667644818623
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,511,0.01716800034046173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,1023,0.04809066653251648
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,1023,0.03142933299144109
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,7,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,fp8,2047,0.050111999114354454
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,2047,0.08609066406885783
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,8,128,1,float16,float16,15,0.013199999928474426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,1,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,3,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,3,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,7,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,15,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,15,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,31,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,31,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,63,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,63,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,127,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,127,0.009663999701539675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,255,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,255,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,511,0.009717333440979322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,511,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,1023,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,1023,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,2047,0.012853333105643591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,2047,0.012831999609867731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,4095,0.01504533365368843
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,4095,0.015103999525308609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,8191,0.016842667013406754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,8191,0.016869333883126576
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,16383,0.033071999748547874
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,fp8,16383,0.01959466685851415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,1,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,1,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,3,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,7,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,2047,0.032138665517171226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,7,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,15,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,15,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,31,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,31,0.008623999853928884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,63,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,63,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,127,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,127,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,255,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,255,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,1,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,511,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,511,0.00984533317387104
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,1023,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,1023,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,4,128,1,float16,float16,7,0.008645333349704742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,2047,0.012485332787036896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,2047,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,4095,0.012794667234023413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,4095,0.014922666052977243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,8191,0.015279999623696009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,8191,0.029301332930723827
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,float16,16383,0.047466665506362915
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,16383,0.03136533250411352
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,1,0.015493333339691162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,1,0.014938666174809137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,3,0.014671999961137772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,7,0.014794666320085526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,7,0.01505600040157636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,15,0.01516266663869222
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,15,0.014874666929244995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,31,0.015109332899252573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,48,8,128,1,float16,fp8,3,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,31,0.01509333277742068
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,63,0.015018666783968607
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,48,4,128,1,float16,fp8,1023,0.017338667064905167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,63,0.014826666563749313
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,127,0.01522133375207583
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,127,0.014671999961137772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,255,0.015397333850463232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,511,0.019013332823912304
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,511,0.03136000037193298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,1,0.018944000204404194
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,1,0.01714133347074191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,1023,0.04959466556708018
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,1023,0.03331733246644338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,3,0.018906666586796444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,3,0.017184000462293625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,7,0.01905599981546402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,7,0.01729600007335345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,float16,3,0.014869333555301031
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,31,0.01720000058412552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,31,0.01903466631968816
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,63,0.019285333653291065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,63,0.01716800034046173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,127,0.018976000448067982
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,15,0.01721599946419398
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,15,0.018885333091020584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,127,0.017237332959969837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,4,128,1,float16,fp8,255,0.014864000181357065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,255,0.02932800104220708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,255,0.01720000058412552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,1023,0.08480532964070638
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,1023,0.050213331977526345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,float16,511,0.048938666780789696
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,48,8,128,1,float16,fp8,511,0.031498665610949196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,float16,1,0.02298133323589961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,fp8,1,0.019215999792019527
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,fp8,3,0.019280000279347103
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,float16,3,0.02329600105683009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,float16,7,0.022266666094462078
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,fp8,15,0.019167999426523846
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,float16,31,0.023232000569502514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,fp8,7,0.019248000035683315
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,float16,15,0.022970666488011677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,fp8,31,0.018992000569899876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,float16,63,0.021536000072956085
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,fp8,63,0.01922133316596349
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,float16,127,0.023157333334287006
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,fp8,127,0.01921066641807556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,float16,255,0.03141866624355316
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,float16,1,0.029215998947620392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,fp8,1,0.02554133286078771
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,4,128,1,float16,fp8,255,0.019226666539907455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,float16,3,0.02916266769170761
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,fp8,3,0.025360000630219776
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,float16,7,0.029301332930723827
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,fp8,7,0.025424001117547352
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,fp8,15,0.025818665822347004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,float16,31,0.02951466788848241
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,float16,15,0.029290666182835896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,fp8,31,0.025439999997615814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,float16,63,0.02903466671705246
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,fp8,63,0.025594666600227356
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,float16,127,0.033215999603271484
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,fp8,127,0.02510400116443634
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,fp8,255,0.03190933416287104
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,float16,1,0.037045332292715706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,fp8,1,0.0312266672650973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,float16,3,0.03729599962631861
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,fp8,3,0.03166399896144867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,float16,7,0.03742400060097376
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,fp8,7,0.03128000100453695
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,float16,15,0.037791999677817024
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,fp8,15,0.03148266673088074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,float16,31,0.03602666656176249
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,fp8,31,0.03146133323510488
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,48,8,128,1,float16,float16,255,0.04966400067011515
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,fp8,63,0.03138133386770884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,float16,63,0.037317333122094475
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,float16,127,0.03859733293453852
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,float16,1,0.05226666728655497
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,fp8,1,0.04385066529115041
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,4,128,1,float16,fp8,127,0.03162666658560435
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,float16,3,0.052576000491778054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,float16,7,0.05273599922657013
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,fp8,7,0.04582933088143667
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,float16,15,0.05236800014972687
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,fp8,3,0.045312002301216125
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,1,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,fp8,15,0.04378666480382284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,1,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,fp8,31,0.04566933214664459
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,float16,63,0.05274133384227753
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,3,0.009952000031868616
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,3,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,15,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,15,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,7,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,7,0.009343999748428663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,31,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,63,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,63,0.00973866693675518
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,127,0.009343999748428663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,fp8,63,0.04453866680463155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,255,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,255,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,511,0.010778666784365972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,1023,0.010874666273593903
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,1023,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,2047,0.012879999975363413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,2047,0.012549333274364471
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,4095,0.014773332824309668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,4095,0.014933332800865173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,fp8,127,0.04688533147176107
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,float16,127,0.05702400207519531
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,8191,0.029850666721661884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,8191,0.01708799973130226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,float16,16383,0.04859200119972229
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,1,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,3,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,7,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,7,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,16383,0.031744000812371574
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,1,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,15,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,48,8,128,1,float16,float16,31,0.05183466772238413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,15,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,31,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,31,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,63,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,63,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,127,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,127,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,255,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,511,0.010853332777818045
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,127,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,1023,0.011301333705584208
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,1023,0.010794666906197866
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,2047,0.013232000172138214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,4,128,1,float16,fp8,511,0.01109333336353302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,4095,0.02939733366171519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,4095,0.01516266663869222
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,8191,0.04785599807898203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,16383,0.08385599652926128
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,16383,0.0491946687301
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,3,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,255,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,float16,511,0.01090666651725769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,2047,0.012991999586423239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,48,8,128,1,float16,fp8,8191,0.031386665999889374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,float16,1,0.06776000062624614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,fp8,1,0.054117331902186074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,float16,3,0.0671253353357315
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,fp8,3,0.055402666330337524
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,float16,7,0.06621866424878438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,fp8,7,0.05522666871547699
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,float16,15,0.06594133377075195
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,fp8,15,0.05606933434804281
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,float16,31,0.0673280010620753
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,fp8,31,0.054048001766204834
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,float16,63,0.06789866586526234
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,4,128,1,float16,fp8,63,0.054714664816856384
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,float16,1,0.09331199526786804
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,fp8,1,0.08107733229796092
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,float16,3,0.09361599882443745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,fp8,3,0.08255466818809509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,float16,7,0.0934399962425232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,fp8,7,0.08227733274300893
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,float16,15,0.09294399619102478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,fp8,15,0.08267199993133545
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,float16,31,0.09289066990216573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,fp8,31,0.08097066481908162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,float16,63,0.09334400296211243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,48,8,128,1,float16,fp8,63,0.08291733264923096
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,4,128,1,float16,float16,1,0.12269866466522217
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,4,128,1,float16,fp8,1,0.10114666819572449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,4,128,1,float16,float16,3,0.12340799967447917
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,4,128,1,float16,fp8,3,0.10140267014503479
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,4,128,1,float16,float16,7,0.1229759951432546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,4,128,1,float16,float16,15,0.12192533413569133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,4,128,1,float16,fp8,7,0.10106133421262105
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,4,128,1,float16,fp8,15,0.10142933328946431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,8,128,1,float16,float16,1,0.17309333880742392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,4,128,1,float16,float16,31,0.1223466694355011
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,4,128,1,float16,fp8,31,0.10102933645248413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,8,128,1,float16,fp8,1,0.1502400040626526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,8,128,1,float16,float16,3,0.17352000872294107
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,1,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,1,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,8,128,1,float16,fp8,3,0.15187733372052512
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,3,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,3,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,8,128,1,float16,fp8,7,0.15133866667747498
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,7,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,7,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,15,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,8,128,1,float16,float16,7,0.1737333337465922
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,31,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,31,0.009642666826645533
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,63,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,63,0.010602666685978571
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,127,0.009296000003814697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,127,0.009322666873534521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,255,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,255,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,511,0.01110400011142095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,511,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,1023,0.013013333082199097
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,1023,0.012778667112191519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,2047,0.014736000448465347
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,2047,0.01681600014368693
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,4095,0.030837332208951313
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,4095,0.016976000120242436
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,float16,8191,0.048997332652409874
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,8191,0.03170666595300039
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,1,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,1,0.009935999910036722
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,3,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,3,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,8,128,1,float16,float16,15,0.17308266957600912
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,7,0.010309333602587381
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,8,128,1,float16,fp8,15,0.15036799510320029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,15,0.009189333145817121
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,31,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,15,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,4,128,1,float16,fp8,15,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,31,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,8,128,1,float16,float16,31,0.1735466718673706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,63,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,63,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,48,8,128,1,float16,fp8,31,0.15060800313949585
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,127,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,127,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,255,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,255,0.00960533320903778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,511,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,1023,0.011168000598748526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,2047,0.015066667149464289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,2047,0.029546665648619335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,4095,0.029103999336560566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,4095,0.04738666613896688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,511,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,8191,0.04826133449872335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,8191,0.08307200173536937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,1,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,float16,1023,0.013104000439246496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,1,0.01071999967098236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,3,0.009402666861812273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,3,0.010911999891201654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,7,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,15,0.010682666053374609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,31,0.009445333232482275
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,31,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,63,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,63,0.009712000067035357
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,48,8,128,1,float16,fp8,7,0.009514666472872099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,127,0.009392000113924345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,127,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,255,0.010703999549150467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,255,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,511,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,511,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,1023,0.013130666067202887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,2047,0.02924799919128418
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,1023,0.012965332716703415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,2047,0.01682666689157486
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,4095,0.04781333108743032
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,fp8,4095,0.03134933362404505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,1,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,1,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,3,0.010949333508809408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,7,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,3,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,15,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,15,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,31,0.010794666906197866
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,31,0.01108266661564509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,63,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,63,0.011141333729028702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,127,0.011109333485364914
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,127,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,255,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,255,0.010874666273593903
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,511,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,1023,0.029178666571776073
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,7,0.010608000059922537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,1023,0.015200000256299973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,4,128,1,float16,float16,15,0.009365333244204521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,2047,0.047781333327293396
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,1,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,2047,0.030858665704727173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,3,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,4095,0.04994133114814758
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,3,0.00960533320903778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,7,0.009226666763424873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,7,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,15,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,15,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,31,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,31,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,63,0.009557333464423815
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,127,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,127,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,255,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,255,0.009866666669646898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,511,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,511,0.01062400018175443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,1023,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,1023,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,2047,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,2047,0.01101333275437355
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,7,0.01110400011142095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,4095,0.011168000598748526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,8191,0.015034666905800501
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,8191,0.015173333386580149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,16383,0.01844800015290578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,16383,0.01738133281469345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,1,0.009317333499590555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,1,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,3,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,3,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,7,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,7,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,15,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,fp8,511,0.012901333471139273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,15,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,31,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,31,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,fp8,1,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,63,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,63,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,40,8,128,1,float16,float16,4095,0.08429333567619324
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,127,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,127,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,255,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,255,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,511,0.009317333499590555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,511,0.009712000067035357
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,1023,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,1023,0.010186666622757912
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,2047,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,4095,0.012789333860079447
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,8191,0.015109332899252573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,8191,0.014021333307027817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,16383,0.016837333639462788
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,16383,0.015552000453074774
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,4095,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,1,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,3,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,3,0.009279999881982803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,7,0.009226666763424873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,15,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,15,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,31,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,31,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,63,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,63,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,127,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,127,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,255,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,255,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,511,0.00949866697192192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,511,0.011120000233252844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,1023,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,1023,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,2047,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,2047,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,4095,0.013013333082199097
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,4095,0.014762666076421738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,8191,0.015119999647140503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,8191,0.015285332997639975
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,16383,0.01764800027012825
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,fp8,16383,0.016906666258970898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,fp8,2047,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,1,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,8,128,1,float16,float16,4095,0.012762666990359625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,3,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,3,0.008645333349704742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,7,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,7,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,1,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,15,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,31,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,31,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,63,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,63,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,127,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,4,128,1,float16,float16,7,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,127,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,255,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,511,0.009754666437705358
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,1023,0.010837333897749582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,1023,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,2047,0.011781333635250727
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,2047,0.011125333607196808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,4095,0.013797332843144735
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,8191,0.016010666886965435
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,float16,16383,0.03142933299144109
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,16383,0.016943999876578648
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,1,0.01180800050497055
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,1,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,3,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,40,4,128,1,float16,float16,63,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,3,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,1,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,7,0.012042666474978128
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,7,0.010837333897749582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,15,0.01108266661564509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,15,0.011253333340088526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,31,0.011173332730929056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,31,0.011461333682139715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,63,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,63,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,127,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,127,0.011973333855470022
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,511,0.009269333134094873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,255,0.011178666104873022
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,255,0.011274666835864386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,4095,0.013114667187134424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,511,0.014815999815861383
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,511,0.014373333503802618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,8191,0.014762666076421738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,1023,0.029696000119050343
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,1023,0.016901332885026932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,float16,2047,0.048538664976755776
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,1,0.013199999928474426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,1,0.012960000584522883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,7,0.013610667238632837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,7,0.012885333349307379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,3,0.012896000097195307
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,15,0.013066666821638743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,15,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,31,0.012874666601419449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,63,0.013167999684810638
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,63,0.013072000195582708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,127,0.014106666048367819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,127,0.012810666114091873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,40,8,128,1,float16,fp8,255,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,255,0.01403733342885971
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,255,0.01301866645614306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,511,0.029866665601730347
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,511,0.01699200024207433
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,1023,0.03145066648721695
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,1023,0.04860266546408335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,4,128,1,float16,fp8,2047,0.032469332218170166
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,2047,0.04986133178075155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,1,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,3,0.009269333134094873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,7,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,1,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,7,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,2047,0.08573333422342937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,15,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,31,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,63,0.008645333349704742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,63,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,127,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,127,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,255,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,255,0.009418666362762451
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,511,0.01090666651725769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,511,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,1023,0.01071999967098236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,1023,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,2047,0.012805332740147909
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,4095,0.014965333044528961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,4095,0.014773332824309668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,8191,0.016842667013406754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,8191,0.016458666572968166
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,16383,0.03310399999221166
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,16383,0.019285333653291065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,1,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,1,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,3,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,3,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,7,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,fp8,3,0.009392000113924345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,15,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,15,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,15,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,31,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,31,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,31,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,63,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,63,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,127,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,127,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,float16,31,0.013104000439246496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,255,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,255,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,511,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,511,0.010015999898314476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,1023,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,4,128,1,float16,float16,2047,0.012896000097195307
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,2047,0.011285333583752314
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,4095,0.012794667234023413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,4095,0.013839999834696451
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,8191,0.029365333418051403
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,16383,0.030085332691669464
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,16383,0.04779199759165446
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,1,0.014767999450365702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,1,0.014789332946141561
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,3,0.015173333386580149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,7,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,3,0.014959999670584997
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,7,0.015082667271296183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,7,0.014949332922697067
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,40,8,128,1,float16,fp8,3,0.01314666618903478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,15,0.015210667004187902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,15,0.014757333944241205
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,31,0.01505600040157636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,31,0.014922666052977243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,63,0.014975999792416891
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,63,0.014922666052977243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,127,0.015087999403476715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,1023,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,127,0.014805333067973455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,float16,2047,0.011760000139474869
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,255,0.014959999670584997
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,40,8,128,1,float16,fp8,8191,0.015024000157912573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,255,0.015082667271296183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,1,0.01882133384545644
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,511,0.031066666046778362
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,511,0.0189280000825723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,1,0.01720000058412552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,3,0.019199999670187633
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,float16,1023,0.04957333207130432
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,4,128,1,float16,fp8,1023,0.03339199970165888
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,3,0.01685333376129468
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,15,0.019088000059127808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,7,0.017050666113694508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,15,0.01718933383623759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,63,0.019088000059127808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,63,0.016890666137139004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,127,0.019066666563351948
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,127,0.017279999951521557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,31,0.01721599946419398
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,31,0.018826667219400406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,255,0.030026666820049286
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,255,0.017103999853134155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,1023,0.050554667909940086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,1023,0.08482133348782857
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,511,0.048469334840774536
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,fp8,511,0.031712000568707786
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,float16,1,0.022954667607943218
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,float16,3,0.022410665949185688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,fp8,1,0.01958400011062622
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,fp8,3,0.019173332800467808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,float16,7,0.021701333423455555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,40,8,128,1,float16,float16,7,0.019050666441520054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,fp8,15,0.019258666783571243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,fp8,7,0.01922133316596349
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,float16,31,0.023007998863856
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,fp8,31,0.019194666296243668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,float16,63,0.0229066660006841
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,fp8,63,0.01905599981546402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,float16,15,0.022613334159056347
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,float16,127,0.022704000274340313
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,fp8,127,0.018992000569899876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,float16,1,0.02917333443959554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,fp8,1,0.025733334322770435
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,float16,255,0.03084266682465871
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,4,128,1,float16,fp8,255,0.019909333437681198
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,float16,3,0.029450667401154835
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,fp8,3,0.025279998779296875
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,float16,7,0.029440000653266907
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,fp8,7,0.02535466601451238
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,fp8,15,0.025487999121348064
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,float16,15,0.02926933268706004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,float16,63,0.029130667448043823
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,fp8,63,0.02537599951028824
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,float16,127,0.03416533271471659
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,float16,31,0.029338667790095013
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,fp8,31,0.025807999074459076
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,fp8,127,0.025429333249727886
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,float16,1,0.03732266773780187
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,fp8,1,0.0315786674618721
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,fp8,255,0.03305066625277201
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,40,8,128,1,float16,float16,255,0.050661335388819374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,fp8,3,0.03143466760714849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,float16,7,0.03754133234421412
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,float16,3,0.03694933404525121
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,fp8,7,0.03158933420976003
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,float16,15,0.03584533433119456
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,fp8,15,0.031504000226656594
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,float16,63,0.03741333385308584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,fp8,63,0.03130666663249334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,fp8,31,0.03138133386770884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,float16,127,0.03938666731119156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,fp8,127,0.03130666663249334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,float16,1,0.05179733534653982
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,fp8,1,0.044026667873064675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,float16,3,0.052890668312708534
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,fp8,3,0.04520000020662943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,float16,7,0.05226666728655497
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,fp8,7,0.044591998060544334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,float16,15,0.05197333296140035
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,fp8,15,0.04391466577847799
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,4,128,1,float16,float16,31,0.037392000357309975
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,1,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,1,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,fp8,31,0.04365866879622141
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,3,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,3,0.009381333366036415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,float16,31,0.05198400219281515
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,7,0.01044800008336703
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,15,0.009162666896979014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,7,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,31,0.009765333185593287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,float16,63,0.05219733218352
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,fp8,63,0.04378666480382284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,63,0.009408000235756239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,127,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,float16,127,0.05570666491985321
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,255,0.009429333110650381
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,511,0.011136000355084738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,127,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,511,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,40,8,128,1,float16,fp8,127,0.0467199981212616
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,1023,0.010901333143313726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,2047,0.012842666357755661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,2047,0.012869333227475485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,1023,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,4095,0.013338666409254074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,8191,0.030037333567937214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,8191,0.017263999829689663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,16383,0.04789866507053375
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,4095,0.01481066644191742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,fp8,16383,0.031530665854612984
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,1,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,1,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,3,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,3,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,7,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,15,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,15,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,31,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,63,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,63,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,127,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,31,0.008661333471536636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,127,0.009445333232482275
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,255,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,255,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,511,0.01062400018175443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,511,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,1023,0.011178666104873022
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,255,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,1023,0.011114666859308878
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,2047,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,2047,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,4095,0.015066667149464289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,4095,0.029509333272775013
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,8191,0.030960001051425934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,8191,0.047744000951449074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,16383,0.04863999783992767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,16383,0.08359466989835103
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,fp8,7,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,8,128,1,float16,float16,31,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,40,4,128,1,float16,float16,63,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,fp8,1,0.05374933282534281
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,float16,1,0.06801066795984904
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,fp8,3,0.05621333420276642
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,float16,7,0.06635733445485432
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,fp8,7,0.054154664278030396
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,float16,15,0.06632533172766368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,fp8,15,0.05479466418425242
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,float16,3,0.0660693347454071
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,float16,31,0.0662773350874583
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,fp8,31,0.053904001911481224
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,float16,63,0.06630933284759521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,float16,1,0.09293867150942485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,4,128,1,float16,fp8,63,0.054058666030565895
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,fp8,1,0.08063466846942902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,float16,3,0.09388800462086995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,fp8,3,0.08061333497365315
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,float16,7,0.0932426651318868
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,fp8,7,0.08206399778525035
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,float16,15,0.09287466605504353
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,fp8,15,0.08057066798210144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,fp8,31,0.08062933385372162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,float16,31,0.09417066971460979
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,float16,63,0.09346666932106018
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,40,8,128,1,float16,fp8,63,0.08097066481908162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,4,128,1,float16,float16,1,0.1202186644077301
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,4,128,1,float16,fp8,1,0.10105066498120625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,4,128,1,float16,float16,3,0.12071999907493591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,4,128,1,float16,fp8,3,0.1011306643486023
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,4,128,1,float16,float16,7,0.12226667006810506
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,4,128,1,float16,fp8,7,0.10132799545923869
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,4,128,1,float16,float16,15,0.124208003282547
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,4,128,1,float16,fp8,15,0.10218666990598042
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,4,128,1,float16,float16,31,0.11982933680216472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,8,128,1,float16,float16,1,0.17282134294509888
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,8,128,1,float16,fp8,1,0.15007999539375305
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,8,128,1,float16,float16,3,0.17274133364359537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,8,128,1,float16,float16,7,0.173962672551473
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,8,128,1,float16,fp8,3,0.15064533551534018
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,1,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,1,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,3,0.009290666629870733
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,3,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,4,128,1,float16,fp8,31,0.10082133611043294
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,7,0.009797333429257074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,7,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,15,0.009194666519761086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,31,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,31,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,63,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,63,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,127,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,127,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,255,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,255,0.009375999992092451
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,511,0.010677333921194077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,8,128,1,float16,fp8,7,0.15043200055758157
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,511,0.010922666639089584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,1023,0.012831999609867731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,1023,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,2047,0.015087999403476715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,2047,0.014767999450365702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,4095,0.03073066721359889
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,4095,0.017130666722853977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,float16,8191,0.047781333327293396
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,8191,0.031311998764673867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,1,0.010026666646202406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,1,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,8,128,1,float16,float16,15,0.17337600390116373
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,3,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,3,0.009317333499590555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,4,128,1,float16,fp8,15,0.009749333063761393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,7,0.009839999799927076
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,8,128,1,float16,fp8,15,0.15007999539375305
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,15,0.00933333362142245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,7,0.00973866693675518
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,15,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,31,0.00960533320903778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,31,0.010117333382368088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,63,0.009573333586255709
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,127,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,127,0.0100426667680343
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,63,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,255,0.009706666693091393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,255,0.009989333028594652
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,511,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,511,0.010656000425418219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,1023,0.012746666868527731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,1023,0.012826666235923767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,2047,0.02942933390537898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,2047,0.014917333920796713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,4095,0.02932800104220708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,4095,0.047413334250450134
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,float16,8191,0.08334933718045552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,40,8,128,1,float16,fp8,8191,0.048101335763931274
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,1,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,1,0.009455999980370203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,8,128,1,float16,fp8,31,0.15019733707110086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,7,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,3,0.009663999701539675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,7,0.00973866693675518
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,15,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,15,0.009232000137368837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,63,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,127,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,127,0.009632000078757605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,31,0.009743999689817429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,255,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,255,0.010245333115259806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,511,0.011050666371981302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,511,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,1023,0.011887999872366587
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,1023,0.012117333710193634
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,2047,0.01469333345691363
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,2047,0.014495999862750372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,4095,0.031258667508761086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,4095,0.017082666357358296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,1,0.009461333354314169
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,1,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,8191,0.03194133440653483
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,40,8,128,1,float16,float16,31,0.17378133535385132
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,3,0.009722666814923286
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,3,0.009194666519761086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,7,0.009466666728258133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,7,0.009381333366036415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,15,0.009999999776482582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,31,0.009525333220760027
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,3,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,31,0.009530666594703993
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,15,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,63,0.009925333162148794
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,127,0.009754666437705358
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,255,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,255,0.010224000240365664
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,31,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,fp8,63,0.009701333319147428
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,511,0.01090666651725769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,511,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,1023,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,1023,0.01139733319481214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,2047,0.014741333822409311
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,4095,0.04771733283996582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,4095,0.03033600002527237
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,8191,0.047770669062932335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,8191,0.08297066887219746
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,1,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,1,0.010901333143313726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,3,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,3,0.010853332777818045
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,2,128,1,float16,float16,8191,0.04943466683228811
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,7,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,7,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,15,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,15,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,31,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,31,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,63,0.01101333275437355
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,127,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,127,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,255,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,63,0.009450666606426239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,255,0.010922666639089584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,fp8,127,0.010213333492477735
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,511,0.012800000607967377
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,511,0.013834666460752487
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,1023,0.02938133229811986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,1023,0.01524266724785169
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,2047,0.04780800143877665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,2047,0.031194667021433514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,4095,0.04990399877230326
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,4,128,1,float16,float16,2047,0.029157333076000214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,4095,0.0843999981880188
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,1,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,1,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,float16,8191,0.15691733360290527
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,3,0.008634666601816813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,3,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,7,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,15,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,15,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,31,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,31,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,63,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,63,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,127,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,255,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,255,0.010112000008424124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,511,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,511,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,1023,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,2047,0.010938666760921478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,2047,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,4095,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,4095,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,63,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,8191,0.015077333897352219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,8191,0.014778666198253632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,16383,0.019130667050679524
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,16383,0.018960000326236088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,32767,0.022997332115968067
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,fp8,32767,0.023103999594847362
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,1,0.008613333106040955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,1,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,3,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,3,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,7,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,15,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,15,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,31,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,31,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,63,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,63,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,127,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,127,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,255,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,255,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,32,8,128,1,float16,fp8,8191,0.08616532882054646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,511,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,511,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,1023,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,1023,0.009269333134094873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,2047,0.010901333143313726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,2047,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,127,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,4095,0.010591999938090643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,4095,0.010640000303586325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,8191,0.014917333920796713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,8191,0.015066667149464289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,16383,0.016917333006858826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,16383,0.016607999801635742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,32767,0.019391999890406925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,1023,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,1,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,1,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,3,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,3,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,7,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,7,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,15,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,15,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,31,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,31,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,63,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,63,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,127,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,255,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,255,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,511,0.009658666948477427
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,511,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,float16,7,0.00854399986565113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,1023,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,1023,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,2047,0.009925333162148794
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,4095,0.012975999464591345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,8191,0.015114666273196539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,8191,0.013701333353916803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,16383,0.015520000209410986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,16383,0.017055999487638474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,float16,32767,0.03346133232116699
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,1,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,2,128,1,float16,float16,7,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,32767,0.018810667097568512
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,1,0.00921066664159298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,3,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,3,0.009706666693091393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,7,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,7,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,15,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,15,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,31,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,31,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,63,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,4,128,1,float16,fp8,32767,0.018543999642133713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,63,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,127,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,127,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,255,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,255,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,511,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,511,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,1023,0.009663999701539675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,2047,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,2047,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,4095,0.010949333508809408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,4095,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,8191,0.015125333021084467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,8191,0.01505600040157636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,127,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,16383,0.017610666652520496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,16383,0.01775466650724411
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,32767,0.0207893339296182
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,fp8,32767,0.01924266666173935
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,1,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,1,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,3,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,2047,0.009423999736706415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,7,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,7,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,32,8,128,1,float16,fp8,4095,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,15,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,31,0.01033599985142549
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,63,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,63,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,127,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,127,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,255,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,255,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,511,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,511,0.009722666814923286
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,1023,0.009648000200589498
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,2047,0.00938666673998038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,1023,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,2047,0.010677333921194077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,4095,0.012917333592971167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,8191,0.014117332796255747
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,8191,0.014869333555301031
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,16383,0.016837333639462788
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,16383,0.015237333873907724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,32767,0.03219733387231827
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,32767,0.018960000326236088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,1,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,2,128,1,float16,float16,1023,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,1,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,3,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,3,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,7,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,7,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,15,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,15,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,31,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,31,0.008613333106040955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,63,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,fp8,3,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,127,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,127,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,255,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,255,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,511,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,511,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,1023,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,1023,0.010608000059922537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,2047,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,2047,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,4095,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,4095,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,8191,0.01504533365368843
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,8191,0.016789333273967106
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,16383,0.031871999303499855
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,16383,0.01708799973130226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,float16,32767,0.04991999765237173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,32767,0.03315199911594391
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,1,0.009375999992092451
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,4095,0.013013333082199097
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,3,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,3,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,7,0.009402666861812273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,7,0.010645333677530289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,15,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,15,0.010794666906197866
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,31,0.010634666929642359
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,31,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,63,0.009408000235756239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,63,0.010602666685978571
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,127,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,127,0.010629333555698395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,255,0.009232000137368837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,255,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,511,0.010938666760921478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,8,128,1,float16,fp8,63,0.009770666559537252
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,511,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,1023,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,1023,0.012885333349307379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,32,4,128,1,float16,float16,31,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,2047,0.02942933390537898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,float16,4095,0.04826666911443075
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,2047,0.016704000532627106
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,4095,0.03130666663249334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,1,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,1,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,3,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,7,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,3,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,15,0.01108266661564509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,2,128,1,float16,fp8,1,0.010768000036478043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,31,0.011055999745925268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,31,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,63,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,127,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,127,0.011130666981140772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,255,0.01109333336353302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,255,0.011055999745925268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,511,0.013253333667914072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,511,0.014042666802803675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,1023,0.015146666516860327
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,1023,0.029674666623274486
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,2047,0.04796266555786133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,2047,0.031173333525657654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,4095,0.08434666196505229
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,1,0.012944000462690989
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,4095,0.050250664353370667
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,1,0.013397333522637686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,3,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,3,0.012831999609867731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,7,0.013114667187134424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,7,0.012986666212479273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,15,0.012831999609867731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,15,0.013151999562978745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,15,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,float16,7,0.011600000162919363
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,31,0.012698666503032049
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,63,0.014442666123310724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,63,0.012944000462690989
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,4,128,1,float16,fp8,63,0.01119999960064888
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,127,0.013306666165590286
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,127,0.01310933381319046
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,255,0.01292266696691513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,255,0.01314666618903478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,1023,0.03126933425664902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,1023,0.048101335763931274
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,511,0.016885332763195038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,511,0.029781334102153778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,2047,0.08429333567619324
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,1,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,2047,0.049679999550183616
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,3,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,7,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,7,0.00960533320903778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,fp8,4095,0.08640533685684204
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,15,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,4095,0.15979199608167013
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,15,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,31,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,63,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,31,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,63,0.008634666601816813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,127,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,255,0.009162666896979014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,255,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,511,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,1023,0.009599999835093817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,2047,0.010645333677530289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,2047,0.01062400018175443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,4095,0.01331199953953425
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,4095,0.013557333499193192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,32,8,128,1,float16,float16,31,0.012847999731699625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,8191,0.015125333021084467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,16383,0.01720000058412552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,16383,0.017103999853134155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,32767,0.033589333295822144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,32767,0.02089600016673406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,1,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,1,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,3,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,3,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,7,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,15,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,15,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,31,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,3,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,31,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,127,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,63,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,511,0.009269333134094873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,63,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,1023,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,127,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,255,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,255,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,511,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,511,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,float16,8191,0.014997333288192749
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,1023,0.009765333185593287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,1023,0.009477333476146063
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,2047,0.011285333583752314
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,2047,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,4095,0.013376000026861826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,4095,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,8191,0.015706667055686314
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,8191,0.014698666830857595
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,16383,0.03136533250411352
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,16383,0.017738666385412216
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,32767,0.032399999598662056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,1,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,3,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,3,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,7,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,fp8,7,0.00916800027092298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,7,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,15,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,31,0.009205333267649015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,31,0.00921066664159298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,2,128,1,float16,fp8,1,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,63,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,63,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,127,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,127,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,255,0.008672000219424566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,255,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,127,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,511,0.009301333377758661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,511,0.009952000031868616
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,1023,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,1023,0.010677333921194077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,2047,0.011098666737476984
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,2047,0.011600000162919363
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,4095,0.013365333278973898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,4095,0.013066666821638743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,8191,0.029157333076000214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,8191,0.015189333508412043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,16383,0.04754666487375895
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,float16,32767,0.08380267024040222
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,32767,0.0491893341143926
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,1,0.012666666259368261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,1,0.011584000041087469
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,4,128,1,float16,float16,32767,0.049829334020614624
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,3,0.012479999413092932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,1,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,7,0.012805332740147909
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,7,0.011141333729028702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,15,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,15,0.011424000064531961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,15,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,31,0.012682666381200155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,63,0.011301333705584208
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,63,0.012383999923865
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,127,0.012709333250919977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,255,0.012661332885424295
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,255,0.011530666301647821
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,511,0.01504533365368843
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,511,0.013525333255529404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,32,8,128,1,float16,fp8,16383,0.030378667016824085
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,1023,0.03084266682465871
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,1,0.01370666672786077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,2047,0.049770668148994446
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,2047,0.03332799921433131
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,3,0.011424000064531961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,1,0.01312000056107839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,3,0.014538666854302088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,float16,31,0.011749333391586939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,7,0.012826666235923767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,15,0.01488000030318896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,15,0.012960000584522883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,3,0.012975999464591345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,127,0.011333333949247995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,31,0.01350933313369751
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,63,0.014848000059525171
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,63,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,127,0.013013333082199097
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,127,0.012810666114091873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,255,0.013167999684810638
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,255,0.013082666943470636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,2,128,1,float16,fp8,1023,0.01701333373785019
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,511,0.029877332349618275
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,511,0.01711999997496605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,1023,0.048954665660858154
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,2047,0.08616000413894653
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,2047,0.05150400102138519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,1,0.019071999937295914
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,1,0.017103999853134155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,3,0.01717866708834966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,7,0.01911466692884763
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,float16,7,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,31,0.013232000172138214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,15,0.019333332777023315
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,15,0.017242666333913803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,31,0.016895999511082966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,31,0.018954666952292126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,63,0.01915733392039935
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,63,0.01721599946419398
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,127,0.019013332823912304
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,127,0.017210666090250015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,4,128,1,float16,fp8,1023,0.031157332162062328
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,255,0.030762667457262676
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,255,0.017130666722853977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,511,0.03125333289305369
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,511,0.04836800197760264
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,7,0.016832000265518825
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,3,0.01899733394384384
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,1023,0.05022933085759481
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,2047,0.15707199772198996
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,float16,1023,0.08540266752243042
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,32,8,128,1,float16,fp8,2047,0.08774399757385254
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,float16,3,0.014874666929244995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,fp8,3,0.01481066644191742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,float16,1,0.015546667079130808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,fp8,1,0.015087999403476715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,float16,7,0.014970666418472925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,float16,15,0.015295999745527903
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,fp8,15,0.014874666929244995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,float16,31,0.015024000157912573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,fp8,31,0.014837333311637243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,fp8,63,0.014794666320085526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,float16,63,0.01515199989080429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,float16,127,0.01498666654030482
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,fp8,127,0.015061333775520325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,float16,255,0.015530666957298914
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,fp8,255,0.014885333677132925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,float16,511,0.031498665610949196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,fp8,511,0.018895999838908512
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,fp8,1,0.017103999853134155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,float16,3,0.019018666197856266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,fp8,3,0.01720000058412552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,float16,7,0.018885333091020584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,2,128,1,float16,fp8,7,0.014762666076421738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,fp8,15,0.017018667111794155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,fp8,7,0.017279999951521557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,float16,31,0.01894933357834816
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,fp8,31,0.01728533332546552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,float16,63,0.018906666586796444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,fp8,63,0.01714666684468587
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,float16,127,0.01911466692884763
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,fp8,127,0.0169813334941864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,float16,255,0.031231999397277832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,fp8,255,0.017077332983414333
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,float16,1,0.018911999960740406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,float16,511,0.04916800061861674
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,fp8,511,0.032144000132878624
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,fp8,1,0.025434667865435284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,float16,1,0.029450667401154835
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,float16,3,0.02920000006755193
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,float16,7,0.029264000554879505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,fp8,3,0.02521066615978877
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,4,128,1,float16,float16,15,0.019098666807015736
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,fp8,15,0.02531733363866806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,float16,31,0.02940800040960312
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,fp8,31,0.02533866713444392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,float16,63,0.02958400050799052
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,fp8,7,0.025253333151340485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,fp8,63,0.025226667523384094
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,float16,127,0.03315199911594391
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,float16,15,0.029370665550231934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,fp8,127,0.02532800038655599
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,fp8,255,0.03244800120592117
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,float16,255,0.04982399940490723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,float16,1,0.022821334501107533
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,fp8,1,0.019130667050679524
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,float16,3,0.023024000227451324
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,float16,511,0.08647466699282329
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,32,8,128,1,float16,fp8,511,0.050373335679372154
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,fp8,3,0.018933333456516266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,float16,7,0.02272533377011617
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,float16,15,0.022064000368118286
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,fp8,7,0.019493332753578823
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,fp8,15,0.019061333189407986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,float16,31,0.023157333334287006
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,fp8,63,0.018976000448067982
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,float16,127,0.02292266736427943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,fp8,127,0.01926400015751521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,fp8,31,0.019066666563351948
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,float16,1,0.029743999242782593
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,fp8,255,0.019120000302791595
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,fp8,1,0.027104000250498455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,float16,255,0.03147733211517334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,float16,3,0.030794667700926464
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,fp8,7,0.02741866558790207
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,fp8,3,0.02737066646416982
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,float16,7,0.029946667452653248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,fp8,15,0.027306665976842243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,float16,15,0.031301334500312805
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,float16,31,0.031471999982992806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,2,128,1,float16,float16,63,0.023007998863856
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,fp8,31,0.0271519993742307
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,float16,63,0.030453334252039593
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,fp8,63,0.027141332626342773
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,float16,127,0.03525333354870478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,fp8,127,0.027434666951497395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,float16,255,0.05192000170548757
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,4,128,1,float16,fp8,255,0.03557866563399633
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,float16,1,0.05198400219281515
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,fp8,1,0.043696001172065735
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,float16,3,0.051738664507865906
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,fp8,3,0.04378666480382284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,float16,7,0.05220800141493479
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,float16,15,0.05186133086681366
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,fp8,7,0.04377600053946177
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,float16,31,0.0518506666024526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,fp8,31,0.04364799956480662
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,float16,63,0.052485331892967224
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,fp8,15,0.04365866879622141
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,1,0.008645333349704742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,fp8,63,0.043562665581703186
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,3,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,3,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,7,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,15,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,15,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,1,0.00914666677514712
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,31,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,7,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,63,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,63,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,127,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,127,0.009599999835093817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,255,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,255,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,511,0.010490667074918747
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,float16,127,0.05606399973233541
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,fp8,127,0.046351999044418335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,511,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,1023,0.009733333562811216
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,1023,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,4095,0.014842666685581207
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,4095,0.014853333433469137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,8191,0.01621333385507266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,16383,0.03192000091075897
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,16383,0.019152000546455383
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,2047,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,fp8,255,0.0563679983218511
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,32767,0.05093333125114441
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,2047,0.012997332960367203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,32767,0.03519999980926514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,32,8,128,1,float16,float16,255,0.08881066242853801
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,1,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,3,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,7,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,15,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,31,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,63,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,63,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,127,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,127,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,255,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,255,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,fp8,31,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,511,0.00961599995692571
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,511,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,3,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,1023,0.010634666929642359
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,1023,0.010672000547250112
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,7,0.010191999996701876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,2047,0.011130666981140772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,2047,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,4095,0.014021333307027817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,4095,0.012853333105643591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,8191,0.015775999675194424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,8191,0.029109333952267964
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,16383,0.031370667119820915
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,32767,0.08398933211962382
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,fp8,32767,0.04947733382383982
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,1,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,2,128,1,float16,float16,8191,0.016789333273967106
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,3,0.00926399976015091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,3,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,7,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,7,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,15,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,15,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,31,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,31,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,63,0.009279999881982803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,127,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,127,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,255,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,255,0.009658666948477427
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,511,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,1023,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,1023,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,2047,0.013173333058754602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,2047,0.012736000120639801
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,4095,0.029370665550231934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,4095,0.015082667271296183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,8191,0.04781333108743032
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,8191,0.032069332897663116
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,16383,0.08356799681981404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,16383,0.04982399940490723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,16383,0.04794133206208547
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,1,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,32767,0.08469333251317342
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,32767,0.15609066685040793
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,fp8,63,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,4,128,1,float16,float16,1,0.009162666896979014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,32,8,128,1,float16,float16,511,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,float16,1,0.03769599894682566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,fp8,1,0.03134933362404505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,fp8,3,0.031354665756225586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,float16,3,0.03728000074625015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,float16,7,0.03732266773780187
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,fp8,7,0.03126933425664902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,float16,15,0.03679466744263967
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,fp8,15,0.031498665610949196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,fp8,31,0.03141333411137263
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,float16,31,0.037503999968369804
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,float16,63,0.03663466622432073
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,fp8,63,0.03154666721820831
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,float16,127,0.03932266682386398
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,2,128,1,float16,fp8,127,0.031583999594052635
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,fp8,1,0.04621866842110952
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,float16,1,0.053823997577031456
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,float16,3,0.05390933156013489
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,fp8,3,0.045893331368764244
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,float16,7,0.054378668467203774
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,fp8,7,0.04625066618124644
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,float16,15,0.05407999952634176
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,float16,31,0.05449600021044413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,fp8,15,0.0462719996770223
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,fp8,31,0.047728002071380615
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,float16,63,0.05427733560403188
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,fp8,63,0.04585599899291992
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,float16,127,0.05934933324654897
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,4,128,1,float16,fp8,127,0.049813335140546165
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,fp8,1,0.08032000064849854
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,float16,1,0.09287466605504353
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,float16,3,0.09332799911499023
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,fp8,3,0.08056533336639404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,float16,7,0.09295466542243958
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,fp8,7,0.08065066734949748
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,float16,31,0.09320533275604248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,float16,15,0.09297066926956177
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,fp8,31,0.08030933141708374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,float16,63,0.0932426651318868
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,fp8,63,0.08058133224646251
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,fp8,15,0.08032000064849854
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,float16,127,0.10019733508427937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,32,8,128,1,float16,fp8,127,0.08140799899895985
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,float16,1,0.06693866848945618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,fp8,1,0.05589333176612854
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,float16,3,0.0676746666431427
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,fp8,3,0.05598933498064677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,float16,7,0.06622399886449178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,fp8,7,0.055786664287249245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,float16,15,0.06840533514817555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,fp8,15,0.056015998125076294
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,fp8,31,0.054474666714668274
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,float16,31,0.06714666883150737
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,float16,63,0.0673333356777827
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,float16,1,0.09687999884287517
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,2,128,1,float16,fp8,63,0.05598933498064677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,fp8,1,0.086709330479304
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,float16,3,0.09686932961146037
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,fp8,3,0.08683733145395915
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,fp8,7,0.08779733379681905
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,float16,7,0.09856533010800679
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,float16,15,0.09809066851933797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,fp8,15,0.08814400434494019
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,float16,31,0.09674666325251262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,fp8,31,0.08658666412035625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,float16,63,0.09661333759625752
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,4,128,1,float16,fp8,63,0.08636800448099773
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,float16,1,0.17307200034459433
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,fp8,1,0.14915200074513754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,float16,3,0.1734773317972819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,fp8,3,0.14973333477973938
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,float16,7,0.17299199104309082
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,1,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,1,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,3,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,3,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,7,0.009226666763424873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,7,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,fp8,15,0.14919466773668924
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,15,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,fp8,7,0.14895466963450113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,15,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,float16,31,0.1728000044822693
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,fp8,31,0.1487573285897573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,31,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,31,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,63,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,63,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,127,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,255,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,511,0.01091733326514562
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,511,0.010591999938090643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,1023,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,127,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,float16,15,0.17356266578038534
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,1023,0.01101333275437355
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,2047,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,2047,0.012618667135636011
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,4095,0.014842666685581207
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,8191,0.030666666726271313
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,8191,0.017071999609470367
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,16383,0.04827733337879181
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,4095,0.014773332824309668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,fp8,16383,0.03190933416287104
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,1,0.00914666677514712
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,1,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,3,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,3,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,7,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,float16,63,0.1731520096460978
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,15,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,15,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,31,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,31,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,63,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,127,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,255,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,255,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,511,0.010672000547250112
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,511,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,1023,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,2047,0.013130666067202887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,127,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,2047,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,4095,0.02962133288383484
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,4095,0.015072000523408255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,2,128,1,float16,float16,255,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,8191,0.04773866633574168
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,8191,0.031386665999889374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,16383,0.08401067058245341
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,16383,0.05020266771316528
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,1,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,1,0.009306666751702627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,3,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,3,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,32,8,128,1,float16,fp8,63,0.14920533696810404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,7,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,7,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,15,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,31,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,7,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,63,0.009392000113924345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,63,0.009232000137368837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,127,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,127,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,255,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,255,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,511,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,1023,0.011765333513418833
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,511,0.01312000056107839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,float16,1023,0.011066666493813196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,31,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,2047,0.014848000059525171
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,2047,0.02940266579389572
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,4095,0.029834667841593426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,4095,0.048325334986050926
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,8191,0.08380800485610962
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,16383,0.08401067058245341
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,1,0.010778666784365972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,3,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,3,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,7,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,15,0.009397333487868309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,7,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,15,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,15,0.009402666861812273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,31,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,4,128,1,float16,fp8,63,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,63,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,63,0.009450666606426239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,127,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,127,0.010581333190202713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,255,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,255,0.009765333185593287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,511,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,1023,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,1023,0.012869333227475485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,1023,0.01292266696691513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,2047,0.014767999450365702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,2047,0.013232000172138214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,4095,0.031199999153614044
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,4095,0.017125333348910015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,fp8,8191,0.04794666667779287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,8191,0.049269333481788635
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,1,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,8191,0.03170666595300039
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,1,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,32,8,128,1,float16,float16,16383,0.15522666772206625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,3,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,float16,1,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,3,0.009258666386206945
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,7,0.009429333110650381
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,15,0.009621333330869675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,31,0.009658666948477427
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,63,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,31,0.010079999764760336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,63,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,127,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,255,0.009258666386206945
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,255,0.009455999980370203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,511,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,1023,0.013104000439246496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,1023,0.012853333105643591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,2047,0.029279999434947968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,2047,0.014848000059525171
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,2,128,1,float16,fp8,511,0.01108266661564509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,4095,0.04761599997679392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,4095,0.030133334298928578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,8191,0.08340799808502197
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,8191,0.048309331138928734
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,1,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,1,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,3,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,3,0.01089599976936976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,15,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,7,0.01091733326514562
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,7,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,15,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,15,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,127,0.009952000031868616
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,31,0.010949333508809408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,63,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,63,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,float16,511,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,127,0.011141333729028702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,127,0.010677333921194077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,255,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,255,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,511,0.012896000097195307
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,511,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,1023,0.01509333277742068
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,2047,0.04757866760094961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,2047,0.031285333136717476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,1,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,4095,0.08411733309427898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,4095,0.04961599906285604
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,fp8,8191,0.0864533285299937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,8191,0.15773866573969522
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,3,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,3,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,7,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,7,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,15,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,15,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,31,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,31,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,63,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,127,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,127,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,255,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,255,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,511,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,511,0.010629333555698395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,1023,0.010640000303586325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,2047,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,2047,0.010565333068370819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,4095,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,4095,0.01089599976936976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,31,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,16383,0.01926933353145917
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,16383,0.019007999449968338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,32767,0.02160000056028366
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,32767,0.021189334491888683
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,1,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,1,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,3,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,3,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,8,128,1,float16,float16,1023,0.029258665939172108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,7,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,7,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,15,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,15,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,31,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,31,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,63,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,1,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,63,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,127,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,127,0.009194666519761086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,255,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,255,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,511,0.010464000205198923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,1023,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,1023,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,2047,0.010768000036478043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,2047,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,4095,0.010682666053374609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,24,4,128,1,float16,fp8,7,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,4095,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,1023,0.009898666913310686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,8191,0.014826666563749313
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,16383,0.01714133347074191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,8191,0.015184000134468079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,32767,0.019173332800467808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,float16,8191,0.014997333288192749
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,2,128,1,float16,fp8,8191,0.01505600040157636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,32767,0.01842133328318596
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,1,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,3,0.008586666857202848
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,3,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,7,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,7,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,15,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,15,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,31,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,63,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,127,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,127,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,255,0.009253333633144697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,255,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,511,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,511,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,float16,511,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,1023,0.010213333492477735
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,1023,0.009232000137368837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,2047,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,2047,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,4095,0.012634667257467905
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,8191,0.01492799942692121
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,8191,0.014858666807413101
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,4,128,1,float16,fp8,16383,0.016906666258970898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,16383,0.01543466622630755
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,16383,0.016976000120242436
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,32767,0.018917333334684372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,32767,0.04193066557248434
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,1,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,1,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,float16,1,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,3,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,3,0.009343999748428663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,15,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,15,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,31,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,31,0.009402666861812273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,63,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,63,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,127,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,63,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,127,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,255,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,255,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,511,0.009957333405812582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,1023,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,1023,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,2047,0.00980266680320104
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,2047,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,4095,0.011055999745925268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,8191,0.014906667172908783
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,8191,0.015439999600251516
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,16383,0.017887999614079792
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,16383,0.01775466650724411
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,32767,0.01958400011062622
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,32767,0.01922133316596349
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,24,8,128,1,float16,fp8,4095,0.01293333371480306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,1,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,1,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,3,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,3,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,7,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,7,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,15,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,31,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,31,0.008581333483258883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,63,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,63,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,7,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,127,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,7,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,127,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,255,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,255,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,511,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,511,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,1023,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,2047,0.00956266683836778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,2047,0.009599999835093817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,31,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,4095,0.012874666601419449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,8191,0.015125333021084467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,8191,0.014378666877746582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,float16,511,0.010538666198650995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,16383,0.01758933315674464
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,32767,0.03316266586383184
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,32767,0.019023999571800232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,2,128,1,float16,fp8,4095,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,3,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,3,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,7,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,7,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,15,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,15,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,31,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,31,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,63,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,63,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,127,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,127,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,255,0.00860799973209699
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,1023,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,float16,4095,0.01293333371480306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,511,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,511,0.00943999985853831
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,4,128,1,float16,fp8,16383,0.015157333264748255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,1023,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,1023,0.009706666693091393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,2047,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,2047,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,4095,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,4095,0.013642666240533194
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,1,0.009359999870260557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,8191,0.01543466622630755
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,8191,0.014959999670584997
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,16383,0.01711999997496605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,16383,0.031370667119820915
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,32767,0.04985600213209788
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,32767,0.03217600037654241
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,1,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,1,0.010543999572594961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,3,0.01062400018175443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,3,0.010661333799362183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,7,0.010703999549150467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,7,0.01089599976936976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,15,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,15,0.010527999450763067
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,31,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,63,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,63,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,127,0.010634666929642359
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,127,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,255,0.009290666629870733
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,255,0.01071999967098236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,511,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,511,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,fp8,255,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,1023,0.013631999492645264
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,1023,0.012773333738247553
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,4095,0.04817600051561991
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,2047,0.029653333127498627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,2047,0.016714667280515034
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,fp8,4095,0.031178665657838184
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,1,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,1,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,3,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,24,8,128,1,float16,float16,1,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,7,0.01071999967098236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,15,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,3,0.010938666760921478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,7,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,31,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,63,0.01109333336353302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,31,0.010837333897749582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,63,0.01101333275437355
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,127,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,127,0.011141333729028702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,255,0.010911999891201654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,255,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,511,0.012858666479587555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,2,128,1,float16,float16,31,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,511,0.012991999586423239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,2047,0.04790399968624115
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,2047,0.031354665756225586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,4095,0.0498986691236496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,1023,0.029637334247430164
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,float16,4095,0.08449066678682964
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,1,0.012970666090647379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,1,0.012965332716703415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,3,0.01314666618903478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,7,0.012981332838535309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,7,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,15,0.013359999905029932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,31,0.013807999591032663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,63,0.014010666559139887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,63,0.012768000364303589
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,127,0.014378666877746582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,15,0.011055999745925268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,127,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,255,0.014122666170199713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,255,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,511,0.02939733366171519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,1023,0.04818133513132731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,1023,0.03141333411137263
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,4,128,1,float16,fp8,1023,0.01516266663869222
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,2047,0.049509331583976746
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,2047,0.08470400174458821
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,1,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,float16,4095,0.1591253379980723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,1,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,3,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,4095,0.08661866188049316
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,3,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,15,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,7,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,31,0.013023999830087027
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,15,0.00860799973209699
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,7,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,15,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,31,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,63,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,127,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,127,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,31,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,255,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,255,0.009599999835093817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,511,0.00922133338948091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,1023,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,1023,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,2047,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,2047,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,4095,0.013343999783198038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,4095,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,8191,0.015114666273196539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,16383,0.01716800034046173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,16383,0.01710933322707812
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,32767,0.03350399931271871
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,511,0.017162666966517765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,32767,0.021242665747801464
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,1,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,3,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,1,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,3,0.008586666857202848
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,7,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,7,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,15,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,15,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,31,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,31,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,63,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,63,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,127,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,127,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,255,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,255,0.008650666723648706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,511,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,511,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,1023,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,1023,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,2047,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,2047,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,63,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,4095,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,4095,0.012928000340859095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,8191,0.014981333166360855
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,8191,0.014837333311637243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,16383,0.0313226655125618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,16383,0.01718933383623759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,float16,32767,0.05004799862702688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,fp8,511,0.011146667102972666
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,1,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,1,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,3,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,7,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,2,128,1,float16,float16,8191,0.015082667271296183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,7,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,15,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,15,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,31,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,31,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,63,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,127,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,127,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,255,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,255,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,511,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,511,0.009423999736706415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,1023,0.010650667051474253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,2047,0.01109333336353302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,2047,0.012272000312805176
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,24,8,128,1,float16,fp8,3,0.012879999975363413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,4095,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,4095,0.012752000242471695
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,8191,0.029493334392706554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,16383,0.04763199885686239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,16383,0.031311998764673867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,32767,0.08432533343633015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,32767,0.04916266600290934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,1,0.012762666990359625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,4,128,1,float16,fp8,32767,0.033514666060606636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,1,0.012736000120639801
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,3,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,3,0.00955200009047985
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,3,0.012869333227475485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,7,0.012975999464591345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,7,0.012842666357755661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,15,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,float16,63,0.008639999975760778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,31,0.011882666498422623
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,15,0.011450666934251785
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,31,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,63,0.012037333101034164
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,1023,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,127,0.011978667229413986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,63,0.013429333766301474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,255,0.01309866706530253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,255,0.012053333222866058
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,24,8,128,1,float16,fp8,8191,0.01488000030318896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,127,0.011685332904259363
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,511,0.015072000523408255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,511,0.013093333691358566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,1023,0.017535999417304993
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,1,0.013503999759753546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,1,0.013013333082199097
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,1023,0.029701332251230877
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,3,0.01341333364446958
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,float16,2047,0.0488373339176178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,2,128,1,float16,fp8,2047,0.03206400076548258
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,7,0.013114667187134424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,15,0.013834666460752487
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,7,0.013013333082199097
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,3,0.01328533391157786
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,31,0.012784000486135483
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,15,0.013183999806642532
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,63,0.012853333105643591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,127,0.013397333522637686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,127,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,255,0.014858666807413101
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,255,0.012975999464591345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,31,0.013338666409254074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,63,0.013242666920026144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,511,0.030074665943781536
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,511,0.017029333859682083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,2047,0.08629866441090901
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,1,0.01911466692884763
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,2047,0.05062933266162872
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,1,0.01691199963291486
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,fp8,1023,0.031301334500312805
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,4,128,1,float16,float16,1023,0.04836266736189524
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,7,0.017157333592573803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,7,0.019029332945744198
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,3,0.018826667219400406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,15,0.018965333700180054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,3,0.016949333250522614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,15,0.016821333517630894
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,63,0.018778666853904724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,63,0.01740266631046931
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,127,0.017162666966517765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,127,0.018965333700180054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,255,0.016864000509182613
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,255,0.02938666691382726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,31,0.016901332885026932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,31,0.01918399954835574
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,1023,0.04983466863632202
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,511,0.04816000163555145
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,2047,0.08799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,2047,0.1564586659272512
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,float16,1,0.01504533365368843
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,fp8,1,0.015072000523408255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,float16,3,0.015146666516860327
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,float16,7,0.015317333241303762
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,fp8,7,0.014901333798964819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,float16,15,0.015119999647140503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,fp8,3,0.014885333677132925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,float16,31,0.015077333897352219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,fp8,15,0.014794666320085526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,fp8,31,0.015034666905800501
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,fp8,63,0.015029333531856537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,float16,127,0.015040000279744467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,float16,1023,0.08468799789746602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,float16,255,0.015119999647140503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,24,8,128,1,float16,fp8,511,0.03134933362404505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,fp8,255,0.014901333798964819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,float16,511,0.03143466760714849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,float16,1,0.01903466631968816
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,fp8,1,0.017130666722853977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,fp8,3,0.017242666333913803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,float16,3,0.019120000302791595
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,fp8,7,0.016864000509182613
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,float16,7,0.019002666076024372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,float16,15,0.019039999693632126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,float16,63,0.01509333277742068
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,fp8,15,0.01692266638080279
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,float16,31,0.019039999693632126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,fp8,31,0.017231999586025875
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,float16,63,0.019274666905403137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,fp8,63,0.017242666333913803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,float16,127,0.019258666783571243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,fp8,511,0.019029332945744198
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,fp8,127,0.016927999754746754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,fp8,255,0.016864000509182613
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,float16,255,0.031386665999889374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,float16,511,0.048613334695498146
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,float16,1,0.02961066613594691
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,4,128,1,float16,fp8,511,0.032698666055997215
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,fp8,1,0.025114665428797405
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,float16,3,0.029109333952267964
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,fp8,3,0.025455998877684276
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,float16,7,0.029487999776999157
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,2,128,1,float16,fp8,127,0.014698666830857595
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,float16,15,0.02924266705910365
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,float16,31,0.029103999336560566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,fp8,7,0.025429333249727886
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,fp8,31,0.025413334369659424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,fp8,15,0.025349333882331848
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,fp8,63,0.025173333783944447
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,float16,63,0.029311999678611755
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,float16,127,0.033071999748547874
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,fp8,127,0.025146665672461193
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,float16,255,0.04959466556708018
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,fp8,255,0.03152533372243246
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,float16,1,0.023200000325838726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,fp8,1,0.019226666539907455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,float16,3,0.021130666136741638
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,fp8,511,0.05022400120894114
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,24,8,128,1,float16,float16,511,0.0863200028737386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,fp8,3,0.019253333409627277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,float16,7,0.023034666975339253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,fp8,7,0.019098666807015736
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,float16,15,0.021920000513394673
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,fp8,15,0.01931200052301089
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,float16,31,0.021914665897687275
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,fp8,63,0.018986667195955913
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,float16,127,0.023007998863856
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,fp8,127,0.01923199991385142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,fp8,31,0.01921066641807556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,float16,63,0.022885332504908245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,float16,1,0.029088000456492107
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,fp8,255,0.019141333798567455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,2,128,1,float16,float16,255,0.03107200066248576
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,fp8,1,0.027482666075229645
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,float16,3,0.030239999294281006
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,float16,7,0.029418667157491047
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,fp8,3,0.027274665733178455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,fp8,7,0.02740799884001414
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,float16,15,0.02935466667016347
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,fp8,31,0.027445333699385326
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,fp8,15,0.027514666318893433
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,float16,31,0.029946667452653248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,float16,63,0.029637334247430164
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,fp8,63,0.0271519993742307
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,float16,127,0.03526400029659271
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,fp8,127,0.027301333844661713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,float16,1,0.052005335688591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,float16,255,0.05157866577307383
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,4,128,1,float16,fp8,255,0.03517866631348928
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,fp8,1,0.043706665436426796
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,float16,3,0.05230933427810669
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,fp8,3,0.043824002146720886
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,float16,7,0.051872000098228455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,float16,15,0.052095999320348106
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,fp8,7,0.04474133253097534
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,fp8,31,0.04371733466784159
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,float16,63,0.05301333467165629
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,fp8,63,0.04374399781227112
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,float16,31,0.052229334910710655
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,fp8,15,0.043920000394185386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,1,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,3,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,7,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,1,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,7,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,15,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,15,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,float16,127,0.05574933191140493
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,31,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,63,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,127,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,fp8,127,0.04621333380540212
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,127,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,255,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,255,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,511,0.010565333068370819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,1023,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,1023,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,63,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,2047,0.012815999488035837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,4095,0.014896000425020853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,511,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,8191,0.01716800034046173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,8191,0.016842667013406754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,16383,0.03322133421897888
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,16383,0.019498666127522785
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,32767,0.051632001996040344
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,32767,0.034474665919939675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,fp8,255,0.05500799914201101
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,1,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,24,8,128,1,float16,float16,255,0.08869333068529765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,1,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,3,0.00984533317387104
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,7,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,15,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,15,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,31,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,31,0.008661333471536636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,63,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,63,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,127,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,float16,3,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,127,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,255,0.008618666479984919
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,255,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,511,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,511,0.010634666929642359
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,1023,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,7,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,1023,0.010661333799362183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,2047,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,2047,0.01109333336353302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,4095,0.01523200049996376
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,4095,0.012709333250919977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,8191,0.029466666281223297
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,8191,0.01515199989080429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,16383,0.048437332113583885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,2047,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,float16,32767,0.0843999981880188
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,2,128,1,float16,fp8,4095,0.014778666198253632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,32767,0.049584001302719116
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,1,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,1,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,3,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,7,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,3,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,7,0.00916800027092298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,15,0.00972800018886725
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,15,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,31,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,63,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,127,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,127,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,255,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,255,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,511,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,511,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,1023,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,2047,0.012725333372751871
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,2047,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,4095,0.02938666691382726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,4095,0.01516266663869222
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,3,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,8191,0.047482664386431374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,8191,0.031082667410373688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,16383,0.08346666892369588
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,16383,0.049498667319615684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,32767,0.08474666873613994
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,32767,0.15633599956830344
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,4,128,1,float16,fp8,16383,0.03137599925200144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,float16,1023,0.01098666712641716
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,float16,1,0.03728000074625015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,fp8,1,0.03141866624355316
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,float16,3,0.037418665985266365
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,fp8,3,0.031248000760873158
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,24,8,128,1,float16,fp8,63,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,fp8,7,0.03156266609827677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,float16,15,0.03705599904060364
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,float16,7,0.03723733375469843
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,fp8,15,0.03159466634194056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,float16,63,0.03726933399836222
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,fp8,31,0.03125333289305369
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,float16,31,0.03753600021203359
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,fp8,63,0.03156266609827677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,float16,127,0.039264000952243805
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,2,128,1,float16,fp8,127,0.03197333216667175
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,float16,1,0.05395199855168661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,fp8,1,0.045893331368764244
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,float16,3,0.05400000015894572
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,fp8,3,0.04585599899291992
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,float16,7,0.05492799977461497
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,fp8,7,0.04691733419895172
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,float16,15,0.055493334929148354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,fp8,15,0.04594666759173075
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,float16,31,0.05394133428732554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,fp8,31,0.04587199787298838
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,float16,63,0.054234668612480164
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,fp8,63,0.045935998360315956
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,float16,127,0.059343998630841575
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,float16,1,0.09333333373069763
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,4,128,1,float16,fp8,127,0.04986133178075155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,fp8,1,0.08056533336639404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,float16,3,0.09276800354321797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,fp8,3,0.08073066671689351
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,float16,7,0.0925600032011668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,fp8,7,0.08035733302434285
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,float16,15,0.09293333689371745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,fp8,31,0.08025066554546356
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,float16,31,0.09356799721717834
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,float16,63,0.0927946666876475
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,fp8,63,0.08038933575153351
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,fp8,15,0.08026666442553203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,float16,127,0.09913600484530131
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,24,8,128,1,float16,fp8,127,0.08098133405049641
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,fp8,1,0.05374933282534281
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,float16,1,0.06614933411280315
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,float16,3,0.06828799843788147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,fp8,3,0.05532266696294149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,fp8,7,0.05394133428732554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,float16,7,0.06645333270231883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,float16,15,0.06740266581376393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,fp8,15,0.05398933092753092
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,fp8,31,0.054378668467203774
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,float16,63,0.06748799979686737
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,fp8,63,0.054661333560943604
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,fp8,1,0.08686932921409607
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,float16,1,0.09734400113423665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,2,128,1,float16,float16,31,0.06703466673692067
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,float16,3,0.09658666451772054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,fp8,3,0.08682666222254436
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,float16,7,0.09682666261990865
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,fp8,7,0.08738133311271667
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,float16,15,0.0976639986038208
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,fp8,15,0.08672533432642619
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,fp8,31,0.0867199997107188
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,float16,63,0.09700799981753032
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,float16,31,0.09742400050163269
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,4,128,1,float16,fp8,63,0.08675733208656311
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,float16,1,0.1726400057474772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,fp8,1,0.1483626663684845
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,float16,3,0.17358932892481485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,fp8,3,0.14900267124176025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,float16,7,0.17326400677363077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,fp8,7,0.15017066399256387
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,1,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,float16,15,0.17270400126775107
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,3,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,fp8,15,0.14855999747912088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,1,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,3,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,7,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,15,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,31,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,31,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,7,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,63,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,63,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,15,0.009962666779756546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,127,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,127,0.009626666704813639
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,255,0.00960533320903778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,511,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,1023,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,1023,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,2047,0.012698666503032049
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,2047,0.01292266696691513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,4095,0.014842666685581207
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,4095,0.014869333555301031
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,8191,0.029743999242782593
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,8191,0.017136000096797943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,16383,0.048112000028292336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,16383,0.03142933299144109
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,1,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,1,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,3,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,3,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,7,0.008672000219424566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,7,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,15,0.00933333362142245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,15,0.009258666386206945
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,float16,31,0.1726133426030477
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,float16,255,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,31,0.009365333244204521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,31,0.009904000287254652
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,2,128,1,float16,fp8,511,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,63,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,63,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,127,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,255,0.008650666723648706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,255,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,511,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,511,0.010837333897749582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,1023,0.010901333143313726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,2047,0.01314666618903478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,2047,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,fp8,31,0.14942933122316995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,4095,0.029157333076000214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,8191,0.04804266492525736
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,8191,0.030981334547201794
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,fp8,63,0.14828266700108847
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,16383,0.04967466493447622
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,1,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,float16,16383,0.08388800422350566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,1,0.010490667074918747
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,3,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,7,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,7,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,15,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,15,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,31,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,31,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,3,0.01009599988659223
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,63,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,63,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,127,0.009594666461149851
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,127,0.0100853331387043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,255,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,255,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,511,0.010645333677530289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,511,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,1023,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,127,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,1023,0.012869333227475485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,1023,0.012608000387748083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,4,128,1,float16,fp8,4095,0.015082667271296183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,2047,0.029274667302767437
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,2047,0.014837333311637243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,4095,0.04725333551565806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,4095,0.029663999875386555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,24,8,128,1,float16,float16,63,0.1739520033200582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,8191,0.08253333469231923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,8191,0.04778666794300079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,1,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,1,0.009343999748428663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,fp8,16383,0.0844693382581075
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,24,8,128,1,float16,float16,16383,0.15577066938082376
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,3,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,3,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,7,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,15,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,15,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,31,0.00933333362142245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,63,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,63,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,127,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,127,0.009413333609700203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,255,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,255,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,511,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,511,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,1023,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,1023,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,2047,0.012714666624863943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,2047,0.012938667088747025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,4095,0.015034666905800501
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,4095,0.014842666685581207
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,8191,0.01764800027012825
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,16383,0.049141332507133484
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,fp8,16383,0.03289066751797994
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,3,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,1,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,1,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,3,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,7,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,7,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,15,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,15,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,31,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,31,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,63,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,127,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,127,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,255,0.008650666723648706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,255,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,511,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,511,0.011194666226704916
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,1023,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,1023,0.01101333275437355
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,2047,0.012837332983811697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,2047,0.012778667112191519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,4095,0.029648000995318096
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,4095,0.01498666654030482
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,8191,0.04823466638724009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,8191,0.03130666663249334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,1,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,1,0.009381333366036415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,fp8,16383,0.04970666766166687
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,3,0.009941333283980688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,2,128,1,float16,float16,16383,0.08558932940165202
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,3,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,7,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,15,0.009839999799927076
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,15,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,31,0.009808000177145004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,31,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,63,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,63,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,8191,0.03002133220434189
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,1,128,1,float16,float16,7,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,127,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,127,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,255,0.00921066664159298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,255,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,511,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,2047,0.02906133234500885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,2047,0.01482133318980535
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,1023,0.011472000430027643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,4095,0.0472320020198822
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,4095,0.030271999537944794
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,8191,0.049082666635513306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,8191,0.08367466926574707
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,16383,0.08648000160853068
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,1,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,1,0.010842667271693548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,3,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,3,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,7,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,7,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,15,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,15,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,31,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,31,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,7,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,63,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,63,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,127,0.010874666273593903
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,127,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,255,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,255,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,511,0.013898666948080063
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,511,0.012768000364303589
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,fp8,511,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,1023,0.01488000030318896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,1023,0.029215998947620392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,1023,0.012805332740147909
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,2047,0.03090133269627889
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,4095,0.04951466619968414
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,4095,0.08434133728345235
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,8191,0.15702399611473083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,1,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,4,128,1,float16,float16,16383,0.15570666392644247
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,8191,0.08581866820653279
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,1,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,3,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,3,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,7,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,fp8,16383,0.1600106656551361
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,15,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,31,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,63,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,31,0.00972800018886725
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,127,0.008613333106040955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,127,0.009690666571259499
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,255,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,255,0.009338666374484697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,511,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,511,0.01071999967098236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,1023,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,1023,0.009861333295702934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,2047,0.010618666807810465
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,2047,0.009743999689817429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,4095,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,4095,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,8191,0.014778666198253632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,8191,0.015040000279744467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,16383,0.018277333428462345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,16383,0.01907733331123988
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,32767,0.027301333844661713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,32767,0.027114666998386383
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,65535,0.029461334149042766
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,65535,0.030586667358875275
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,1,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,1,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,3,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,3,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,7,0.008623999853928884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,7,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,2047,0.04757866760094961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,15,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,15,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,63,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,31,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,127,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,127,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,255,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,255,0.009685333197315535
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,7,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,511,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,float16,15,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,16,8,128,1,float16,float16,16383,0.3043733239173889
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,1023,0.009455999980370203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,2047,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,2047,0.009365333244204521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,4095,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,1,128,1,float16,fp8,63,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,4095,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,16383,0.019013332823912304
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,16383,0.018826667219400406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,32767,0.021322667598724365
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,32767,0.020997333029905956
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,65535,0.023205332458019257
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,65535,0.022815999885400135
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,1,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,1,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,3,0.0086666668454806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,7,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,7,0.008634666601816813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,15,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,15,0.008613333106040955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,31,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,31,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,63,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,127,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,63,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,127,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,255,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,511,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,255,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,1023,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,511,0.00985599992175897
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,1023,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,fp8,8191,0.013280000537633896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,1023,0.009695999945203463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,2047,0.009370666618148485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,4095,0.010640000303586325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,4095,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,8191,0.014869333555301031
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,8191,0.015087999403476715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,16383,0.017130666722853977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,16383,0.017162666966517765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,32767,0.01915733392039935
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,32767,0.0184906671444575
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,65535,0.02049066623051961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,1,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,1,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,3,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,3,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,7,0.008602666358153025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,7,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,15,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,15,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,511,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,2,128,1,float16,float16,8191,0.013157332936922709
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,31,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,31,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,63,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,2047,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,127,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,255,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,3,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,511,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,511,0.00997866690158844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,1023,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,1023,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,2047,0.009695999945203463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,2047,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,float16,65535,0.03559466699759165
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,4095,0.012842666357755661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,4095,0.013023999830087027
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,8191,0.014885333677132925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,8191,0.014117332796255747
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,16383,0.017237332959969837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,4,128,1,float16,fp8,63,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,32767,0.03333866596221924
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,32767,0.018543999642133713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,65535,0.03443199892838796
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,1,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,1,0.009797333429257074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,3,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,7,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,7,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,15,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,15,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,31,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,63,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,127,0.0086666668454806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,63,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,127,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,127,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,255,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,255,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,511,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,511,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,1023,0.009599999835093817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,1023,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,2047,0.009216000015536943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,4095,0.010682666053374609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,4095,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,8191,0.015082667271296183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,8191,0.015002666662136713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,16383,0.01921066641807556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,16383,0.019274666905403137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,32767,0.021253332495689392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,32767,0.02207999924818675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,float16,65535,0.024714666108290356
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,65535,0.023333333432674408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,1,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,1,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,3,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,3,0.0086666668454806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,7,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,fp8,16383,0.015178666760524115
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,7,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,15,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,15,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,31,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,63,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,63,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,127,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,127,0.008645333349704742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,255,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,3,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,255,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,511,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,511,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,1023,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,1023,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,2047,0.010538666198650995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,2047,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,4095,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,4095,0.010677333921194077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,8191,0.014959999670584997
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,8191,0.014688000082969666
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,16383,0.01681600014368693
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,16383,0.016970666746298473
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,32767,0.019189332922299702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,32767,0.01717866708834966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,65535,0.034559999903043113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,fp8,65535,0.020874666670958202
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,255,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,1,0.008586666857202848
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,1,128,1,float16,fp8,2047,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,3,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,3,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,7,0.008618666479984919
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,7,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,15,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,15,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,31,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,63,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,63,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,127,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,127,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,255,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,255,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,511,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,511,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,1023,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,2,128,1,float16,float16,31,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,2047,0.009674666449427605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,2047,0.009381333366036415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,16,8,128,1,float16,float16,65535,0.0507893313964208
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,4095,0.01302933320403099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,4095,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,8191,0.014874666929244995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,16383,0.017221332838137943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,8191,0.01488000030318896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,16383,0.015173333386580149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,32767,0.017152000218629837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,32767,0.03357866654793421
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,float16,65535,0.051962668697039284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,65535,0.03379733363787333
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,1,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,1,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,3,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,3,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,7,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,7,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,15,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,15,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,31,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,31,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,63,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,127,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,127,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,1,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,255,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,511,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,1023,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,1023,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,31,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,2047,0.010869332899649939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,2047,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,4095,0.013258667041858038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,4095,0.012800000607967377
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,8191,0.014853333433469137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,16383,0.03147733211517334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,16383,0.017263999829689663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,32767,0.04959466556708018
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,32767,0.0322773332397143
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,65535,0.05151999990145365
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,65535,0.08563199639320374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,1,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,1,0.00938666673998038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,3,0.009786666681369146
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,3,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,7,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,7,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,15,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,63,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,15,0.009808000177145004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,31,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,31,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,63,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,fp8,511,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,63,0.009695999945203463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,127,0.009743999689817429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,127,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,255,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,255,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,511,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,511,0.010821333775917688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,8191,0.01628799984852473
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,1023,0.012634667257467905
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,2047,0.014938666174809137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,2047,0.014767999450365702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,4,128,1,float16,fp8,1023,0.009930666536092758
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,4095,0.030437332888444264
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,4095,0.01758933315674464
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,1,0.009919999788204828
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,8191,0.049216002225875854
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,fp8,8191,0.03251733382542928
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,1,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,3,0.009359999870260557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,7,0.009632000078757605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,15,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,15,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,31,0.009941333283980688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,31,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,63,0.010015999898314476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,63,0.010314666976531347
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,16,8,128,1,float16,float16,255,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,127,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,127,0.009583999713261923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,255,0.009269333134094873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,255,0.009690666571259499
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,511,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,511,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,1023,0.012917333592971167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,1,128,1,float16,float16,1023,0.013066666821638743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,1023,0.012736000120639801
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,2047,0.029109333952267964
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,2047,0.014794666320085526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,4095,0.04790933430194855
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,4095,0.030405332644780476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,8191,0.08507733543713887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,fp8,8191,0.049551998575528465
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,3,0.01110400011142095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,3,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,1,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,7,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,1,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,7,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,3,0.009450666606426239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,2,128,1,float16,float16,7,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,15,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,31,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,63,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,63,0.010602666685978571
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,127,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,127,0.010778666784365972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,255,0.011050666371981302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,255,0.010869332899649939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,511,0.013781332721312841
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,511,0.012789333860079447
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,1023,0.030394665896892548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,1023,0.015317333241303762
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,2047,0.031040000418821972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,4095,0.048954665660858154
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,4095,0.08444266517957051
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,1,0.013776000589132309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,1,0.012970666090647379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,3,0.01314666618903478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,fp8,8191,0.0881866713364919
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,3,0.01292266696691513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,7,0.01302933320403099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,15,0.013765333841244379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,7,0.012896000097195307
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,15,0.01109333336353302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,31,0.011050666371981302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,31,0.012917333592971167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,31,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,63,0.013237333546082178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,63,0.012831999609867731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,127,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,255,0.01301866645614306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,255,0.012815999488035837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,2047,0.047968000173568726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,511,0.02958933264017105
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,1023,0.04808533191680908
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,1023,0.031173333525657654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,2047,0.08522666494051616
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,2047,0.0498933345079422
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,4,128,1,float16,float16,8191,0.15625066558519998
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,4095,0.08514666557312012
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,4095,0.15847466389338175
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,1,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,1,0.009621333330869675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,3,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,3,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,7,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,15,0.012981332838535309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,float16,8191,0.30912532409032184
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,7,0.009381333366036415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,15,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,8191,0.15878933668136597
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,31,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,63,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,31,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,127,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,255,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,255,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,511,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,511,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,1023,0.009285333255926767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,1023,0.010591999938090643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,2047,0.010543999572594961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,2047,0.010565333068370819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,4095,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,127,0.01310933381319046
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,8191,0.014933332800865173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,8191,0.015141333142916361
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,16383,0.017279999951521557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,16383,0.017407999684413273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,32767,0.0205226664741834
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,32767,0.019002666076024372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,float16,65535,0.036144000788529716
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,16,8,128,1,float16,fp8,511,0.017018667111794155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,1,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,1,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,3,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,7,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,3,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,7,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,15,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,15,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,31,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,31,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,63,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,63,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,127,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,127,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,255,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,255,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,511,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,511,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,1023,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,1023,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,2047,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,15,0.00978133330742518
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,2047,0.009253333633144697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,4095,0.012858666479587555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,4095,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,8191,0.014864000181357065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,127,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,8191,0.014442666123310724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,16383,0.015087999403476715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,32767,0.03297599901755651
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,32767,0.019013332823912304
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,1,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,fp8,65535,0.03342933456103007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,1,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,3,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,3,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,7,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,4095,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,15,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,15,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,31,0.008592000231146812
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,31,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,63,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,127,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,127,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,255,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,511,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,511,0.01080000028014183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,1023,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,1023,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,2047,0.011503999431928
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,2047,0.011125333607196808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,4095,0.013754667093356451
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,4095,0.012821332861979803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,8191,0.017018667111794155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,16383,0.031285333136717476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,16383,0.01701333373785019
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,16383,0.016757333030303318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,32767,0.0498986691236496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,2,128,1,float16,float16,65535,0.050240000089009605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,32767,0.03310399999221166
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,7,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,65535,0.05202666421731313
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,float16,65535,0.08562133709589641
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,1,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,1,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,63,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,7,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,1,128,1,float16,fp8,65535,0.022650666534900665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,7,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,255,0.008661333471536636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,15,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,15,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,31,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,31,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,63,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,63,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,127,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,127,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,255,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,255,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,511,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,511,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,1023,0.01071999967098236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,1023,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,2047,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,2047,0.011503999431928
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,4095,0.014538666854302088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,4,128,1,float16,fp8,8191,0.014858666807413101
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,4095,0.012847999731699625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,8191,0.015130666395028433
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,8191,0.029450667401154835
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,16383,0.04775999983151754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,16383,0.030506665507952373
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,32767,0.08407466610272725
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,32767,0.0481333335240682
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,65535,0.15653866529464722
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,65535,0.08608532945315044
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,1,0.009695999945203463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,1,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,3,0.00997866690158844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,3,0.010672000547250112
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,7,0.010634666929642359
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,7,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,float16,3,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,15,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,15,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,31,0.010869332899649939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,31,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,63,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,63,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,127,0.010629333555698395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,127,0.010597333312034607
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,255,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,255,0.010922666639089584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,511,0.011120000233252844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,511,0.011168000598748526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,1023,0.014101333916187286
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,2047,0.030031998952229817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,1,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,2047,0.016901332885026932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,1,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,float16,4095,0.047775998711586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,4095,0.031189332405726116
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,3,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,3,0.011226666470368704
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,7,0.011215999722480774
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,16,8,128,1,float16,fp8,3,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,31,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,31,0.01109333336353302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,63,0.010869332899649939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,15,0.010778666784365972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,63,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,15,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,127,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,127,0.01121066634853681
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,255,0.01098666712641716
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,255,0.010837333897749582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,511,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,511,0.012789333860079447
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,1,128,1,float16,fp8,1023,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,1023,0.029626667499542236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,1023,0.015087999403476715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,4095,0.08572799960772197
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,1,0.013765333841244379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,4095,0.050255998969078064
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,float16,2047,0.04790399968624115
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,2047,0.03109866629044215
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,1,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,3,0.012981332838535309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,3,0.013114667187134424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,7,0.013343999783198038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,7,0.013242666920026144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,31,0.01313599944114685
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,31,0.012986666212479273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,63,0.012981332838535309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,127,0.01310933381319046
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,15,0.013104000439246496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,15,0.013594667116800943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,127,0.012858666479587555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,255,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,255,0.013023999830087027
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,511,0.02916266769170761
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,511,0.01695466662446658
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,1023,0.04855466882387797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,2047,0.05054933329423269
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,2047,0.08567999800046285
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,4095,0.15832533439000449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,4095,0.08866133292516072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,1,0.017173333714405697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,3,0.018981333822011948
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,3,0.01720533271630605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,1,0.019071999937295914
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,float16,63,0.013850666582584381
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,7,0.018933333456516266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,7,0.016997333616018295
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,15,0.019029332945744198
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,15,0.01693333312869072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,31,0.017008000363906223
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,63,0.016869333883126576
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,127,0.019093333433071773
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,127,0.017210666090250015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,255,0.029487999776999157
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,4,128,1,float16,fp8,1023,0.03181866556406021
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,255,0.017173333714405697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,2,128,1,float16,fp8,7,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,511,0.04775466521581014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,511,0.03143466760714849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,1023,0.0839573343594869
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,1023,0.049770668148994446
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,2047,0.1560426652431488
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,2047,0.08712533116340637
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,31,0.018906666586796444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,63,0.018789333601792652
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,1,0.012986666212479273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,1,0.011711999773979187
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,float16,4095,0.299562672773997
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,3,0.01268799975514412
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,3,0.012757333616415659
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,7,0.011834666132926941
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,7,0.012997332960367203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,15,0.011770666887362799
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,15,0.012917333592971167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,31,0.012863999853531519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,31,0.011839999506870905
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,63,0.012736000120639801
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,63,0.011760000139474869
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,127,0.012191999703645706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,127,0.012965332716703415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,255,0.012757333616415659
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,255,0.012730666746695837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,511,0.014975999792416891
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,511,0.014874666929244995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,16,8,128,1,float16,fp8,4095,0.15983999768892923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,float16,1023,0.030410667260487873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,1,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,1,128,1,float16,fp8,1023,0.017263999829689663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,3,0.014138666292031607
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,3,0.013141332815090815
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,7,0.013461332768201828
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,7,0.013194666554530462
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,15,0.012847999731699625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,15,0.013114667187134424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,31,0.014479999740918478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,63,0.012949333836634954
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,63,0.012837332983811697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,127,0.012901333471139273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,127,0.012901333471139273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,255,0.013104000439246496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,255,0.013082666943470636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,511,0.02991466720898946
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,511,0.016794666647911072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,1,0.013381333400805792
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,float16,1023,0.04940799872080485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,1,0.019120000302791595
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,1023,0.03146666785081228
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,1,0.01735466718673706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,3,0.018858666221300762
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,3,0.016821333517630894
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,7,0.0189280000825723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,2,128,1,float16,fp8,31,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,15,0.018853332847356796
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,31,0.019002666076024372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,7,0.0170666662355264
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,31,0.01693333312869072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,63,0.01916266605257988
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,63,0.01714133347074191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,127,0.019189332922299702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,127,0.016943999876578648
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,255,0.03049066662788391
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,255,0.017162666966517765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,511,0.04822400212287903
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,511,0.031530665854612984
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,1,0.029103999336560566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,1,0.02537599951028824
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,3,0.02923733244339625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,15,0.016842667013406754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,7,0.02513599892457326
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,fp8,1023,0.049829334020614624
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,4,128,1,float16,float16,1023,0.08468266328175862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,15,0.02917333443959554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,15,0.02537599951028824
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,31,0.028991999725500744
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,31,0.02532266577084859
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,127,0.025045332809289295
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,127,0.031717332700888314
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,255,0.03275199979543686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,255,0.0487360010544459
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,7,0.029306667546431225
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,63,0.02924799919128418
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,63,0.025455998877684276
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,511,0.08467732866605122
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,511,0.04958933095137278
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,float16,1023,0.15662933389345804
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,float16,1,0.015493333339691162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,fp8,1,0.014773332824309668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,float16,3,0.015392000476519266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,3,0.025285333395004272
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,fp8,7,0.014842666685581207
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,float16,15,0.015647999942302704
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,fp8,15,0.01481066644191742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,fp8,3,0.01505600040157636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,float16,7,0.015658666690190632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,float16,31,0.015658666690190632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,16,8,128,1,float16,fp8,1023,0.08640533685684204
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,fp8,31,0.01469333345691363
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,float16,63,0.015141333142916361
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,fp8,63,0.014837333311637243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,float16,127,0.015018666783968607
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,fp8,127,0.01504533365368843
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,float16,255,0.015210667004187902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,fp8,255,0.01482133318980535
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,float16,511,0.03147733211517334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,1,128,1,float16,fp8,511,0.01897066707412402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,float16,1,0.019013332823912304
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,fp8,1,0.017071999609470367
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,float16,3,0.01886933296918869
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,fp8,3,0.01721599946419398
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,float16,7,0.019141333798567455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,fp8,7,0.017338667064905167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,fp8,15,0.01692266638080279
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,float16,31,0.01918399954835574
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,fp8,31,0.01717866708834966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,float16,15,0.018992000569899876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,float16,63,0.019141333798567455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,fp8,63,0.017125333348910015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,float16,127,0.018863999595244724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,fp8,127,0.017077332983414333
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,fp8,255,0.017258666455745697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,float16,255,0.030799999833106995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,float16,511,0.04956266780694326
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,2,128,1,float16,fp8,511,0.03148266673088074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,float16,1,0.02996266633272171
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,fp8,1,0.027317332724730175
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,float16,3,0.029359998802344005
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,fp8,3,0.027301333844661713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,float16,7,0.03128000100453695
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,float16,31,0.030069333811601002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,fp8,31,0.02735999971628189
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,fp8,7,0.02733866622050603
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,float16,15,0.029440000653266907
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,fp8,15,0.027056001126766205
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,float16,63,0.02942399928967158
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,fp8,63,0.027461332579453785
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,float16,127,0.035216001172860466
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,fp8,127,0.02703999976317088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,float16,255,0.05053333441416422
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,float16,1,0.05186133086681366
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,fp8,255,0.03550933301448822
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,fp8,1,0.04364799956480662
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,float16,511,0.08713600039482117
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,float16,3,0.052832002441088356
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,fp8,3,0.04385066529115041
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,fp8,7,0.04354133208592733
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,float16,7,0.0517439991235733
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,fp8,15,0.04365866879622141
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,fp8,31,0.04385599990685781
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,float16,63,0.051669334371884666
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,float16,15,0.05193066596984863
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,4,128,1,float16,fp8,511,0.0536053329706192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,fp8,63,0.04386133452256521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,float16,127,0.05574400226275126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,1,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,float16,31,0.051632001996040344
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,3,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,fp8,127,0.04645866652329763
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,7,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,15,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,15,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,3,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,31,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,31,0.009237333511312803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,63,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,1,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,63,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,127,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,127,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,255,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,255,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,511,0.00933333362142245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,float16,255,0.08846400181452434
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,511,0.010821333775917688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,1023,0.010015999898314476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,7,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,fp8,255,0.056320001681645714
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,1023,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,2047,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,2047,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,4095,0.012901333471139273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,16383,0.018207999567190807
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,16383,0.016997333616018295
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,32767,0.021002667645613354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,32767,0.033376000821590424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,65535,0.051914667089780174
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,65535,0.03561066587766012
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,1,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,3,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,3,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,7,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,7,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,15,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,4095,0.014218666901191076
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,31,0.008597333605090777
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,float16,8191,0.015157333264748255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,31,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,63,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,63,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,127,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,127,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,255,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,float16,511,0.16012799739837646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,255,0.008623999853928884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,511,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,1023,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,1023,0.010618666807810465
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,2047,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,2047,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,1,128,1,float16,fp8,8191,0.01504533365368843
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,4095,0.01312000056107839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,4095,0.013151999562978745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,8191,0.01470400020480156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,16383,0.031317333380381264
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,16383,0.017184000462293625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,1,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,32767,0.033530667424201965
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,511,0.01062400018175443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,65535,0.0867199997107188
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,16,8,128,1,float16,fp8,511,0.08989333113034566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,65535,0.05243733525276184
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,1,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,1,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,fp8,15,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,3,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,7,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,7,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,15,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,15,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,31,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,31,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,63,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,63,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,127,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,127,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,255,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,255,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,511,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,511,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,1023,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,1023,0.010842667271693548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,2047,0.011120000233252844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,2047,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,4095,0.012965332716703415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,4095,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,8191,0.01515199989080429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,8191,0.029290666182835896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,16383,0.04781333108743032
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,16383,0.03109866629044215
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,32767,0.08409600456555684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,32767,0.04966400067011515
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,3,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,8191,0.01578666642308235
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,float16,65535,0.15616533160209656
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,1,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,4,128,1,float16,fp8,65535,0.08676266670227051
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,3,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,3,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,7,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,7,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,2,128,1,float16,float16,32767,0.049584001302719116
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,15,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,15,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,31,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,31,0.009658666948477427
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,63,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,127,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,127,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,255,0.009242666885256767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,255,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,511,0.010778666784365972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,511,0.010853332777818045
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,1023,0.011120000233252844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,2047,0.012805332740147909
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,4095,0.029215998947620392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,4095,0.014837333311637243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,8191,0.047482664386431374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,8191,0.03046933313210805
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,16383,0.08428800106048584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,16383,0.04860266546408335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,32767,0.08524266878763835
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,32767,0.15562666455904642
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,65535,0.1566933294137319
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,65535,0.29972267150878906
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,fp8,1,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,63,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,1023,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,16,8,128,1,float16,float16,2047,0.013343999783198038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,fp8,1,0.019637333850065868
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,float16,1,0.023077333966890972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,fp8,3,0.019285333653291065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,float16,3,0.023157333334287006
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,float16,7,0.022991999983787537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,fp8,7,0.0191040001809597
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,float16,15,0.02260799954334895
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,fp8,15,0.020213333268960316
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,float16,31,0.02309333284695943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,fp8,31,0.020010666300853092
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,float16,63,0.022618666291236877
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,fp8,63,0.01964266722400983
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,float16,127,0.023002666731675465
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,fp8,127,0.01926933353145917
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,float16,255,0.03143999973932902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,1,128,1,float16,fp8,255,0.01942933350801468
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,float16,1,0.03030399978160858
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,fp8,1,0.027376001079877216
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,float16,3,0.031231999397277832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,fp8,3,0.027493332823117573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,float16,7,0.030789333085219067
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,fp8,7,0.02771199991305669
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,fp8,15,0.02739733209212621
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,float16,15,0.030058667063713074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,float16,31,0.02943466603755951
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,fp8,31,0.027119999130566914
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,fp8,63,0.02738133321205775
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,float16,63,0.029605334003766377
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,float16,127,0.035536001125971474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,fp8,127,0.02749866743882497
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,float16,1,0.05376533170541128
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,float16,255,0.05154666801293691
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,fp8,1,0.04576533536116282
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,2,128,1,float16,fp8,255,0.03602133442958196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,float16,3,0.05468800167242686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,fp8,3,0.04593066871166229
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,float16,7,0.054010664423306785
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,fp8,7,0.04586133360862732
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,float16,15,0.05386666456858317
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,fp8,15,0.045781334241231285
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,float16,31,0.0539626677831014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,float16,63,0.053770666321118675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,fp8,31,0.04572266836961111
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,fp8,63,0.04562666515509287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,float16,127,0.0581279993057251
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,float16,1,0.09265066186587016
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,fp8,127,0.0498879998922348
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,fp8,1,0.080485333998998
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,fp8,255,0.05834133426348368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,4,128,1,float16,float16,255,0.09082667032877605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,float16,3,0.09283199906349182
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,fp8,3,0.08035733302434285
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,float16,7,0.09245866537094116
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,fp8,7,0.07970666885375977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,float16,15,0.09220266342163086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,fp8,15,0.08065066734949748
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,float16,63,0.09311999877293904
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,fp8,63,0.0804799993832906
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,float16,31,0.09258666634559631
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,float16,127,0.09868266185124715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,fp8,31,0.08057066798210144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,fp8,127,0.0809440016746521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,fp8,255,0.09794666369756062
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,float16,3,0.037503999968369804
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,float16,1,0.03734400123357773
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,fp8,1,0.03165333221356074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,16,8,128,1,float16,float16,255,0.16321600476900736
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,fp8,3,0.03147733211517334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,float16,7,0.037317333122094475
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,float16,15,0.03756800045569738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,fp8,7,0.031311998764673867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,fp8,15,0.03145066648721695
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,fp8,31,0.03123733401298523
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,float16,31,0.03755199909210205
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,fp8,63,0.0314026673634847
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,float16,63,0.03693866729736328
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,float16,127,0.03940266619126002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,float16,1,0.053674668073654175
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,fp8,1,0.045738667249679565
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,1,128,1,float16,fp8,127,0.03324266771475474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,float16,7,0.05394133428732554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,float16,3,0.05465066432952881
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,fp8,3,0.04574400186538696
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,fp8,7,0.046762665112813316
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,float16,15,0.05449066559473673
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,fp8,15,0.04584000011285146
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,float16,31,0.05398400127887726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,fp8,31,0.047226667404174805
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,float16,63,0.053861334919929504
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,fp8,63,0.0459199994802475
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,float16,127,0.059658666451772056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,2,128,1,float16,fp8,127,0.05026666820049286
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,float16,1,0.09703999757766724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,fp8,1,0.08542399605115254
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,float16,3,0.0969599982102712
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,fp8,3,0.08683733145395915
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,float16,7,0.0979360044002533
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,fp8,7,0.08675199747085571
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,float16,15,0.09726400176684062
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,fp8,15,0.08659733335177104
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,float16,31,0.09716266393661499
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,fp8,31,0.08463467160860698
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,float16,63,0.0969599982102712
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,fp8,63,0.08596799770991008
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,float16,127,0.10548800230026245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,4,128,1,float16,fp8,127,0.08684266606966655
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,fp8,1,0.14890666802724203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,float16,3,0.1730239987373352
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,fp8,3,0.1483733355998993
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,float16,1,0.17333332697550455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,fp8,7,0.15016532937685648
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,float16,7,0.17287466923395792
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,float16,15,0.1730560064315796
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,1,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,float16,31,0.17291200160980225
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,1,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,3,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,7,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,fp8,31,0.14884266257286072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,fp8,15,0.14898666739463806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,7,0.009941333283980688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,15,0.009519999846816063
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,3,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,31,0.009232000137368837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,63,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,fp8,63,0.14983466267585754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,127,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,float16,63,0.17335999011993408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,31,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,15,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,255,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,255,0.009258666386206945
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,1023,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,1023,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,2047,0.012879999975363413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,2047,0.012975999464591345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,4095,0.014901333798964819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,4095,0.014938666174809137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,8191,0.01727466657757759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,8191,0.016885332763195038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,16383,0.01916266605257988
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,16383,0.03287466615438461
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,32767,0.05096533397833506
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,32767,0.03573866685231527
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,1,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,1,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,3,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,3,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,7,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,7,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,15,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,float16,511,0.010656000425418219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,511,0.010522666076819101
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,15,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,31,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,31,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,63,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,63,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,127,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,127,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,255,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,255,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,511,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,511,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,1,128,1,float16,fp8,127,0.009850666547815004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,1023,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,1023,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,2047,0.012842666357755661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,4095,0.014794666320085526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,2047,0.011781333635250727
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,4095,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,8191,0.015813333292802174
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,16383,0.048112000028292336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,16383,0.031530665854612984
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,32767,0.08597333232561748
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,fp8,32767,0.05004266897837321
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,1,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,3,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,3,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,7,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,7,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,15,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,15,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,31,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,63,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,63,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,127,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,127,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,255,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,float16,127,0.1856693426767985
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,255,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,511,0.010682666053374609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,511,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,1023,0.010842667271693548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,1023,0.01118933285276095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,2047,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,4095,0.02938666691382726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,4095,0.015125333021084467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,8191,0.031008000175158184
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,8191,0.04773866633574168
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,16383,0.04953599969546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,16383,0.08362666765848796
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,fp8,32767,0.08690667152404785
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,32767,0.1564853290716807
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,1,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,1,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,3,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,7,0.009818666925032934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,1,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,4,128,1,float16,float16,2047,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,16,8,128,1,float16,fp8,127,0.15000533064206442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,15,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,15,0.010453333457310995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,31,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,31,0.009450666606426239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,63,0.010640000303586325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,127,0.009999999776482582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,127,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,255,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,255,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,511,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,511,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,1023,0.012896000097195307
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,1023,0.012746666868527731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,2047,0.03523733218510946
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,2047,0.014837333311637243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,4095,0.04714666803677877
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,4095,0.029696000119050343
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,8191,0.04775999983151754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,8191,0.0830506682395935
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,63,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,16383,0.08427733182907104
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,3,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,16383,0.1553759972254435
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,1,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,3,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,1,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,7,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,float16,32767,0.3001493414243062
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,32767,0.1569706698258718
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,7,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,15,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,15,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,63,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,31,0.009541333342591921
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,127,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,127,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,255,0.00927466650803884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,255,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,511,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,511,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,1023,0.011215999722480774
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,1023,0.010682666053374609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,2047,0.012885333349307379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,2047,0.012709333250919977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,4095,0.01481066644191742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,4095,0.015008000036080679
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,8191,0.030741333961486816
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,8191,0.0169813334941864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,16383,0.03332799921433131
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,16383,0.04922133187452952
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,1,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,1,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,2,128,1,float16,float16,8191,0.02972800036271413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,3,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,3,0.008672000219424566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,7,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,16,8,128,1,float16,fp8,7,0.009450666606426239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,7,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,fp8,3,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,15,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,31,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,31,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,63,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,63,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,127,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,127,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,31,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,255,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,511,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,511,0.01090666651725769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,1023,0.011152000476916632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,1023,0.01110400011142095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,2047,0.012815999488035837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,4095,0.015082667271296183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,8191,0.04817600051561991
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,8191,0.031248000760873158
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,16383,0.08567999800046285
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,fp8,16383,0.04987733562787374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,1,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,1,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,3,0.009866666669646898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,3,0.00943999985853831
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,7,0.00926399976015091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,7,0.009519999846816063
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,15,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,31,0.010181333248813948
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,31,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,15,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,63,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,127,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,127,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,255,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,255,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,1,128,1,float16,float16,63,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,511,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,511,0.010608000059922537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,1023,0.012826666235923767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,4095,0.02935466667016347
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,1023,0.011695999652147293
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,2047,0.029135999580224354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,2047,0.014890667051076889
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,4095,0.047594666481018066
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,4095,0.030031998952229817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,8191,0.08374933401743571
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,8191,0.049327999353408813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,1,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,1,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,16383,0.15562132994333902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,fp8,16383,0.08705600102742513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,3,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,3,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,7,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,15,0.008623999853928884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,7,0.009509333098928133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,15,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,31,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,31,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,63,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,127,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,127,0.00966933307548364
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,255,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,255,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,511,0.010661333799362183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,511,0.01071999967098236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,4,128,1,float16,float16,63,0.009253333633144697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,1023,0.009712000067035357
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,1023,0.010650667051474253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,2047,0.009413333609700203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,4095,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,2047,0.010768000036478043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,4095,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,8191,0.014912000546852747
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,8191,0.015237333873907724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,16383,0.018922666708628338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,32767,0.027056001126766205
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,16383,0.019007999449968338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,32767,0.02718399961789449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,255,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,65535,0.02956266701221466
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,fp8,65535,0.03046400099992752
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,1,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,1,0.008634666601816813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,3,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,7,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,15,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,15,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,31,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,12,2,128,1,float16,float16,2047,0.012869333227475485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,63,0.008650666723648706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,63,0.008581333483258883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,127,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,255,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,127,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,511,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,511,0.010026666646202406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,1023,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,1023,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,2047,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,2047,0.009824000298976898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,4095,0.01020800011853377
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,4095,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,8191,0.013248000293970108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,8191,0.013167999684810638
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,16383,0.018837332725524902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,16383,0.018383999665578205
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,32767,0.021242665747801464
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,32767,0.020975999534130096
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,65535,0.02327466756105423
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,65535,0.021594665944576263
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,1,128,1,float16,float16,63,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,1,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,1,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,3,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,3,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,7,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,7,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,15,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,31,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,63,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,63,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,127,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,3,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,float16,7,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,255,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,255,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,31,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,511,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,511,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,1023,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,1023,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,2047,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,2047,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,2,128,1,float16,fp8,255,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,4095,0.010650667051474253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,4095,0.010661333799362183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,8191,0.014757333944241205
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,16383,0.016890666137139004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,16383,0.017018667111794155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,32767,0.019333332777023315
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,32767,0.01781333362062772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,65535,0.03586133321126302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,65535,0.019167999426523846
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,1,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,3,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,3,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,7,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,7,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,15,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,15,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,31,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,15,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,31,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,63,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,63,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,fp8,31,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,127,0.009813333551088968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,255,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,255,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,511,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,511,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,1023,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,1023,0.009685333197315535
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,127,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,2047,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,2047,0.010005333150426546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,4095,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,4095,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,8191,0.014885333677132925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,8191,0.015013333410024643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,16383,0.01933866615096728
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,16383,0.018992000569899876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,32767,0.02201066662867864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,32767,0.021562665700912476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,65535,0.02399466683467229
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,fp8,65535,0.023018665611743927
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,1,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,1,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,3,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,3,0.008602666358153025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,7,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,7,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,15,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,15,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,31,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,63,0.008672000219424566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,63,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,127,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,127,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,255,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,255,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,511,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,1023,0.009472000102202097
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,2047,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,2047,0.010117333382368088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,4095,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,4095,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,8191,0.014773332824309668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,8191,0.014896000425020853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,16383,0.017024000485738117
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,16383,0.01695999999841054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,32767,0.019007999449968338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,65535,0.036362667878468834
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,127,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,65535,0.020634666085243225
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,1,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,3,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,3,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,7,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,7,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,15,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,15,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,31,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,31,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,63,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,12,4,128,1,float16,float16,8191,0.015024000157912573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,127,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,255,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,511,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,float16,1023,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,255,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,511,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,511,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,1023,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,1023,0.009152000149091085
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,2047,0.009301333377758661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,4095,0.012906666845083237
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,4095,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,2,128,1,float16,fp8,32767,0.01738133281469345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,8191,0.01498666654030482
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,8191,0.01471466695268949
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,16383,0.017125333348910015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,1,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,32767,0.033215999603271484
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,32767,0.017840000490347546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,65535,0.051557332277297974
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,65535,0.0351200004418691
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,1,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,1,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,3,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,63,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,3,0.009279999881982803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,127,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,7,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,7,0.009872000043590864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,1,128,1,float16,float16,1,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,15,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,15,0.009423999736706415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,31,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,31,0.010384000216921171
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,63,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,63,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,127,0.009413333609700203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,127,0.010133333504199982
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,float16,2047,0.010037333394090334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,255,0.009589333087205887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,511,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,511,0.011098666737476984
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,1023,0.012794667234023413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,1023,0.012885333349307379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,12,4,128,1,float16,fp8,16383,0.01523200049996376
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,2047,0.014767999450365702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,2047,0.015173333386580149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,8191,0.04931733508904775
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,8191,0.03212266663710276
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,fp8,4095,0.017093333105246227
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,4095,0.030799999833106995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,1,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,3,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,3,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,7,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,15,0.009418666362762451
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,7,0.009226666763424873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,31,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,31,0.00997866690158844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,63,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,63,0.009813333551088968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,127,0.009898666913310686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,127,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,255,0.0102613332370917
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,255,0.010112000008424124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,511,0.010874666273593903
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,511,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,1023,0.012784000486135483
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,2047,0.029365333418051403
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,2047,0.015029333531856537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,4095,0.04772266745567322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,4095,0.030085332691669464
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,1,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,1,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,8191,0.08472533027331035
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,fp8,8191,0.04886400202910105
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,3,0.010874666273593903
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,3,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,7,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,7,0.011055999745925268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,15,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,15,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,31,0.01090666651725769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,31,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,15,0.009866666669646898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,63,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,127,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,255,0.011050666371981302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,255,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,511,0.013183999806642532
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,511,0.012821332861979803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,1023,0.029167999823888142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,1,128,1,float16,float16,255,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,1023,0.014943999548753103
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,2047,0.047637333472569786
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,4095,0.08477866649627686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,4095,0.049226666490236916
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,1,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,8191,0.15679466724395752
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,1,0.009797333429257074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,1,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,8191,0.08807466427485149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,3,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,3,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,7,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,15,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,7,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,31,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,31,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,63,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,127,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,127,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,255,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,127,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,255,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,511,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,511,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,1023,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,1023,0.011855999628702799
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,2047,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,2,128,1,float16,float16,1023,0.01309866706530253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,2047,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,4095,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,8191,0.015082667271296183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,4095,0.011055999745925268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,8191,0.015072000523408255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,16383,0.018405333161354065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,16383,0.01748266691962878
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,32767,0.01921066641807556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,65535,0.037690666814645134
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,65535,0.02298133323589961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,fp8,2047,0.03108799954255422
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,1,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,1,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,3,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,3,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,7,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,7,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,15,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,31,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,31,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,fp8,15,0.009359999870260557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,63,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,63,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,63,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,127,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,127,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,255,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,12,4,128,1,float16,float16,63,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,255,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,511,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,1023,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,511,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,1023,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,4095,0.0129120002190272
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,4095,0.012917333592971167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,8191,0.01482133318980535
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,16383,0.015109332899252573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,16383,0.01693333312869072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,32767,0.03408533334732056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,32767,0.0191040001809597
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,65535,0.0516533354918162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,65535,0.03457599878311157
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,1,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,1,128,1,float16,float16,32767,0.020981334149837494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,1,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,3,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,3,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,7,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,15,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,15,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,31,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,31,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,63,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,63,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,127,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,15,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,127,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,255,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,255,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,511,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,511,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,1023,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,2047,0.011343999455372492
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,2047,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,4095,0.013162666310866674
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,4095,0.012949333836634954
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,8191,0.016762666404247284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,8191,0.014922666052977243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,fp8,2047,0.01055466632048289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,16383,0.01736533393462499
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,32767,0.049551998575528465
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,8191,0.01479999969402949
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,32767,0.03345066557327906
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,65535,0.08560533324877422
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,65535,0.05156800150871277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,1,0.009839999799927076
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,1,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,3,0.009717333440979322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,3,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,7,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,7,0.009642666826645533
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,7,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,15,0.00996800015370051
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,15,0.010656000425418219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,31,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,31,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,63,0.009642666826645533
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,63,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,127,0.010586666564146677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,fp8,1023,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,127,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,255,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,255,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,511,0.011157333850860596
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,511,0.01090666651725769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,4,128,1,float16,float16,16383,0.031386665999889374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,12,2,128,1,float16,float16,2047,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,1023,0.01292266696691513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,1023,0.012805332740147909
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,2047,0.017157333592573803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,2047,0.02979733298222224
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,1,0.011498666057984034
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,1,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,3,0.011098666737476984
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,float16,4095,0.04821866750717163
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,1,128,1,float16,fp8,4095,0.031370667119820915
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,7,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,3,0.01108266661564509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,7,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,15,0.01108266661564509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,15,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,63,0.010922666639089584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,63,0.01101333275437355
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,127,0.011066666493813196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,127,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,255,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,255,0.011055999745925268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,511,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,511,0.012938667088747025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,31,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,1023,0.02942933390537898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,1023,0.014885333677132925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,4095,0.08582933743794759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,1,0.013167999684810638
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,2047,0.03108799954255422
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,2047,0.04833599925041199
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,1,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,3,0.012831999609867731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,3,0.013141332815090815
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,7,0.013690666606028875
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,15,0.012975999464591345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,31,0.013013333082199097
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,31,0.01349866638580958
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,7,0.012714666624863943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,63,0.01414399966597557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,15,0.012928000340859095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,63,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,float16,31,0.011109333485364914
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,127,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,127,0.013253333667914072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,255,0.012762666990359625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,255,0.012965332716703415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,511,0.02942933390537898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,511,0.016800000021855038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,2047,0.08509332935015361
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,2047,0.05008000135421753
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,1023,0.03152533372243246
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,2,128,1,float16,fp8,4095,0.05048533280690511
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,4095,0.15872533122698465
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,float16,1023,0.04852266609668732
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,12,4,128,1,float16,fp8,4095,0.08774933218955994
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,1,0.011402666568756104
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,1,0.012928000340859095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,3,0.011519999553759893
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,7,0.012602667013804117
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,7,0.011855999628702799
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,15,0.01219733307758967
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,31,0.012026666353146235
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,3,0.012815999488035837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,63,0.011733333269755045
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,31,0.012853333105643591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,15,0.012698666503032049
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,63,0.012879999975363413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,127,0.011957333733638128
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,127,0.01202133297920227
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,255,0.0120319997270902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,255,0.012901333471139273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,511,0.015018666783968607
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,511,0.014757333944241205
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,float16,1023,0.031189332405726116
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,1,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,1,0.013178666432698568
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,1,128,1,float16,fp8,1023,0.01735466718673706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,3,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,3,0.013130666067202887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,7,0.013354666531085968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,7,0.012874666601419449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,15,0.01414399966597557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,15,0.012986666212479273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,31,0.014074667046467463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,63,0.013178666432698568
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,63,0.012901333471139273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,31,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,127,0.014671999961137772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,127,0.012815999488035837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,255,0.01320533330241839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,255,0.012885333349307379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,511,0.030207999050617218
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,1,0.018960000326236088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,511,0.016805333395799
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,1,0.016965333372354507
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,float16,1023,0.04957866668701172
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,3,0.019194666296243668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,2,128,1,float16,fp8,1023,0.031850665807724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,3,0.017221332838137943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,7,0.017029333859682083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,15,0.016970666746298473
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,7,0.019146667172511418
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,31,0.01717866708834966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,63,0.0170666662355264
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,127,0.0189280000825723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,31,0.019018666197856266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,63,0.019130667050679524
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,127,0.017301333447297413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,255,0.03001066545645396
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,255,0.016927999754746754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,511,0.048895999789237976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,511,0.031445334355036415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,15,0.01894933357834816
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,fp8,1023,0.050581331054369606
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,float16,1,0.01522133375207583
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,fp8,1,0.01516266663869222
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,float16,3,0.015008000036080679
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,fp8,3,0.015141333142916361
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,fp8,7,0.014954666296641031
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,fp8,15,0.014746667196353277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,float16,31,0.015589332828919092
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,float16,15,0.014943999548753103
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,fp8,31,0.014917333920796713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,float16,63,0.015173333386580149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,fp8,63,0.014725333700577417
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,float16,127,0.01515199989080429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,fp8,127,0.015002666662136713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,float16,255,0.015189333508412043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,float16,7,0.015317333241303762
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,fp8,255,0.015008000036080679
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,float16,511,0.03145066648721695
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,12,4,128,1,float16,float16,1023,0.08467732866605122
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,float16,1,0.019130667050679524
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,1,128,1,float16,fp8,511,0.018992000569899876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,fp8,1,0.01718933383623759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,float16,3,0.018986667195955913
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,fp8,3,0.017157333592573803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,float16,7,0.019237333287795384
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,fp8,7,0.017162666966517765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,fp8,15,0.017136000096797943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,float16,31,0.018895999838908512
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,fp8,31,0.017221332838137943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,float16,63,0.019007999449968338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,fp8,63,0.017290666699409485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,float16,15,0.019066666563351948
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,float16,127,0.01921066641807556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,fp8,127,0.017029333859682083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,float16,255,0.031023999055226643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,fp8,255,0.017263999829689663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,float16,1,0.02959466725587845
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,float16,511,0.04934399823347727
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,2,128,1,float16,fp8,511,0.03219199925661087
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,fp8,1,0.02743999908367793
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,fp8,3,0.027136000494162243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,float16,7,0.0305173322558403
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,float16,15,0.02951466788848241
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,fp8,7,0.027066667874654133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,fp8,31,0.027280000348885853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,float16,63,0.03012266755104065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,fp8,63,0.02719466636578242
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,fp8,15,0.027434666951497395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,float16,31,0.03018666555484136
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,float16,3,0.02938133229811986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,float16,127,0.03538133452335993
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,fp8,127,0.02701333413521449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,1,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,1,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,7,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,7,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,15,0.00926399976015091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,15,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,3,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,31,0.009173333023985228
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,3,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,31,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,63,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,63,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,127,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,127,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,255,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,255,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,511,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,511,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,fp8,255,0.03497066597143809
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,1023,0.009957333405812582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,1023,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,2047,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,2047,0.010618666807810465
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,4095,0.012901333471139273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,float16,255,0.050426666935284935
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,16383,0.017045332739750545
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,4095,0.013248000293970108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,32767,0.021002667645613354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,65535,0.052373334765434265
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,32767,0.03384533276160558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,65535,0.03586133321126302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,float16,511,0.08643733461697896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,1,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,1,0.009829333052039146
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,3,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,12,4,128,1,float16,fp8,511,0.05355200171470642
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,7,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,7,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,15,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,15,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,31,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,31,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,63,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,63,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,127,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,127,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,255,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,3,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,255,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,511,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,511,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,1023,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,2047,0.011173332730929056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,2047,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,4095,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,4095,0.012789333860079447
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,8191,0.015061333775520325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,16383,0.03143999973932902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,16383,0.0169813334941864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,8191,0.015029333531856537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,32767,0.050250664353370667
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,float16,16383,0.01756799966096878
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,1,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,65535,0.053397332628568016
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,1,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,3,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,3,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,7,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,7,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,15,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,15,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,31,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,31,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,63,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,63,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,127,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,127,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,255,0.008634666601816813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,255,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,511,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,1023,0.009359999870260557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,511,0.010794666906197866
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,1023,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,1023,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,2047,0.012015999605258306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,2047,0.010911999891201654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,4095,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,4095,0.012826666235923767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,8191,0.015178666760524115
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,8191,0.029370665550231934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,16383,0.04751466711362203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,16383,0.03140799949566523
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,32767,0.049738665421803795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,32767,0.08410132924715678
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,fp8,65535,0.08674133817354839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,65535,0.08841066559155782
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,4,128,1,float16,float16,65535,0.15637866655985513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,float16,8191,0.015061333775520325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,2,128,1,float16,fp8,32767,0.03330666571855545
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,float16,1,0.0229120006163915
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,fp8,1,0.01922133316596349
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,float16,7,0.022837333381175995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,float16,3,0.022976001103719074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,fp8,3,0.019226666539907455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,12,1,128,1,float16,fp8,8191,0.015168000012636185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,fp8,7,0.019333332777023315
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,float16,15,0.022970666488011677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,fp8,15,0.018944000204404194
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,float16,31,0.022986667851607006
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,fp8,31,0.019194666296243668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,float16,63,0.022944000860055287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,fp8,63,0.019487999379634857
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,float16,127,0.022997332115968067
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,fp8,255,0.02045866722861926
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,float16,255,0.03166399896144867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,1,128,1,float16,fp8,127,0.01915733392039935
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,fp8,1,0.027450665831565857
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,float16,1,0.029509333272775013
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,float16,3,0.029877332349618275
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,fp8,3,0.027621333797772724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,float16,7,0.03073599934577942
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,float16,31,0.030213333666324615
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,fp8,7,0.02741333345572154
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,float16,15,0.031152000029881794
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,fp8,15,0.02739733209212621
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,float16,63,0.03035733352104823
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,fp8,63,0.027386667827765148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,fp8,31,0.02722666660944621
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,float16,127,0.0355679988861084
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,fp8,127,0.02717866748571396
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,float16,1,0.05394133428732554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,float16,255,0.05190933247407278
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,2,128,1,float16,fp8,255,0.035487999518712364
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,fp8,1,0.04555733501911163
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,float16,3,0.05384000142415365
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,fp8,3,0.04560000201066335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,float16,7,0.0539626677831014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,fp8,7,0.045610666275024414
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,float16,15,0.05425066749254862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,fp8,31,0.045647998650868736
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,fp8,15,0.045941332976023354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,float16,31,0.05420800050099691
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,float16,63,0.053802669048309326
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,fp8,63,0.04557333389918009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,float16,127,0.05770133435726166
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,fp8,127,0.049360002080599465
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,fp8,255,0.05797333518664042
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,12,4,128,1,float16,float16,255,0.09021866321563721
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,float16,1,0.03691199918588003
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,fp8,1,0.03142933299144109
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,float16,3,0.035589332381884255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,fp8,3,0.03147733211517334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,float16,7,0.03736533224582672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,fp8,7,0.03164266546567281
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,fp8,15,0.03129599988460541
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,float16,31,0.03559466699759165
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,float16,15,0.037658666570981346
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,fp8,31,0.031370667119820915
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,float16,63,0.03743999948104223
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,float16,127,0.039247999588648476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,float16,1,0.05397333204746246
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,fp8,63,0.03141333411137263
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,fp8,1,0.04573333263397217
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,1,128,1,float16,fp8,127,0.031680000325044
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,float16,3,0.05392000079154968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,float16,7,0.05425600210825602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,fp8,7,0.04576000074545542
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,float16,15,0.05402133365472158
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,fp8,3,0.04599999884764353
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,float16,31,0.05389333268006643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,fp8,31,0.04615999758243561
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,float16,63,0.05388266841570536
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,fp8,15,0.045781334241231285
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,float16,127,0.05841066439946493
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,fp8,63,0.04607999821503957
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,fp8,1,0.08685866991678874
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,float16,1,0.0965119997660319
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,2,128,1,float16,fp8,127,0.05045333504676819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,float16,3,0.09724799791971843
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,fp8,3,0.08673600355784099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,float16,7,0.09566932916641235
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,float16,15,0.09634133179982503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,fp8,7,0.08668800195058186
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,1,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,1,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,3,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,3,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,fp8,15,0.08502399921417236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,7,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,7,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,15,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,15,0.00916800027092298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,31,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,63,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,fp8,31,0.08674666285514832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,float16,31,0.09683199723561604
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,127,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,127,0.009818666925032934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,255,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,511,0.011098666737476984
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,1023,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,255,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,1023,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,511,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,2047,0.012815999488035837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,2047,0.012810666114091873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,4095,0.014965333044528961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,4095,0.015146666516860327
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,8191,0.01695466662446658
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,8191,0.016282666474580765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,16383,0.03348266581694285
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,16383,0.019493332753578823
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,float16,32767,0.05128000179926554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,1,128,1,float16,fp8,32767,0.03453333427508672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,1,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,1,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,3,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,3,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,7,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,7,0.00927466650803884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,15,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,float16,63,0.0972106655438741
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,15,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,31,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,31,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,fp8,63,0.0851093331972758
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,63,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,127,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,255,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,255,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,511,0.009861333295702934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,1023,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,1023,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,127,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,2047,0.011365332951148352
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,2047,0.011322667201360067
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,float16,127,0.10474133491516113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,511,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,4095,0.01302933320403099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,8191,0.029215998947620392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,8191,0.015087999403476715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,16383,0.0314026673634847
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,32767,0.08498133222262065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,fp8,32767,0.05014933149019877
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,1,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,1,0.009893333539366722
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,3,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,12,4,128,1,float16,fp8,127,0.08672533432642619
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,3,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,16383,0.047509332497914634
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,7,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,15,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,15,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,31,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,63,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,63,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,127,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,63,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,127,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,255,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,255,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,511,0.010853332777818045
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,511,0.011760000139474869
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,31,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,1023,0.010837333897749582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,2047,0.013658666362365087
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,2047,0.012874666601419449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,4095,0.015119999647140503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,4095,0.029306667546431225
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,8191,0.031199999153614044
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,2,128,1,float16,float16,4095,0.01516266663869222
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,8191,0.047584002216657005
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,16383,0.049653331438700356
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,16383,0.08448533217112224
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,32767,0.15613866845766702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,1,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,1,0.009301333377758661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,3,0.008581333483258883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,32767,0.08546666304270427
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,7,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,7,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,15,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,31,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,31,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,63,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,127,0.009173333023985228
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,127,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,255,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,255,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,511,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,fp8,7,0.009984000275532404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,511,0.009839999799927076
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,1023,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,1023,0.01110400011142095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,2047,0.012736000120639801
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,4095,0.012944000462690989
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,4095,0.014671999961137772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,8191,0.02938133229811986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,8191,0.015237333873907724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,16383,0.04862933357556661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,1,0.008672000219424566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,16383,0.03181866556406021
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,32767,0.049685334165891014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,1,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,float16,32767,0.08591467142105103
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,3,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,3,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,7,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,12,4,128,1,float16,float16,1023,0.011136000355084738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,15,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,31,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,31,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,63,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,127,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,127,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,255,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,255,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,511,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,511,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,3,0.009354666496316591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,1023,0.011178666104873022
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,1023,0.010922666639089584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,2047,0.01321600005030632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,2047,0.012949333836634954
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,4095,0.015066667149464289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,4095,0.029834667841593426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,8191,0.04836800197760264
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,16383,0.0496319979429245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,8191,0.031583999594052635
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,16383,0.08582933743794759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,1,128,1,float16,fp8,2047,0.011055999745925268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,32767,0.08646399776140849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,1,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,32767,0.16089066863059998
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,3,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,1,0.01002133327225844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,3,0.010474666953086853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,7,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,15,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,7,0.010341333225369453
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,15,0.010656000425418219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,31,0.009162666896979014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,float16,7,0.009392000113924345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,63,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,2,128,1,float16,fp8,15,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,63,0.009381333366036415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,255,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,127,0.010319999729593595
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,511,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,511,0.011226666470368704
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,1023,0.012906666845083237
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,2047,0.02920000006755193
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,1023,0.013023999830087027
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,4095,0.03033600002527237
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,4095,0.047775998711586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,8191,0.0830506682395935
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,8191,0.04828799764315287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,16383,0.15633599956830344
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,16383,0.08711999654769897
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,32767,0.299727996190389
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,1,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,32767,0.1612320045630137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,float16,31,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,3,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,127,0.009872000043590864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,7,0.01098666712641716
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,7,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,255,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,15,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,15,0.011130666981140772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,31,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,31,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,63,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,63,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,127,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,127,0.011178666104873022
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,255,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,4,128,1,float16,fp8,2047,0.015450666348139444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,255,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,511,0.012954667210578918
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,511,0.013104000439246496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,1023,0.02945599953333537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,1023,0.015146666516860327
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,2047,0.047456001242001854
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,4095,0.04942933221658071
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,8191,0.1576533317565918
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,16383,0.301744004090627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,16383,0.15946132938067117
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,1,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,32767,0.5988639990488688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,1,0.00916800027092298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,32767,0.3088586727778117
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,3,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,3,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,7,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,7,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,15,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,31,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,31,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,63,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,63,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,127,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,127,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,255,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,255,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,2047,0.030832000076770782
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,511,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,511,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,1023,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,1023,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,2047,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,2047,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,4095,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,4095,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,8191,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,8191,0.012954667210578918
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,16383,0.017349333812793095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,16383,0.017184000462293625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,32767,0.02532800038655599
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,32767,0.02521066615978877
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,65535,0.029440000653266907
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,65535,0.029391999046007793
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,131071,0.03139200061559677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,fp8,131071,0.031146667897701263
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,1,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,1,0.00860799973209699
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,1,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,3,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,3,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,7,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,7,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,15,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,15,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,31,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,63,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,63,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,127,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,127,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,255,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,1,128,1,float16,float16,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,255,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,511,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,1023,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,1023,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,2047,0.009648000200589498
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,2047,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,4095,0.010608000059922537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,4095,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,8191,0.013007999708255133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,8191,0.015082667271296183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,16383,0.0186666672428449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,16383,0.019173332800467808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,32767,0.021317332983016968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,32767,0.021061333517233532
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,65535,0.0236160010099411
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,65535,0.023071999351183575
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,131071,0.04322133461634318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,fp8,131071,0.02462399999300639
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,8191,0.0867146650950114
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,1,0.008650666723648706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,1,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,3,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,3,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,7,0.010474666953086853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,15,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,15,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,31,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,fp8,3,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,63,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,63,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,127,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,127,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,255,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,255,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,511,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,511,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,1023,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,1023,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,2047,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,2047,0.009338666374484697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,4095,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,4095,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,2,128,1,float16,float16,511,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,8191,0.015957333147525787
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,8191,0.014762666076421738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,16383,0.017077332983414333
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,16383,0.01704000060757001
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,32767,0.01932266727089882
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,32767,0.019029332945744198
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,65535,0.035546667873859406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,65535,0.02019199977318446
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,fp8,131071,0.03748800108830134
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,131071,0.05392533540725708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,1,0.008645333349704742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,3,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,3,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,7,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,7,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,15,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,15,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,31,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,31,0.008650666723648706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,8,8,128,1,float16,float16,4095,0.08384000261624654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,63,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,63,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,127,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,7,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,255,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,511,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,511,0.011999999483426413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,4,128,1,float16,float16,31,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,1023,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,1023,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,2047,0.011141333729028702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,2047,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,4095,0.012944000462690989
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,4095,0.012863999853531519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,8191,0.01504533365368843
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,16383,0.016970666746298473
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,32767,0.03315199911594391
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,32767,0.01877333347996076
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,65535,0.05160533388455709
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,65535,0.0345920001467069
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,131071,0.08701866865158081
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,1,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,131071,0.05300800005594889
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,3,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,1,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,3,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,7,0.008645333349704742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,7,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,15,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,15,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,31,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,31,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,63,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,63,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,127,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,127,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,255,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,255,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,127,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,511,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,float16,255,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,1023,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,1023,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,2047,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,2047,0.00961599995692571
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,4095,0.010672000547250112
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,4095,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,8191,0.014570667097965876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,8191,0.013007999708255133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,16383,0.0182239996890227
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,16383,0.017397332936525345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,32767,0.02146133283774058
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,16383,0.015210667004187902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,32767,0.020821332931518555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,65535,0.023242667317390442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,65535,0.023344000180562336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,float16,131071,0.039333333571751915
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,131071,0.024362665911515553
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,1,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,1,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,3,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,3,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,7,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,7,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,15,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,15,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,63,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,127,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,127,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,255,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,1,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,511,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,511,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,1023,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,1023,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,2047,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,2047,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,4095,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,4095,0.010629333555698395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,8191,0.014783999572197596
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,1,128,1,float16,fp8,511,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,8191,0.014912000546852747
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,16383,0.01695999999841054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,16383,0.01681600014368693
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,32767,0.019205333044131596
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,32767,0.017050666113694508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,65535,0.03659733384847641
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,65535,0.01972266659140587
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,131071,0.055306668082873024
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,131071,0.03754133234421412
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,1,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,1,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,3,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,3,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,7,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,15,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,15,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,31,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,31,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,63,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,31,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,63,0.0086666668454806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,127,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,127,0.008618666479984919
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,fp8,255,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,255,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,255,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,511,0.010069333637754122
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,511,0.009813333551088968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,1023,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,1023,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,2047,0.009941333283980688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,2047,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,4095,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,4095,0.012778667112191519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,8191,0.01481066644191742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,8191,0.014362666755914688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,16383,0.016805333395799
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,16383,0.015168000012636185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,32767,0.018735999862353008
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,65535,0.0516533354918162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,65535,0.034272000193595886
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,131071,0.053173333406448364
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,131071,0.0872373382250468
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,1,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,1,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,3,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,3,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,8,8,128,1,float16,fp8,8191,0.015173333386580149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,7,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,7,0.009162666896979014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,15,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,15,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,fp8,7,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,31,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,31,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,63,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,63,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,2,128,1,float16,float16,31,0.008586666857202848
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,127,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,127,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,255,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,255,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,511,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,511,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,1023,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,1023,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,2047,0.011253333340088526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,4095,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,8191,0.015040000279744467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,8191,0.014848000059525171
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,16383,0.03156266609827677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,16383,0.017173333714405697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,32767,0.0496373325586319
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,32767,0.03350933392842611
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,32767,0.03286399940649668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,65535,0.050901333491007485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,65535,0.08611733714739482
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,1,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,131071,0.1575093368689219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,1,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,3,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,3,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,7,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,7,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,15,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,15,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,31,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,31,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,63,0.009503999724984169
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,127,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,127,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,255,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,255,0.009290666629870733
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,511,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,511,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,2047,0.011125333607196808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,1023,0.01137599969903628
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,float16,4095,0.01333333303531011
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,1023,0.01091733326514562
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,2047,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,2047,0.012821332861979803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,8191,0.04808000226815542
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,8191,0.031082667410373688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,4095,0.015301333119471868
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,4095,0.030058667063713074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,float16,16383,0.08540800213813782
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,1,128,1,float16,fp8,16383,0.049546668926874794
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,1,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,8,128,1,float16,fp8,131071,0.08938133716583252
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,7,0.009375999992092451
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,15,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,3,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,63,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,15,0.010314666976531347
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,31,0.009519999846816063
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,63,0.009509333098928133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,127,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,8,4,128,1,float16,float16,63,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,127,0.010677333921194077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,255,0.00933333362142245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,255,0.009695999945203463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,511,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,511,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,1023,0.012719999998807907
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,1023,0.013280000537633896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,2047,0.015034666905800501
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,4095,0.04769066472848257
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,2047,0.029557332396507263
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,4095,0.030533333619435627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,1,0.00966933307548364
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,8191,0.04966933528582255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,1,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,8191,0.08509866396586101
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,16383,0.1584213376045227
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,1,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,3,0.01003200002014637
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,3,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,fp8,16383,0.08732266227404277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,3,0.010773333410422007
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,7,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,15,0.010922666639089584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,7,0.011296000331640244
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,15,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,31,0.010938666760921478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,63,0.010938666760921478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,31,0.011136000355084738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,63,0.01089599976936976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,127,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,255,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,255,0.010794666906197866
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,127,0.01109333336353302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,511,0.013738666971524557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,1023,0.029311999678611755
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,511,0.0136266661187013
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,2047,0.04785066843032837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,1023,0.01505600040157636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,2047,0.031173333525657654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,4095,0.08473599950472514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,8191,0.08673600355784099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,4095,0.04905066887537638
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,8191,0.15747732917467752
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,1,0.01302933320403099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,1,0.013130666067202887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,2,128,1,float16,float16,7,0.00943999985853831
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,3,0.012954667210578918
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,float16,16383,0.3027199904123942
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,4,128,1,float16,fp8,16383,0.16446933150291443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,7,0.013162666310866674
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,3,0.01331199953953425
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,7,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,15,0.012981332838535309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,15,0.013973332941532135
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,31,0.013466666142145792
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,63,0.014469332993030548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,31,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,127,0.013210666676362356
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,127,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,63,0.01331199953953425
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,255,0.013552000125249227
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,255,0.013141332815090815
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,511,0.029882666965325672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,1023,0.04782933493455251
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,2047,0.08486933509508769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,2047,0.04993066688378652
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,511,0.017071999609470367
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,4095,0.086517333984375
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,1,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,4095,0.15955733259518942
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,3,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,8191,0.15894400080045065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,7,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,16383,0.30500266949335736
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,7,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,15,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,31,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,3,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,16383,0.6074826717376709
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,float16,8191,0.3092213273048401
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,15,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,63,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,127,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,255,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,127,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,255,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,511,0.009904000287254652
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,1023,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,1023,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,2047,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,2047,0.009573333586255709
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,4095,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,4095,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,8191,0.015135999768972397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,8191,0.015205333630243937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,16383,0.017279999951521557
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,32767,0.019237333287795384
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,32767,0.017237332959969837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,8,8,128,1,float16,fp8,1023,0.031658666829268135
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,65535,0.03461866577466329
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,65535,0.01956266661485036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,131071,0.05388799806435903
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,1,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,1,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,3,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,3,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,7,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,7,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,15,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,15,0.00922133338948091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,1,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,31,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,31,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,63,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,63,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,127,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,31,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,255,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,63,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,255,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,511,0.010661333799362183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,1023,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,2047,0.009717333440979322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,float16,511,0.009338666374484697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,2047,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,4095,0.012671999633312225
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,4095,0.012874666601419449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,8191,0.014858666807413101
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,8191,0.014752000570297241
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,16383,0.01725333308180173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,16383,0.017024000485738117
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,32767,0.03345066557327906
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,32767,0.01858666663368543
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,65535,0.05194133520126343
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,65535,0.03372266640265783
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,1,128,1,float16,fp8,131071,0.03578133384386698
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,131071,0.08890133102734883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,1,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,131071,0.05292266607284546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,1,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,3,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,3,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,7,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,15,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,15,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,31,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,31,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,63,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,127,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,63,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,127,0.009189333145817121
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,float16,511,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,255,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,1023,0.009269333134094873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,255,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,511,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,1023,0.009408000235756239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,1023,0.010506667196750641
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,2047,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,4095,0.012997332960367203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,8191,0.015210667004187902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,8191,0.01522133375207583
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,2,128,1,float16,fp8,16383,0.014906667172908783
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,16383,0.031386665999889374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,16383,0.016976000120242436
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,32767,0.04970666766166687
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,32767,0.03219733387231827
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,65535,0.05138133466243744
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,131071,0.158160001039505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,131071,0.08937066793441772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,7,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,1,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,1,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,3,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,3,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,7,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,7,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,15,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,15,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,127,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,63,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,fp8,511,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,2047,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,63,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,127,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,4095,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,255,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,511,0.009930666536092758
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,1023,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,1023,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,4,128,1,float16,float16,65535,0.08501866459846497
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,2047,0.012383999923865
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,2047,0.01108266661564509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,4095,0.012853333105643591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,4095,0.014384000251690546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,8191,0.029461334149042766
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,8191,0.015210667004187902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,16383,0.03109866629044215
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,32767,0.08400000135103862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,32767,0.04971733192602793
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,65535,0.0870293378829956
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,65535,0.15637866655985513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,127,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,131071,0.15918933351834616
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,131071,0.30102399984995526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,511,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,1,0.009194666519761086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,1,0.009423999736706415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,3,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,3,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,7,0.009285333255926767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,7,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,15,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,15,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,16383,0.047882666190465294
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,31,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,31,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,63,0.010682666053374609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,63,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,127,0.009952000031868616
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,float16,31,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,255,0.01062400018175443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,255,0.010682666053374609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,511,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,511,0.010805333654085795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,1023,0.01301866645614306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,255,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,1023,0.013114667187134424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,2047,0.02938133229811986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,2047,0.01504533365368843
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,4095,0.04795733094215393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,8191,0.04980266590913137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,4095,0.03142400085926056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,1,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,1,0.010928000013033548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,3,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,fp8,127,0.009338666374484697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,7,0.011066666493813196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,8,8,128,1,float16,fp8,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,7,0.010901333143313726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,3,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,15,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,15,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,31,0.011231999844312668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,31,0.01098666712641716
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,63,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,63,0.01091733326514562
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,127,0.011125333607196808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,127,0.01091733326514562
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,255,0.010778666784365972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,255,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,511,0.013232000172138214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,511,0.012837332983811697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,1023,0.015119999647140503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,1023,0.02958400050799052
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,4095,0.08555733164151509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,1,128,1,float16,float16,8191,0.08551999926567078
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,4095,0.05003199974695841
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,2047,0.048512001832326256
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,2047,0.03102933367093404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,float16,8191,0.1588053305943807
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,1,0.013722666849692663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,1,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,3,0.012954667210578918
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,3,0.013269333789745966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,2,128,1,float16,fp8,8191,0.08699199557304382
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,15,0.013178666432698568
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,15,0.013248000293970108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,31,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,31,0.013466666142145792
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,7,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,127,0.014309333016475042
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,63,0.013605333864688873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,63,0.013306666165590286
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,255,0.012949333836634954
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,255,0.013151999562978745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,511,0.030063999195893604
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,511,0.017162666966517765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,2047,0.05064533154169718
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,1023,0.04862933357556661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,1023,0.03136533250411352
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,4095,0.15852266550064087
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,4095,0.08787199854850769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,8191,0.3075466752052307
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,1,0.016837333639462788
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,7,0.01309866706530253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,3,0.019082666685183842
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,127,0.012768000364303589
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,3,0.017157333592573803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,1,0.019280000279347103
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,7,0.018874666343132656
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,7,0.017071999609470367
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,15,0.01904533306757609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,15,0.016869333883126576
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,fp8,8191,0.16191466649373373
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,4,128,1,float16,float16,2047,0.08530132969220479
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,31,0.01923199991385142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,63,0.01907733331123988
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,31,0.017173333714405697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,63,0.016789333273967106
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,127,0.019173332800467808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,127,0.01692266638080279
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,1023,0.050016000866889954
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,1023,0.08436800042788188
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,255,0.01714666684468587
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,2047,0.15587199727694193
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,2047,0.08604799707730611
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,511,0.03127466638882955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,511,0.048197334011395775
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,255,0.030085332691669464
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,4095,0.2985706726710002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,4095,0.1604373355706533
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,fp8,8191,0.30767999092737836
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,1,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,7,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,7,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,15,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,15,0.011050666371981302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,3,0.010768000036478043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,31,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,1,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,3,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,31,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,63,0.011168000598748526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,127,0.011146667102972666
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,255,0.011130666981140772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,255,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,511,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,511,0.012752000242471695
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,1023,0.01552533358335495
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,1023,0.02985599885384242
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,8,8,128,1,float16,float16,8191,0.5913333495457967
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,float16,2047,0.048512001832326256
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,2047,0.03178666780392329
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,1,0.014042666802803675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,1,0.01302933320403099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,63,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,3,0.01479999969402949
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,7,0.012869333227475485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,15,0.014096000542243322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,15,0.012821332861979803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,3,0.013248000293970108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,7,0.012917333592971167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,31,0.014261333892742792
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,31,0.013248000293970108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,63,0.013637332866589228
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,63,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,127,0.013157332936922709
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,127,0.012837332983811697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,255,0.012928000340859095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,255,0.012970666090647379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,511,0.03036266565322876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,511,0.016800000021855038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,1,128,1,float16,fp8,127,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,1023,0.04966400067011515
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,1023,0.031744000812371574
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,1,0.01912533367673556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,1,0.01714666684468587
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,3,0.017237332959969837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,3,0.018933333456516266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,float16,2047,0.08538132905960083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,2,128,1,float16,fp8,2047,0.050160000721613564
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,15,0.01915733392039935
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,7,0.019088000059127808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,15,0.01709866647919019
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,31,0.01709866647919019
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,31,0.01918399954835574
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,7,0.016943999876578648
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,127,0.019178666174411774
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,127,0.017077332983414333
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,255,0.01711999997496605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,255,0.030368000268936157
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,511,0.0310506671667099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,511,0.04756799836953481
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,63,0.0189280000825723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,63,0.016890666137139004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,2047,0.15622933705647787
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,2047,0.08798933029174805
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,float16,1023,0.08463467160860698
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,1,0.028858666618665058
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,4,128,1,float16,fp8,1023,0.050581331054369606
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,3,0.029264000554879505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,3,0.025439999997615814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,7,0.029018667836983997
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,7,0.02535466601451238
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,15,0.025114665428797405
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,31,0.02537599951028824
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,63,0.029215998947620392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,63,0.025392000873883564
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,127,0.032655999064445496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,127,0.02515733242034912
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,1,0.02516266703605652
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,255,0.04821866750717163
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,31,0.02951466788848241
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,255,0.031888000667095184
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,1023,0.08593066533406575
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,15,0.029157333076000214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,1023,0.15606932838757834
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,511,0.08437866965929668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,511,0.04987200101216634
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,fp8,2047,0.15838399529457092
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,8,8,128,1,float16,float16,2047,0.29928000768025714
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,fp8,1,0.012703999876976013
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,float16,1,0.014896000425020853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,fp8,3,0.01312000056107839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,float16,15,0.013882666826248169
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,fp8,15,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,float16,31,0.012928000340859095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,fp8,31,0.01320533330241839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,float16,63,0.01471466695268949
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,float16,7,0.015040000279744467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,fp8,7,0.01301866645614306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,fp8,63,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,float16,3,0.013338666409254074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,float16,127,0.013807999591032663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,fp8,127,0.012736000120639801
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,float16,255,0.014975999792416891
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,fp8,255,0.012757333616415659
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,float16,511,0.029669334491093952
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,fp8,511,0.018122666825850803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,float16,1,0.01886933296918869
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,fp8,1,0.017130666722853977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,float16,1023,0.049813335140546165
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,1,128,1,float16,fp8,1023,0.03179733455181122
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,fp8,3,0.016917333006858826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,float16,7,0.01899733394384384
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,float16,15,0.019205333044131596
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,fp8,7,0.01706133286158244
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,fp8,15,0.016794666647911072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,float16,63,0.019018666197856266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,float16,31,0.018981333822011948
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,fp8,31,0.01711999997496605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,fp8,63,0.01709866647919019
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,float16,3,0.019237333287795384
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,fp8,127,0.01691199963291486
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,float16,127,0.019066666563351948
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,float16,255,0.03009066730737686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,fp8,255,0.016970666746298473
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,float16,511,0.04839999973773956
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,float16,1,0.029733332494894665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,fp8,511,0.03146666785081228
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,fp8,1,0.02703999976317088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,float16,3,0.02942399928967158
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,float16,1023,0.08647466699282329
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,fp8,3,0.027215999861558277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,float16,7,0.029338667790095013
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,fp8,15,0.027034667630990345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,fp8,7,0.027263998985290527
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,float16,31,0.029824001093705494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,float16,15,0.02939733366171519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,float16,63,0.029818666477998097
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,2,128,1,float16,fp8,1023,0.05031466484069824
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,fp8,63,0.02735466758410136
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,float16,127,0.03515200068553289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,fp8,31,0.02752000093460083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,fp8,127,0.027285332481066387
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,float16,255,0.05083199838797251
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,fp8,255,0.03550933301448822
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,fp8,511,0.05317866802215576
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,float16,511,0.08752000331878662
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,float16,1,0.05232533315817515
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,fp8,1,0.043712000052134194
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,float16,3,0.051541333397229515
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,fp8,3,0.043765331308046974
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,float16,15,0.05190399785836538
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,fp8,15,0.043653334180514015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,fp8,1023,0.09019200007120769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,4,128,1,float16,float16,1023,0.1588106652100881
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,float16,7,0.051829333106676735
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,fp8,7,0.04358933369318644
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,float16,31,0.051829333106676735
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,fp8,31,0.04367466767628988
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,float16,63,0.051685333251953125
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,fp8,63,0.04355733096599579
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,float16,255,0.08676266670227051
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,fp8,255,0.054799998799959816
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,float16,1023,0.30197866757710773
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,3,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,fp8,1023,0.16174399852752686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,3,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,7,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,7,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,15,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,15,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,31,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,31,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,63,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,63,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,127,0.008639999975760778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,float16,127,0.05566399792830149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,127,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,fp8,511,0.08842133482297261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,1,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,511,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,511,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,float16,511,0.15898133317629495
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,1023,0.010298666854699453
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,1,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,2047,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,4095,0.012960000584522883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,8,8,128,1,float16,fp8,127,0.04683200021584829
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,4095,0.012815999488035837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,8191,0.01498666654030482
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,8191,0.014725333700577417
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,16383,0.015279999623696009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,32767,0.019098666807015736
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,65535,0.05161066850026449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,255,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,65535,0.03345066557327906
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,131071,0.08905067046483357
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,131071,0.052186667919158936
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,1,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,3,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,16383,0.017194667210181553
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,3,0.00922133338948091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,7,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,7,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,15,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,15,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,31,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,31,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,63,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,255,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,63,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,127,0.008661333471536636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,127,0.00916800027092298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,255,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,255,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,511,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,511,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,1023,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,1023,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,2047,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,2047,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,4095,0.01302933320403099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,1023,0.009205333267649015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,fp8,2047,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,4095,0.01293333371480306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,1,128,1,float16,float16,32767,0.033146666983763375
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,8191,0.014890667051076889
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,8191,0.01505600040157636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,16383,0.03183999905983607
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,16383,0.01706133286158244
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,32767,0.0498933345079422
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,32767,0.03316800047953924
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,65535,0.051445335149765015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,65535,0.08673066894213359
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,1,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,1,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,fp8,131071,0.09113066395123799
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,131071,0.16025599837303162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,3,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,3,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,2,128,1,float16,float16,1,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,15,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,15,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,63,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,63,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,127,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,127,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,255,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,255,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,511,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,511,0.01033599985142549
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,1023,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,1023,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,2047,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,2047,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,4095,0.013440000514189402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,4095,0.012661332885424295
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,8191,0.029189333319664
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,8191,0.015135999768972397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,16383,0.04778666794300079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,16383,0.031023999055226643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,32767,0.08412800232569377
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,32767,0.048895999789237976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,65535,0.15634666879971823
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,65535,0.08939733107884724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,7,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,1,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,1,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,131071,0.29997867345809937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,131071,0.16337066888809204
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,float16,31,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,3,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,3,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,7,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,7,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,15,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,15,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,31,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,63,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,63,0.008623999853928884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,127,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,255,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,255,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,511,0.010778666784365972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,511,0.011114666859308878
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,1023,0.010949333508809408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,2047,0.013082666943470636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,2047,0.012741333494583765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,4095,0.029311999678611755
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,4095,0.015168000012636185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,8191,0.04757866760094961
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,8191,0.03014400104681651
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,16383,0.084389328956604
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,16383,0.04886933167775472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,32767,0.08474133412043254
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,32767,0.15587199727694193
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,65535,0.15651733676592508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,65535,0.2996799945831299
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,7,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,131071,0.30239999294281006
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,4,128,1,float16,fp8,31,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,131071,0.5890453259150187
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,fp8,127,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,8,8,128,1,float16,float16,1023,0.011130666981140772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,float16,1,0.019039999693632126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,fp8,1,0.01708799973130226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,float16,3,0.01912533367673556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,fp8,7,0.01716800034046173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,fp8,3,0.017103999853134155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,float16,7,0.019215999792019527
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,float16,15,0.018917333334684372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,fp8,15,0.01720000058412552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,float16,31,0.018965333700180054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,float16,63,0.019258666783571243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,fp8,31,0.01691199963291486
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,fp8,63,0.017152000218629837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,float16,127,0.019567999988794327
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,fp8,127,0.01711999997496605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,float16,255,0.03036266565322876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,fp8,1,0.027109332382678986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,fp8,255,0.017269333203633625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,float16,1,0.029653333127498627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,float16,511,0.049584001302719116
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,1,128,1,float16,fp8,511,0.031530665854612984
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,float16,3,0.030938667555650074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,float16,7,0.030495998760064442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,fp8,3,0.027488000690937042
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,fp8,7,0.027269333600997925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,fp8,15,0.027552001178264618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,float16,15,0.02943466603755951
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,float16,63,0.029338667790095013
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,float16,31,0.030074665943781536
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,fp8,63,0.027104000250498455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,fp8,31,0.027221334477265675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,float16,127,0.03570133447647095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,fp8,127,0.027317332724730175
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,float16,1,0.05403733253479004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,fp8,1,0.0455626646677653
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,fp8,255,0.034602666894594826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,float16,255,0.0517546683549881
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,float16,511,0.08909866213798523
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,2,128,1,float16,fp8,511,0.05400000015894572
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,float16,3,0.054117331902186074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,fp8,3,0.045824001232783
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,float16,7,0.053904001911481224
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,fp8,7,0.04593066871166229
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,float16,15,0.05395199855168661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,float16,63,0.054602667689323425
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,fp8,63,0.045594667394955955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,fp8,15,0.04567466676235199
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,float16,31,0.0539626677831014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,float16,127,0.05808533231417338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,fp8,31,0.04593066871166229
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,fp8,127,0.049546668926874794
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,float16,255,0.09030933181444804
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,fp8,255,0.05788266658782959
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,float16,1,0.09152533610661824
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,float16,3,0.09321066737174988
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,fp8,1,0.08051200211048126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,float16,511,0.16239466269810995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,4,128,1,float16,fp8,511,0.09359467029571533
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,fp8,3,0.08035199840863545
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,float16,15,0.09284266829490662
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,fp8,15,0.0803466687599818
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,float16,31,0.09288000067075093
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,fp8,63,0.08086933195590973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,float16,7,0.09261332949002583
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,fp8,7,0.0806826651096344
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,float16,127,0.09726400176684062
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,fp8,127,0.0817493349313736
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,fp8,31,0.08029866715272267
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,float16,63,0.09311466415723164
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,float16,255,0.1622666617234548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,fp8,255,0.09825600186983745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,float16,1,0.029882666965325672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,float16,511,0.30554133653640747
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,float16,3,0.030234667162100475
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,8,8,128,1,float16,fp8,511,0.1644533375898997
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,fp8,3,0.02731200059254964
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,fp8,1,0.02741866558790207
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,float16,7,0.030533333619435627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,fp8,7,0.027503999571005504
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,float16,15,0.029466666281223297
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,float16,31,0.029946667452653248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,float16,63,0.029365333418051403
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,fp8,63,0.027248000105222065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,float16,127,0.035504000882307686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,fp8,15,0.027263998985290527
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,fp8,127,0.027434666951497395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,float16,255,0.05162133276462555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,fp8,31,0.027456000447273254
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,1,128,1,float16,fp8,255,0.036933332681655884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,fp8,1,0.04550399879614512
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,float16,1,0.05382933219273885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,float16,3,0.05415999889373779
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,fp8,3,0.04560533165931702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,float16,7,0.05403733253479004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,fp8,7,0.04557866851488749
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,float16,15,0.05379199981689453
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,fp8,15,0.04552533229192098
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,float16,31,0.05390933156013489
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,fp8,31,0.04577599962552389
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,float16,63,0.053957333167394005
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,fp8,63,0.045552000403404236
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,float16,127,0.05916800101598104
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,fp8,127,0.049509331583976746
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,float16,1,0.09734400113423665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,float16,255,0.09101333220799764
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,2,128,1,float16,fp8,255,0.058677335580190025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,fp8,1,0.08529067039489746
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,float16,3,0.09707732995351155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,fp8,3,0.08577066659927368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,float16,7,0.09730666875839233
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,float16,31,0.09702933828035991
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,fp8,15,0.08667199810345967
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,fp8,31,0.08461866776148479
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,fp8,7,0.08574933807055156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,float16,15,0.09686400492986043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,float16,63,0.09704533219337463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,fp8,63,0.08509866396586101
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,float16,1,0.17280532916386923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,fp8,1,0.15035733580589294
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,fp8,127,0.08742400010426839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,float16,255,0.16818133989969888
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,fp8,255,0.10449600219726562
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,4,128,1,float16,float16,127,0.10417067011197408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,fp8,3,0.14856533209482828
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,float16,3,0.17256534099578857
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,fp8,7,0.14841600259145102
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,float16,7,0.1727466583251953
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,float16,15,0.17312000195185342
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,fp8,31,0.14899200201034546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,fp8,63,0.14842133720715842
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,float16,31,0.1726400057474772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,float16,63,0.17282666762669882
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,1,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,3,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,7,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,7,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,15,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,1,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,float16,127,0.18370133638381958
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,15,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,31,0.0086666668454806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,fp8,127,0.1502400040626526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,31,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,63,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,127,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,255,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,255,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,511,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,511,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,1023,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,1023,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,2047,0.011749333391586939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,fp8,15,0.1488800048828125
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,2047,0.01108266661564509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,4095,0.013866666704416275
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,4095,0.012837332983811697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,8191,0.014688000082969666
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,16383,0.03146133323510488
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,16383,0.017082666357358296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,3,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,32767,0.03344533344109853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,8191,0.015552000453074774
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,65535,0.08669333656628926
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,1,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,65535,0.05187733471393585
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,1,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,3,0.009162666896979014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,3,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,7,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,7,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,15,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,15,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,31,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,31,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,63,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,63,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,127,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,127,0.008639999975760778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,255,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,255,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,511,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,511,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,1023,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,1023,0.010672000547250112
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,2047,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,2047,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,4095,0.013536000003417334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,63,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,4095,0.013189333180586496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,fp8,127,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,8191,0.014917333920796713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,16383,0.04891733328501383
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,1,128,1,float16,float16,32767,0.049685334165891014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,16383,0.03141333411137263
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,32767,0.05028266708056132
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,32767,0.08612799644470215
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,1,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,1,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,65535,0.1599999964237213
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,3,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,fp8,65535,0.08981333176294963
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,3,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,7,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,7,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,15,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,31,0.009306666751702627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,31,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,15,0.009253333633144697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,63,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,127,0.00916800027092298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,255,0.009392000113924345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,255,0.010037333394090334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,511,0.010672000547250112
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,511,0.011264000087976456
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,1023,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,1023,0.011434666812419891
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,2047,0.01312000056107839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,4095,0.02921066681543986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,2,128,1,float16,float16,8191,0.02939733366171519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,2047,0.013386666774749756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,float16,255,0.31218665838241577
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,4095,0.015168000012636185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,8191,0.04776533444722494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,8191,0.03188266605138779
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,16383,0.08295999964078267
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,32767,0.15611199537913004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,32767,0.08784000078837077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,63,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,65535,0.3001706600189209
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,1,0.009914666414260864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,1,0.00949866697192192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,65535,0.16032532850901285
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,3,0.009317333499590555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,3,0.009141333401203156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,7,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,7,0.009370666618148485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,15,0.010293333480755487
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,31,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,63,0.010368000095089277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,127,0.009632000078757605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,63,0.009455999980370203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,127,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,255,0.01062400018175443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,8,8,128,1,float16,fp8,255,0.1842613418896993
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,511,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,511,0.011114666859308878
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,1023,0.012826666235923767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,1023,0.011498666057984034
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,2047,0.014831999937693277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,2047,0.02934933453798294
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,8191,0.08368000388145447
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,8191,0.0479360024134318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,float16,127,0.010389333590865135
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,4,128,1,float16,fp8,16383,0.04952533543109894
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,16383,0.15493866801261902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,16383,0.08515200018882751
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,15,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,32767,0.2993759910265605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,1,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,32767,0.156549334526062
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,3,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,255,0.009408000235756239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,65535,0.30084266265233356
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,3,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,7,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,7,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,15,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,31,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,15,0.009306666751702627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,31,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,63,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,127,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,fp8,4095,0.030042665700117748
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,255,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,127,0.009898666913310686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,255,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,511,0.009685333197315535
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,511,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,1023,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,1023,0.011157333850860596
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,2047,0.012682666381200155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,2047,0.01098666712641716
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,4095,0.014885333677132925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,4095,0.012997332960367203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,8191,0.03018666555484136
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,8191,0.015765332927306492
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,16383,0.04855466882387797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,32767,0.050474668542544045
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,32767,0.08705600102742513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,16383,0.03197333216667175
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,1,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,1,0.009530666594703993
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,4095,0.04794666667779287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,3,0.009381333366036415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,3,0.009317333499590555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,7,0.009397333487868309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,15,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,15,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,31,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,31,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,63,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,63,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,127,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,127,0.009258666386206945
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,255,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,255,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,511,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,511,0.010703999549150467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,1023,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,1023,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,2047,0.012938667088747025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,1,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,2047,0.012805332740147909
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,8,8,128,1,float16,float16,65535,0.5863039890925089
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,4095,0.015130666395028433
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,8191,0.04860266546408335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,8191,0.03108799954255422
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,16383,0.08517332871754964
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,16383,0.0499946673711141
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,32767,0.08704533179601033
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,65535,0.3066026568412781
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,1,0.009754666437705358
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,65535,0.16084800163904825
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,1,0.009535999968647957
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,3,0.009050666665037474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,3,0.010005333150426546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,7,0.009413333609700203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,7,0.010224000240365664
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,15,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,4095,0.029648000995318096
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,15,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,31,0.010058666889866194
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,31,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,63,0.00956266683836778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,127,0.01003200002014637
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,127,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,255,0.009242666885256767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,float16,32767,0.15838399529457092
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,float16,65535,0.1595253348350525
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,511,0.01089599976936976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,1023,0.012815999488035837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,1023,0.012703999876976013
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,2047,0.02932800104220708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,2047,0.014981333166360855
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,4095,0.029663999875386555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,4095,0.047498668233553566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,8191,0.08309866487979889
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,8191,0.04876266419887543
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,16383,0.15478932857513428
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,16383,0.08574933807055156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,32767,0.3006666700045268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,63,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,1,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,1,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,255,0.009269333134094873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,65535,0.5890346765518188
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,1,128,1,float16,fp8,65535,0.08834667007128398
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,7,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,3,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,65535,0.3031253417332967
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,15,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,7,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,31,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,63,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,63,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,127,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,127,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,255,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,255,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,511,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,511,0.01089599976936976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,1023,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,1023,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,2047,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,2047,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,4095,0.010053333515922228
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,4095,0.010682666053374609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,8191,0.012794667234023413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,8191,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,16383,0.016906666258970898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,16383,0.017093333105246227
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,fp8,32767,0.159578671058019
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,32767,0.025279998779296875
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,32767,0.025370667378107708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,65535,0.02903999884923299
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,131071,0.03141866624355316
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,65535,0.02922133356332779
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,131071,0.029359998802344005
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,1,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,1,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,2,128,1,float16,fp8,7,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,3,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,3,0.00860799973209699
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,3,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,7,0.008634666601816813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,7,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,15,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,15,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,31,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,31,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,63,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,63,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,127,0.009850666547815004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,fp8,15,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,127,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,255,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,1,128,1,float16,float16,31,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,511,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,255,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,1023,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,2047,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,2047,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,4095,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,4095,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,8191,0.014896000425020853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,8191,0.013936000565687815
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,16383,0.01911466692884763
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,16383,0.020495999604463577
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,32767,0.021333334346612293
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,32767,0.02088533341884613
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,65535,0.022154666483402252
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,131071,0.04321600000063578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,131071,0.023669332265853882
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,1,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,1,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,3,0.008469333251317343
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,3,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,7,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,7,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,15,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,15,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,31,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,31,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,63,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,63,0.008623999853928884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,127,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,127,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,255,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,255,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,511,0.009493333597977957
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,511,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,1023,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,1023,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,4,4,128,1,float16,float16,511,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,2047,0.010309333602587381
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,4095,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,4095,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,8191,0.014741333822409311
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,8191,0.015008000036080679
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,16383,0.017162666966517765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,16383,0.016800000021855038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,32767,0.019130667050679524
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,fp8,511,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,1023,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,65535,0.03690666705369949
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,65535,0.019199999670187633
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,131071,0.05365866422653198
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,1,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,131071,0.03730666637420654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,1,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,3,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,7,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,3,0.008618666479984919
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,7,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,15,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,31,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,31,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,2,128,1,float16,float16,65535,0.023525332411130268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,63,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,127,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,127,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,255,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,255,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,511,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,511,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,1023,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,1023,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,2047,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,2047,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,4095,0.009658666948477427
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,4095,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,8191,0.013893333574136099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,16383,0.0189280000825723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,16383,0.01889066646496455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,32767,0.021344001094500225
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,32767,0.02015999952952067
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,65535,0.023530667026837666
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,65535,0.02276800076166789
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,131071,0.041722665230433144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,131071,0.024986666937669117
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,1,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,1,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,3,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,3,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,7,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,7,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,15,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,float16,2047,0.00960533320903778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,15,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,31,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,31,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,63,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,63,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,127,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,127,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,255,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,255,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,4,4,128,1,float16,fp8,32767,0.01838933303952217
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,511,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,511,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,1023,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,2047,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,2047,0.009829333052039146
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,4095,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,4095,0.010794666906197866
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,8191,0.014618666221698126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,15,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,16383,0.016976000120242436
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,16383,0.01655999943614006
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,fp8,63,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,32767,0.0189280000825723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,32767,0.019205333044131596
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,65535,0.03519999980926514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,65535,0.01960533360640208
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,131071,0.05532800157864889
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,1,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,131071,0.03754133234421412
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,1,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,3,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,1,128,1,float16,float16,8191,0.013264000415802002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,7,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,7,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,15,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,15,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,31,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,31,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,63,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,127,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,127,0.009205333267649015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,255,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,255,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,511,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,511,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,1023,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,1023,0.00956266683836778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,2047,0.00922133338948091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,2047,0.010357333347201347
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,4095,0.012773333738247553
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,4095,0.012842666357755661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,8191,0.015125333021084467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,8191,0.01479999969402949
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,16383,0.016949333250522614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,16383,0.015072000523408255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,fp8,1023,0.009663999701539675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,32767,0.03349333256483078
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,32767,0.018976000448067982
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,2,128,1,float16,float16,8191,0.015072000523408255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,65535,0.03393599887688955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,131071,0.05320000151793162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,1,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,1,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,3,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,3,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,7,0.008661333471536636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,7,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,fp8,3,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,15,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,15,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,31,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,31,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,63,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,63,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,63,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,127,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,127,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,255,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,511,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,511,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,1023,0.011109333485364914
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,1023,0.010911999891201654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,2047,0.01313599944114685
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,4095,0.029098667204380035
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,4095,0.015013333410024643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,8191,0.048026666045188904
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,8191,0.031109333038330078
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,65535,0.05128000179926554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,4,4,128,1,float16,float16,131071,0.08663466572761536
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,16383,0.050000001986821495
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,32767,0.08701866865158081
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,32767,0.16029866536458334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,16383,0.08568533261617024
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,1,0.009194666519761086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,1,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,3,0.009973333527644476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,3,0.009285333255926767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,7,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,15,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,31,0.009493333597977957
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,31,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,63,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,63,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,float16,255,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,127,0.010288000106811523
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,127,0.010581333190202713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,255,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,255,0.01055466632048289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,511,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,511,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,1,128,1,float16,fp8,2047,0.012741333494583765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,1023,0.01309866706530253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,1023,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,2047,0.02924799919128418
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,2047,0.014896000425020853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,4095,0.04771733283996582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,4095,0.02992533395687739
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,8191,0.08481066425641377
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,8191,0.04921066761016846
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,16383,0.08892800410588582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,16383,0.1592586636543274
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,1,0.011055999745925268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,7,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,15,0.009695999945203463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,fp8,32767,0.1602079967657725
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,3,0.010629333555698395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,2,128,1,float16,float16,32767,0.3054879903793335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,1,0.011125333607196808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,7,0.011125333607196808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,15,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,15,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,31,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,63,0.010922666639089584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,63,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,127,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,31,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,255,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,255,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,511,0.012981332838535309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,511,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,1023,0.02942933390537898
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,2047,0.047770669062932335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,1023,0.015018666783968607
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,2047,0.031184000273545582
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,4095,0.08461333314577739
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,8191,0.08684800068537395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,8191,0.15652267138163248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,16383,0.3028320074081421
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,16383,0.16174399852752686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,7,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,3,0.011050666371981302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,1,0.009205333267649015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,3,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,3,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,7,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,15,0.008650666723648706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,127,0.010853332777818045
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,float16,32767,0.5938133398691813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,32767,0.31382399797439575
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,31,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,127,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,63,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,127,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,63,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,255,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,255,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,511,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,511,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,1023,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,1023,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,2047,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,2047,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,4095,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,4,4,128,1,float16,fp8,4095,0.04942933221658071
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,4095,0.010842667271693548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,8191,0.014943999548753103
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,8191,0.015087999403476715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,16383,0.01718933383623759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,16383,0.016821333517630894
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,32767,0.018954666952292126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,32767,0.017722666263580322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,65535,0.0356480007370313
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,65535,0.01966933285196622
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,131071,0.05462400118509928
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,131071,0.03798400113979975
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,1,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,3,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,3,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,7,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,15,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,15,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,31,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,31,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,7,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,63,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,63,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,15,0.008597333605090777
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,float16,31,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,127,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,255,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,255,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,511,0.009690666571259499
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,511,0.00979200005531311
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,1023,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,1023,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,2047,0.009743999689817429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,2047,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,4095,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,4095,0.01302933320403099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,8191,0.015077333897352219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,8191,0.01392000044385592
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,16383,0.0170666662355264
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,16383,0.01515199989080429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,32767,0.03344533344109853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,32767,0.01815466706951459
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,65535,0.05161066850026449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,65535,0.03453866640726725
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,float16,131071,0.08878933389981587
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,131071,0.052576000491778054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,1,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,1,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,3,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,7,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,1,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,15,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,1,128,1,float16,fp8,1,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,31,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,63,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,63,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,127,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,127,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,127,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,255,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,255,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,511,0.009599999835093817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,511,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,1023,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,1023,0.009930666536092758
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,2047,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,2047,0.01137599969903628
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,4095,0.013082666943470636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,4095,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,8191,0.01623999948302905
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,8191,0.014954666296641031
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,16383,0.031328000128269196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,16383,0.01703466723362605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,32767,0.05018133421738943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,32767,0.033344000577926636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,65535,0.08598933617273967
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,3,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,65535,0.051685333251953125
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,7,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,float16,131071,0.15743999679883322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,131071,0.089519997437795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,1,0.010757333288590113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,1,0.010853332777818045
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,4,128,1,float16,fp8,15,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,3,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,4,2,128,1,float16,fp8,7,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,3,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,7,0.009152000149091085
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,7,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,15,0.010698666175206503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,15,0.010768000036478043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,31,0.010474666953086853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,31,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,63,0.010645333677530289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,63,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,127,0.009749333063761393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,127,0.010821333775917688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,255,0.01062400018175443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,255,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,511,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,511,0.011061333119869232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,1023,0.012037333101034164
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,2047,0.02957333376010259
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,2047,0.015061333775520325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,8191,0.08602133393287659
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,8191,0.0499839981396993
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,4095,0.04789333542188009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,16383,0.15870933731396994
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,4095,0.03054400036732356
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,1,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,1,0.01116266722480456
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,3,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,7,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,fp8,16383,0.08700266480445862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,15,0.011258666714032492
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,15,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,31,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,31,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,3,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,7,0.01110400011142095
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,63,0.011077333241701126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,63,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,127,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,127,0.011087999989589056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,1,128,1,float16,float16,1023,0.01267733300725619
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,255,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,255,0.011141333729028702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,511,0.014495999862750372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,511,0.013034666577974955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,1023,0.029130667448043823
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,1023,0.015008000036080679
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,4095,0.04957333207130432
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,4095,0.08608532945315044
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,2047,0.031136001149813335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,8191,0.16065067052841187
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,8191,0.08804800113042195
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,1,0.014111999422311783
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,2047,0.0476800004641215
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,7,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,3,0.013130666067202887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,15,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,15,0.012778667112191519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,float16,16383,0.3062880039215088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,2,128,1,float16,fp8,16383,0.1625653306643168
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,3,0.013712000101804733
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,7,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,31,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,31,0.013141332815090815
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,127,0.014501333236694336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,127,0.013056000073750814
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,63,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,1,0.01302933320403099
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,255,0.012991999586423239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,255,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,511,0.030533333619435627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,511,0.017050666113694508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,2047,0.08535466591517131
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,1023,0.04855999847253164
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,2047,0.04994666576385498
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,1023,0.03150933235883713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,4095,0.08849599957466125
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,4095,0.15846400459607443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,8191,0.16181866327921549
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,8191,0.30773866176605225
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,63,0.012746666868527731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,3,0.010944000134865442
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,1,0.010981333752473196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,1,0.011173332730929056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,fp8,16383,0.3085493246714274
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,3,0.011120000233252844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,4,4,128,1,float16,float16,16383,0.6065119902292887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,15,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,7,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,15,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,31,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,7,0.011434666812419891
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,63,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,127,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,127,0.010650667051474253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,255,0.011450666934251785
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,31,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,255,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,511,0.014639999717473984
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,511,0.012719999998807907
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,63,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,1023,0.029711998999118805
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,1023,0.015087999403476715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,1,0.013077333569526672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,1,0.012981332838535309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,2047,0.049738665421803795
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,2047,0.03139200061559677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,3,0.014522666732470194
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,3,0.012757333616415659
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,7,0.013744000345468521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,7,0.01292266696691513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,float16,4095,0.08565333485603333
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,1,128,1,float16,fp8,4095,0.049829334020614624
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,31,0.01292266696691513
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,63,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,31,0.014671999961137772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,15,0.012970666090647379
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,63,0.01310933381319046
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,255,0.014394666999578476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,255,0.01310933381319046
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,511,0.029482667644818623
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,511,0.01691199963291486
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,1023,0.048783997694651283
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,127,0.013157332936922709
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,1023,0.03196266790231069
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,127,0.013301332791646322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,4095,0.1599573294321696
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,1,0.018976000448067982
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,4095,0.08790399630864461
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,2047,0.08653333783149719
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,fp8,2047,0.05045333504676819
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,1,0.017071999609470367
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,3,0.01718933383623759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,3,0.019189332922299702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,7,0.018944000204404194
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,7,0.01687466725707054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,31,0.019050666441520054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,2,128,1,float16,float16,15,0.01498666654030482
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,31,0.017029333859682083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,63,0.018960000326236088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,63,0.01711999997496605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,127,0.018800000349680584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,127,0.016895999511082966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,255,0.01701333373785019
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,255,0.029893333713213604
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,15,0.01684800038735072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,15,0.01912533367673556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,511,0.04786133269468943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,511,0.031311998764673867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,2047,0.08710933725039165
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,1023,0.04974400003751119
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,fp8,4095,0.16209066907564798
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,4095,0.29869333902994794
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,1,0.013749333719412485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,1,0.012869333227475485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,3,0.013477332890033722
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,3,0.012858666479587555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,7,0.013157332936922709
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,2047,0.15596266587575278
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,15,0.013989333063364029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,7,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,31,0.014757333944241205
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,15,0.012789333860079447
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,31,0.013072000195582708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,63,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,63,0.013130666067202887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,127,0.014501333236694336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,255,0.014826666563749313
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,255,0.013125333935022354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,511,0.030106666187445324
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,4,4,128,1,float16,float16,1023,0.08442667126655579
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,511,0.016949333250522614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,1,0.018981333822011948
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,1023,0.048783997694651283
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,1023,0.03147733211517334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,127,0.012778667112191519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,float16,2047,0.08715200424194336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,1,128,1,float16,fp8,2047,0.04995200037956238
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,3,0.018992000569899876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,3,0.01711999997496605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,7,0.018922666708628338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,15,0.019007999449968338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,7,0.017130666722853977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,31,0.01710933322707812
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,63,0.018842666099468868
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,1,0.01724799970785777
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,15,0.01714133347074191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,31,0.018986667195955913
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,127,0.019002666076024372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,127,0.0170666662355264
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,255,0.02939733366171519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,255,0.01687466725707054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,511,0.04760533571243286
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,511,0.031328000128269196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,63,0.01703466723362605
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,1,0.029552000264326733
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,1023,0.050255998969078064
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,1023,0.08489066362380981
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,1,0.027141332626342773
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,3,0.02733866622050603
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,3,0.029535998900731403
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,7,0.029525332152843475
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,float16,2047,0.15901333093643188
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,2,128,1,float16,fp8,2047,0.08833600083986919
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,7,0.02714666724205017
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,15,0.029520000020662945
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,15,0.027210667729377747
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,31,0.02752000093460083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,31,0.02972800036271413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,63,0.030192000170548756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,127,0.0272533322374026
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,255,0.03324799984693527
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,255,0.05017066498597463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,511,0.08633599678675334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,511,0.05216533442338308
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,2047,0.3022560079892476
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,3,0.008703999842206636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,1023,0.1581439971923828
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,2047,0.16432533661524454
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,3,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,1023,0.0888320008913676
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,15,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,31,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,7,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,63,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,1,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,63,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,127,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,127,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,fp8,63,0.027488000690937042
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,255,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,1,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,511,0.009402666861812273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,1023,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,1023,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,2047,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,511,0.009503999724984169
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,2047,0.010741333166758219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,4095,0.012800000607967377
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,8191,0.015125333021084467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,8191,0.014602666099866232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,4095,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,16383,0.0169813334941864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,16383,0.015066667149464289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,32767,0.033439998825391136
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,32767,0.01916266605257988
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,65535,0.033488000432650246
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,131071,0.08850133419036865
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,1,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,131071,0.05386666456858317
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,1,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,3,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,4,4,128,1,float16,float16,127,0.0342399999499321
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,3,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,7,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,15,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,31,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,7,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,7,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,15,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,31,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,63,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,63,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,127,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,127,0.008858666444818178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,255,0.008869333192706108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,255,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,fp8,255,0.008645333349704742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,511,0.00997866690158844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,1023,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,1023,0.010762666662534079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,2047,0.01099733387430509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,4095,0.013370666652917862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,4095,0.012847999731699625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,8191,0.014917333920796713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,8191,0.015141333142916361
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,16383,0.03175999969244003
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,16383,0.01708799973130226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,32767,0.05018133421738943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,32767,0.033189333975315094
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,65535,0.08734933535257976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,65535,0.05115200082461039
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,1,128,1,float16,float16,65535,0.05216533442338308
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,131071,0.16049066185951233
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,131071,0.08947199583053589
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,1,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,3,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,3,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,7,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,7,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,15,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,15,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,31,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,31,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,63,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,15,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,127,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,127,0.008645333349704742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,255,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,255,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,511,0.010319999729593595
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,511,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,1023,0.01090666651725769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,1023,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,2047,0.010869332899649939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,2047,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,4095,0.013834666460752487
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,float16,511,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,8191,0.015130666395028433
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,8191,0.029167999823888142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,2,128,1,float16,fp8,2047,0.011168000598748526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,16383,0.031221332649389904
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,16383,0.04753066599369049
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,32767,0.049226666490236916
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,65535,0.08783466617266338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,65535,0.15687466661135355
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,131071,0.3014293313026428
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,131071,0.1609173317750295
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,1,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,63,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,fp8,4095,0.012917333592971167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,4,4,128,1,float16,float16,32767,0.08362133304278056
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,float16,1,0.019061333189407986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,fp8,1,0.017008000363906223
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,float16,3,0.019167999426523846
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,fp8,3,0.017008000363906223
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,float16,7,0.01905599981546402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,fp8,7,0.01717866708834966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,float16,15,0.01913600042462349
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,fp8,15,0.017210666090250015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,float16,31,0.019007999449968338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,float16,63,0.020175999651352566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,fp8,63,0.017130666722853977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,fp8,31,0.016997333616018295
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,float16,127,0.019130667050679524
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,fp8,127,0.017157333592573803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,float16,255,0.031082667410373688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,fp8,255,0.017231999586025875
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,float16,511,0.048341333866119385
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,fp8,511,0.031712000568707786
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,float16,1,0.030394665896892548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,fp8,1,0.0271573339899381
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,float16,1023,0.0867680013179779
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,1,128,1,float16,fp8,1023,0.050997331738471985
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,fp8,3,0.027285332481066387
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,float16,3,0.030837332208951313
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,float16,7,0.03011200080315272
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,fp8,15,0.028160000840822857
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,float16,31,0.03017599880695343
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,fp8,31,0.027306665976842243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,float16,63,0.02985599885384242
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,fp8,63,0.027274665733178455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,fp8,7,0.027434666951497395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,float16,15,0.02977599948644638
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,float16,127,0.035173334181308746
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,fp8,127,0.02718399961789449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,float16,255,0.05129066606362661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,fp8,255,0.035562666753927864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,float16,511,0.08707732955614726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,float16,1,0.053946668903032936
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,fp8,511,0.05301866432030996
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,fp8,1,0.045498669147491455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,fp8,1023,0.090421328941981
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,float16,3,0.05409066875775655
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,fp8,3,0.045824001232783
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,2,128,1,float16,float16,1023,0.16142933567365012
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,float16,7,0.053727999329566956
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,fp8,7,0.045221333702405296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,fp8,15,0.04553600152333578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,float16,15,0.0539680023988088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,fp8,31,0.04560533165931702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,float16,63,0.05417599777380625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,fp8,127,0.04985066751639048
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,float16,127,0.05773333211739858
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,float16,255,0.08840533097585042
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,fp8,255,0.057775999108950295
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,float16,31,0.054058666030565895
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,fp8,63,0.04576000074545542
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,fp8,511,0.09306666254997253
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,float16,511,0.1612266699473063
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,float16,1,0.030410667260487873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,fp8,1,0.027263998985290527
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,fp8,1023,0.16504533092180887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,float16,3,0.030031998952229817
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,4,4,128,1,float16,float16,1023,0.304149329662323
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,fp8,3,0.02733866622050603
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,float16,7,0.02958933264017105
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,fp8,7,0.027237333357334137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,float16,31,0.02942399928967158
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,float16,63,0.029861333469549816
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,float16,15,0.03012266755104065
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,fp8,63,0.027503999571005504
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,float16,127,0.03522666543722153
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,fp8,127,0.027045334378878277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,fp8,15,0.027237333357334137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,fp8,31,0.02717866748571396
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,float16,255,0.0514933317899704
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,fp8,255,0.035461333890755974
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,float16,511,0.08900800347328186
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,1,128,1,float16,fp8,511,0.05323733389377594
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,float16,1,0.05397333204746246
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,fp8,1,0.04571733375390371
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,float16,3,0.053818667928377785
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,fp8,3,0.045941332976023354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,float16,7,0.05388266841570536
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,fp8,7,0.045824001232783
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,float16,15,0.05398400127887726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,float16,31,0.053898667295773826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,fp8,31,0.045381332437197365
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,fp8,15,0.04576533536116282
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,float16,63,0.05367999772230784
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,fp8,63,0.04543466866016388
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,float16,127,0.058090666929880776
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,fp8,127,0.049728001157442726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,float16,255,0.09087999661763509
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,fp8,255,0.05880000193913778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,fp8,1,0.08516266942024231
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,float16,1,0.09738133351008098
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,float16,511,0.1641493340333303
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,float16,3,0.09659199913342793
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,fp8,3,0.08530132969220479
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,2,128,1,float16,fp8,511,0.09283199906349182
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,fp8,7,0.08635733524958293
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,float16,7,0.09671466549237569
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,fp8,15,0.08500799536705017
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,fp8,31,0.08468799789746602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,float16,15,0.09567466378211975
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,float16,63,0.0969493289788564
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,fp8,63,0.08522133032480876
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,float16,31,0.09731200337409973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,float16,127,0.10291199882825215
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,1,0.009285333255926767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,1,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,3,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,7,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,7,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,15,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,15,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,fp8,127,0.08700799942016602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,31,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,3,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,63,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,127,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,127,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,255,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,255,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,511,0.009706666693091393
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,511,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,1023,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,1023,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,2047,0.011130666981140772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,2047,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,4095,0.013306666165590286
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,4095,0.013066666821638743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,63,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,8191,0.016048000504573185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,8191,0.01492799942692121
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,16383,0.03169599920511246
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,16383,0.017082666357358296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,32767,0.05023466547330221
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,32767,0.03342399994532267
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,65535,0.08717333277066548
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,65535,0.05287466446558634
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,float16,255,0.16680532693862915
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,fp8,131071,0.08963732918103536
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,1,128,1,float16,float16,131071,0.16220800081888834
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,3,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,3,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,1,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,7,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,7,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,15,0.008629333227872849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,15,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,31,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,31,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,63,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,63,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,1,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,127,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,255,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,fp8,255,0.10410666465759277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,511,0.009935999910036722
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,511,0.010661333799362183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,1023,0.011231999844312668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,2047,0.011717333147923151
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,4095,0.015135999768972397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,4095,0.013045333325862885
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,fp8,511,0.17285333077112833
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,8191,0.01509333277742068
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,8191,0.029696000119050343
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,16383,0.04776533444722494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,32767,0.0505920002857844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,32767,0.08743466933568318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,2047,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,65535,0.15890133380889893
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,65535,0.08709866801897685
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,127,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,255,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,1,0.009589333087205887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,1023,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,1,0.009872000043590864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,3,0.00938666673998038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,7,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,3,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,15,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,15,0.00933333362142245
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,float16,131071,0.3052053252855937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,131071,0.16152532895406088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,31,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,31,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,63,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,127,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,127,0.009248000259200731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,255,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,63,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,255,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,511,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,511,0.011045332998037338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,1023,0.011855999628702799
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,1023,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,2047,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,2047,0.0129120002190272
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,4095,0.029343999922275543
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,4095,0.015109332899252573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,8191,0.03147733211517334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,8191,0.04780800143877665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,16383,0.04931733508904775
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,16383,0.08450133601824443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,32767,0.08674133817354839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,32767,0.15681599577267966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,1,0.009370666618148485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,65535,0.16102400422096252
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,65535,0.2998080054918925
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,1,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,131071,0.5885386864344279
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,float16,7,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,4,4,128,1,float16,float16,511,0.309717337290446
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,3,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,4,128,1,float16,fp8,131071,0.3069546620051066
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,7,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,3,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,15,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,15,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,7,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,31,0.009269333134094873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,63,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,31,0.009242666885256767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,127,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,255,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,255,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,511,0.011152000476916632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,1023,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,2047,0.011029332876205444
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,2047,0.011002667248249054
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,4095,0.012960000584522883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,4095,0.013007999708255133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,8191,0.029450667401154835
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,8191,0.015376000354687372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,16383,0.04827199876308441
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,16383,0.03126933425664902
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,32767,0.08611733714739482
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,32767,0.05004266897837321
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,4,2,128,1,float16,fp8,16383,0.031119999786218006
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,127,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,511,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,1023,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,1,0.009162666896979014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,1,0.00926399976015091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,3,0.009349333122372627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,65535,0.08869866530100505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,3,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,7,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,131071,0.30640000104904175
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,7,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,15,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,31,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,15,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,31,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,fp8,131071,0.16695467631022134
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,63,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,63,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,127,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,255,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,255,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,511,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,1023,0.011247999966144562
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,511,0.010581333190202713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,1023,0.011136000355084738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,4095,0.02917333443959554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,2047,0.013007999708255133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,4095,0.015119999647140503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,8191,0.04775466521581014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,8191,0.031093334158261616
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,16383,0.08543466528256734
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,16383,0.0498986691236496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,32767,0.15812266866366068
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,1,128,1,float16,float16,65535,0.1611146628856659
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,65535,0.30586665868759155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,1,0.009205333267649015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,65535,0.1604639987150828
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,131071,0.31221866607666016
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,3,0.008634666601816813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,1,0.009189333145817121
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,3,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,7,0.008613333106040955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,7,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,15,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,127,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,31,0.008576000109314919
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,31,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,63,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,15,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,127,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,255,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,255,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,511,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,511,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,1023,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,1023,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,2047,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,2047,0.01313599944114685
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,2047,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,4095,0.010784000158309937
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,4095,0.010687999427318573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,8191,0.013141332815090815
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,16383,0.01741333305835724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,16383,0.017301333447297413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,32767,0.02510400116443634
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,32767,0.02537599951028824
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,65535,0.029498666524887085
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,65535,0.029311999678611755
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,131071,0.03151999910672506
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,131071,0.03123733401298523
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,fp8,32767,0.08685866991678874
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,1,0.008736000085870424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,3,0.008618666479984919
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,7,0.009248000259200731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,3,0.00922133338948091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,7,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,15,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,15,0.009205333267649015
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,31,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,63,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,127,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,127,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,255,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,255,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,511,0.009765333185593287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,2,2,128,1,float16,float16,131071,0.5987786849339803
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,511,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,fp8,63,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,127,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,1023,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,2047,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,2047,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,4095,0.009653333574533463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,1023,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,4095,0.010656000425418219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,8191,0.013007999708255133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,16383,0.019167999426523846
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,16383,0.01897066707412402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,32767,0.021269333859284718
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,1,128,1,float16,float16,8191,0.013007999708255133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,32767,0.019920000185569126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,65535,0.023370665808518726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,65535,0.02231466770172119
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,131071,0.02458133300145467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,1,0.008661333471536636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,1,0.0085333331177632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,3,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,3,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,7,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,1,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,7,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,15,0.008709333216150602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,15,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,31,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,31,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,63,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,127,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,127,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,255,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,31,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,63,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,255,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,511,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,511,0.009253333633144697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,1023,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,1023,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,2047,0.009077333534757296
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,2047,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,4095,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,4095,0.010693332801262537
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,8191,0.01369599997997284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,8191,0.014794666320085526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,16383,0.019152000546455383
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,16383,0.017269333203633625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,32767,0.021375998854637146
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,32767,0.021125334004561108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,65535,0.023306667804718018
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,65535,0.02231466770172119
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,float16,131071,0.04274133344491323
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,1,128,1,float16,fp8,131071,0.024800000091393787
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,1,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,1,0.008623999853928884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,3,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,3,0.008874666566650072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,7,0.008725333337982496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,7,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,15,0.008672000219424566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,15,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,31,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,31,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,63,0.009808000177145004
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,63,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,127,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,127,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,255,0.009466666728258133
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,511,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,511,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,1023,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,1023,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,2047,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,fp8,8191,0.014442666123310724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,2047,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,4095,0.010746666540702185
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,4095,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,8191,0.01481066644191742
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,8191,0.015034666905800501
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,16383,0.016741332908471424
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,32767,0.01932266727089882
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,32767,0.01823466643691063
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,65535,0.03555733213822047
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,65535,0.019253333409627277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,2,2,128,1,float16,float16,131071,0.043050666650136314
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,131071,0.03728533287843069
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,1,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,1,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,3,0.008778666456540426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,3,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,7,0.009354666496316591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,7,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,15,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,15,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,31,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,31,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,63,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,63,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,127,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,127,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,255,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,255,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,511,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,511,0.010591999938090643
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,1023,0.010933333386977514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,1023,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,2047,0.013082666943470636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,2047,0.012853333105643591
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,4095,0.029717333614826202
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,fp8,255,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,4095,0.015130666395028433
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,8191,0.04857600231965383
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,8191,0.031231999397277832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,16383,0.050016000866889954
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,16383,0.017050666113694508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,2,2,128,1,float16,float16,131071,0.05532266696294149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,16383,0.08561600248018901
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,1,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,32767,0.08867200215657552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,1,0.010053333515922228
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,3,0.010453333457310995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,32767,0.1597653329372406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,7,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,15,0.009797333429257074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,7,0.010117333382368088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,float16,65535,0.30798933903376263
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,15,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,31,0.009119999905427298
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,1,128,1,float16,fp8,65535,0.16100800037384033
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,31,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,63,0.009258666386206945
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,63,0.01002133327225844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,127,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,255,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,511,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,127,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,511,0.010911999891201654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,1023,0.012837332983811697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,1023,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,255,0.00938666673998038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,4095,0.04772266745567322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,4095,0.030058667063713074
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,2047,0.02938133229811986
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,16383,0.0862666666507721
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,8191,0.08494399984677632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,16383,0.15843733151753744
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,3,0.009253333633144697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,1,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,1,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,3,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,32767,0.30582932631174725
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,7,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,15,0.009301333377758661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,15,0.008863999818762144
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,31,0.008672000219424566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,32767,0.15971733132998148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,7,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,65535,0.30828799804051715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,31,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,63,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,127,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,255,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,255,0.008613333106040955
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,511,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,2047,0.014741333822409311
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,511,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,127,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,1023,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,2047,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,2047,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,4095,0.010597333312034607
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,1023,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,4095,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,8191,0.014773332824309668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,8191,0.014741333822409311
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,16383,0.017184000462293625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,16383,0.01709866647919019
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,32767,0.017925333231687546
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,65535,0.03589866558710734
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,65535,0.020997333029905956
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,131071,0.05454400181770325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,131071,0.037834666669368744
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,1,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,fp8,3,0.010037333394090334
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,1,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,3,0.009088000282645226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,3,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,7,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,7,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,15,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,15,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,31,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,63,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,63,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,127,0.008992000172535578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,255,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,255,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,511,0.00922133338948091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,511,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,float16,65535,0.5997493267059326
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,1023,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,1023,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,2047,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,4095,0.012842666357755661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,4095,0.012725333372751871
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,8191,0.015098666151364645
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,2047,0.010885333021481832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,8191,0.014762666076421738
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,63,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,16383,0.01720000058412552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,16383,0.01515199989080429
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,32767,0.03316266586383184
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,32767,0.01907733331123988
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,65535,0.051669334371884666
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,65535,0.034901333351929985
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,float16,131071,0.08858666817347209
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,131071,0.05352533360322317
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,2,2,128,1,float16,fp8,8191,0.04852266609668732
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,1,0.01098666712641716
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,1,0.010714666297038397
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,3,0.009423999736706415
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,3,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,7,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,7,0.010634666929642359
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,15,0.00919999989370505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,15,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,31,0.010879999647537867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,31,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,2,128,1,float16,fp8,127,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,63,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,63,0.010677333921194077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,127,0.010703999549150467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,127,0.009306666751702627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,255,0.010954666882753372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,511,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,511,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,1023,0.012944000462690989
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,1023,0.011749333391586939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,2047,0.02903466671705246
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,2047,0.015237333873907724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,8191,0.08571199576059978
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,2,1,128,1,float16,float16,32767,0.019274666905403137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,8191,0.04859733581542969
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,4095,0.047695999344189964
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,4095,0.031045332551002502
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,16383,0.15929599603017172
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,1,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,16383,0.08636266986529033
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,3,0.010677333921194077
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,float16,32767,0.3078719973564148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,7,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,1,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,32767,0.16051200032234192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,1,128,1,float16,fp8,255,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,3,0.011226666470368704
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,7,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,15,0.01081066702802976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,31,0.01090666651725769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,15,0.011034666250149408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,31,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,63,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,127,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,127,0.010901333143313726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,255,0.01121066634853681
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,255,0.010901333143313726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,63,0.01098666712641716
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,511,0.01310933381319046
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,511,0.01268799975514412
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,1023,0.02905600021282832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,4095,0.0499893327554067
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,1023,0.015024000157912573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,4095,0.08573866883913676
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,2047,0.03138133386770884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,8191,0.08769067128499348
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,8191,0.15901333093643188
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,16383,0.16192000110944113
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,1,0.013082666943470636
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,3,0.01163200040658315
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,1,0.011381333072980246
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,16383,0.3072906732559204
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,fp8,32767,0.3144426743189494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,7,0.01090666651725769
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,3,0.010890666395425797
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,2047,0.04785066843032837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,7,0.010911999891201654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,2,2,128,1,float16,float16,32767,0.6015733480453491
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,15,0.010960000256697336
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,15,0.010938666760921478
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,31,0.011146667102972666
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,63,0.010826667149861654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,127,0.011365332951148352
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,255,0.010911999891201654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,255,0.010794666906197866
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,511,0.01313599944114685
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,31,0.011130666981140772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,1023,0.029696000119050343
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,1023,0.015114666273196539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,63,0.01089599976936976
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,2047,0.04828799764315287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,127,0.011120000233252844
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,2047,0.031504000226656594
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,8191,0.16114133596420288
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,1,0.014565333724021912
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,511,0.012762666990359625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,4095,0.050330668687820435
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,float16,4095,0.0858026643594106
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,3,0.012981332838535309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,1,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,3,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,1,128,1,float16,fp8,8191,0.08707732955614726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,7,0.013482666263977686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,31,0.012938667088747025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,15,0.013141332815090815
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,63,0.015034666905800501
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,7,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,127,0.013221333424250284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,63,0.01320533330241839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,127,0.012725333372751871
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,255,0.013807999591032663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,15,0.014335999886194864
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,255,0.01332266628742218
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,511,0.016901332885026932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,511,0.029472000896930695
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,1023,0.04881600042184194
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,1023,0.03162133445342382
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,31,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,4095,0.16062399744987488
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,4095,0.08682133754094441
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,8191,0.3109920024871826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,float16,2047,0.08601599931716919
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,1,0.013151999562978745
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,2047,0.05004799862702688
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,1,0.013386666774749756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,2,2,128,1,float16,fp8,8191,0.16065067052841187
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,3,0.015024000157912573
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,3,0.01312000056107839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,7,0.014896000425020853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,15,0.01331199953953425
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,31,0.014671999961137772
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,7,0.012917333592971167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,63,0.013141332815090815
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,63,0.01492799942692121
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,31,0.01322666679819425
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,127,0.014917333920796713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,15,0.013199999928474426
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,127,0.01303999995191892
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,255,0.014730667074521383
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,255,0.013066666821638743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,511,0.017103999853134155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,511,0.030069333811601002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,1023,0.049125333627065025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,1023,0.0313226655125618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,2047,0.05020266771316528
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,1,0.01727466657757759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,1,0.019088000059127808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,3,0.018922666708628338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,3,0.017610666652520496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,7,0.01720000058412552
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,15,0.0191040001809597
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,7,0.01912533367673556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,4095,0.1627840002377828
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,fp8,4095,0.08772266904513042
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,1,128,1,float16,float16,2047,0.08707732955614726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,15,0.01717866708834966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,31,0.01922133316596349
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,31,0.016986666868130367
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,63,0.01718933383623759
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,255,0.017386666188637417
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,511,0.048026666045188904
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,511,0.031248000760873158
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,63,0.019173332800467808
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,255,0.02957333376010259
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,1023,0.08527466654777527
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,1023,0.0498933345079422
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,127,0.018960000326236088
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,3,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,3,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,4095,0.304149329662323
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,7,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,15,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,15,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,4095,0.1623253325621287
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,31,0.008757333581646284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,63,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,2047,0.08701866865158081
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,63,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,1,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,7,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,127,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,255,0.008821333448092142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,fp8,127,0.017018667111794155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,255,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,511,0.009589333087205887
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,511,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,1023,0.008853333070874214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,2047,0.009066666786869368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,31,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,2047,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,4095,0.012975999464591345
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,4095,0.012671999633312225
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,1,0.008656000097592672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,8191,0.01434133326013883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,16383,0.01714666684468587
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,32767,0.033439998825391136
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,32767,0.018917333334684372
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,65535,0.051925331354141235
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,65535,0.03526400029659271
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,131071,0.0537120004494985
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,131071,0.08874666690826416
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,127,0.008682666967312494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,1,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,2,2,128,1,float16,float16,2047,0.158053328593572
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,3,0.00879466657837232
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,3,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,7,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,15,0.00874133345981439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,31,0.008618666479984919
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,31,0.008623999853928884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,63,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,63,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,127,0.008581333483258883
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,127,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,float16,8191,0.015034666905800501
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,1023,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,255,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,15,0.008698666468262672
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,1,128,1,float16,fp8,16383,0.015077333897352219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,1023,0.009455999980370203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,1023,0.009045333291093508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,2047,0.01102399950226148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,2047,0.011370666325092316
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,4095,0.013183999806642532
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,4095,0.013093333691358566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,8191,0.015290666371583939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,8191,0.01471466695268949
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,16383,0.03180799881617228
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,16383,0.01685333376129468
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,32767,0.04993600149949392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,32767,0.03346133232116699
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,65535,0.08681600292523702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,1,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,7,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,131071,0.1597653329372406
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,131071,0.09016000231107076
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,255,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,511,0.01009599988659223
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,fp8,65535,0.05178666611512502
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,1,0.019258666783571243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,1,0.017152000218629837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,3,0.019018666197856266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,3,0.01704000060757001
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,2,2,128,1,float16,float16,511,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,7,0.018895999838908512
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,7,0.01741333305835724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,15,0.019274666905403137
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,31,0.019071999937295914
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,15,0.017263999829689663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,31,0.016895999511082966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,63,0.018858666221300762
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,63,0.017258666455745697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,127,0.018922666708628338
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,127,0.017050666113694508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,255,0.03011200080315272
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,255,0.017194667210181553
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,511,0.031119999786218006
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,511,0.048394665122032166
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,1023,0.08584533135096233
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,1023,0.04991999765237173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,1,0.03051200012365977
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,1,0.027429332335789997
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,float16,2047,0.15842666228612265
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,1,128,1,float16,fp8,2047,0.08924266695976257
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,3,0.029829333225886028
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,3,0.027285332481066387
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,15,0.029893333713213604
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,7,0.029733332494894665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,7,0.027215999861558277
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,15,0.027424000203609467
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,63,0.02939733366171519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,63,0.027317332724730175
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,127,0.0351946676770846
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,127,0.02717866748571396
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,31,0.029477333029111225
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,255,0.03519999980926514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,255,0.050570666790008545
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,31,0.027306665976842243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,511,0.08801600337028503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,511,0.0521919975678126
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,1023,0.1607039968172709
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,1023,0.08947199583053589
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,float16,1,0.029301332930723827
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,fp8,1,0.027295999228954315
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,fp8,2047,0.16353600223859152
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,2,2,128,1,float16,float16,2047,0.30751466751098633
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,float16,3,0.029552000264326733
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,fp8,3,0.027210667729377747
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,float16,7,0.030282666285832722
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,float16,15,0.029546665648619335
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,fp8,7,0.02737066646416982
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,float16,63,0.029738667110602062
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,fp8,15,0.027306665976842243
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,fp8,63,0.027562665442625683
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,float16,127,0.03523733218510946
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,float16,31,0.03031466652949651
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,fp8,31,0.02717866748571396
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,fp8,127,0.027301333844661713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,float16,255,0.050954664746920265
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,fp8,255,0.0354720006386439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,float16,1,0.054010664423306785
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,fp8,1,0.04571199913819631
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,float16,511,0.08869866530100505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,fp8,511,0.053557331363360085
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,float16,3,0.05399466554323832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,float16,1023,0.1616266667842865
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,1,128,1,float16,fp8,1023,0.09013332923253377
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,fp8,3,0.04563733438650767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,fp8,7,0.04550399879614512
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,float16,7,0.05387733379999796
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,float16,15,0.054570664962132774
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,float16,31,0.05366399884223938
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,fp8,63,0.045610666275024414
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,fp8,15,0.045850664377212524
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,fp8,127,0.04974933465321859
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,float16,127,0.05892266829808553
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,fp8,31,0.04553600152333578
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,float16,63,0.05399466554323832
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,1,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,3,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,3,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,1,0.008896000062425932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,7,0.008799999952316284
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,7,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,15,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,15,0.009093333035707474
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,31,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,31,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,63,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,63,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,127,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,127,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,float16,255,0.0890933374563853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,fp8,255,0.058330665032068886
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,255,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,1023,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,1023,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,2047,0.011338666081428528
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,2047,0.011519999553759893
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,4095,0.013749333719412485
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,4095,0.012906666845083237
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,8191,0.01551466683546702
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,8191,0.014933332800865173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,255,0.008570666735370954
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,511,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,16383,0.03126399964094162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,16383,0.017162666966517765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,32767,0.03322133421897888
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,511,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,65535,0.08718933661778767
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,65535,0.05286933481693268
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,1,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,131071,0.16208533445994058
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,fp8,131071,0.09158933162689209
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,1,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,3,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,3,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,7,0.008623999853928884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,7,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,15,0.009296000003814697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,15,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,31,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,31,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,63,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,fp8,511,0.09196266531944275
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,127,0.008949333180983862
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,127,0.00884799969693025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,float16,511,0.16325866182645163
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,255,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,511,0.010949333508809408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,1023,0.010661333799362183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,1023,0.010911999891201654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,1,128,1,float16,float16,32767,0.04985600213209788
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,2047,0.011098666737476984
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,2047,0.011146667102972666
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,4095,0.014149333039919535
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,4095,0.012874666601419449
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,8191,0.029658667743206024
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,8191,0.01488000030318896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,16383,0.031141333281993866
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,32767,0.08476266264915466
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,32767,0.04975999891757965
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,65535,0.15953600406646729
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,65535,0.08768533666928609
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,63,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,1,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,255,0.011786667009194693
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,1,0.009429333110650381
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,3,0.009301333377758661
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,511,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,float16,1023,0.30883200963338214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,3,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,7,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,fp8,131071,0.1618880033493042
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,2,2,128,1,float16,fp8,1023,0.1648426651954651
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,15,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,131071,0.305567999680837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,15,0.00897066667675972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,2,2,128,1,float16,float16,16383,0.04781333108743032
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,63,0.009103999783595404
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,63,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,127,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,255,0.009056000038981438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,255,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,511,0.009322666873534521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,511,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,1023,0.011066666493813196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,1023,0.010661333799362183
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,2047,0.012261333564917246
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,2047,0.010725333044926325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,4095,0.014933332800865173
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,4095,0.012773333738247553
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,8191,0.02957333376010259
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,8191,0.015119999647140503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,16383,0.049322664737701416
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,32767,0.08546132842699687
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,16383,0.03140799949566523
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,32767,0.05036266644795736
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,7,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,65535,0.0897759993871053
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,1,0.00873066671192646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,31,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,65535,0.1607253352801005
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,3,0.00915733352303505
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,3,0.009296000003814697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,7,0.009306666751702627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,7,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,15,0.0086666668454806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,15,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,131071,0.3089013298352559
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,31,0.008746666833758354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,31,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,63,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,fp8,131071,0.16580800215403238
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,127,0.008634666601816813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,127,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,255,0.008650666723648706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,511,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,511,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,1023,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,1023,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,63,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,2047,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,2047,0.008816000074148178
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,4095,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,4095,0.009786666681369146
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,8191,0.012810666114091873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,8191,0.01312000056107839
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,16383,0.01691199963291486
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,16383,0.017301333447297413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,32767,0.02533866713444392
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,32767,0.025392000873883564
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,65535,0.029493334392706554
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,65535,0.029157333076000214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,float16,131071,0.031328000128269196
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,131071,0.031194667021433514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,1,0.008890666688481966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,1,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,3,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,3,0.0086666668454806
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,7,0.00860799973209699
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,7,0.00867733359336853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,15,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,15,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,31,0.008693333094318708
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,63,0.008602666358153025
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,63,0.00860799973209699
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,127,0.008634666601816813
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,127,0.008687999720374743
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,255,0.008714666590094566
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,255,0.008650666723648706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,1,0.009082666908701261
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,511,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,511,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,1023,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,1023,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,2047,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,2047,0.009072000160813332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,4095,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,4095,0.010911999891201654
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,8191,0.013301332791646322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,16383,0.0189280000825723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,16383,0.019152000546455383
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,32767,0.02111999938885371
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,31,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,32767,0.02128000060717265
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,65535,0.023200000325838726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,65535,0.02312533309062322
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,32,1,1,1,128,1,float16,float16,127,0.00871999996403853
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,fp8,131071,0.02516266703605652
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,1,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,1,0.009455999980370203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,3,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,3,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,7,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1,1,1,1,128,1,float16,fp8,255,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,15,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,15,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,31,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,63,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,31,0.008805333326260248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,31,0.008810666700204214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,63,0.009018666421373686
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,127,0.008922666932145754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,127,0.00961599995692571
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,255,0.009061333412925402
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,255,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,511,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,511,0.010949333508809408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,1023,0.010858666151762009
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,8191,0.013850666582584381
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,1023,0.01119999960064888
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,2047,0.012794667234023413
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,2047,0.012901333471139273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2,1,1,1,128,1,float16,float16,131071,0.04293866455554962
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,8191,0.0497920016447703
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,4095,0.01498666654030482
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,8191,0.030832000076770782
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,16383,0.04965866605440775
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,7,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,16383,0.08565866947174072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,32767,0.16034666697184244
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,32767,0.08872532844543457
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,1,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,3,0.00922133338948091
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,7,0.008901333436369896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,3,0.009269333134094873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,7,0.009194666519761086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,1,0.009194666519761086
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,15,0.009226666763424873
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,65535,0.3083840012550354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,31,0.009338666374484697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,31,0.00927466650803884
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,4095,0.030037333567937214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,63,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,63,0.008986666798591614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,255,0.009178666397929192
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,255,0.009253333633144697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,127,0.008938666433095932
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,511,0.00943999985853831
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,511,0.009663999701539675
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,1023,0.009248000259200731
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,1023,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,131071,0.3107466697692871
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,2047,0.00938666673998038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,2047,0.01003200002014637
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,4095,0.009930666536092758
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,8191,0.015050667027632395
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,float16,131071,0.6043200095494589
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,16383,0.01613866661985715
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,4095,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,32767,0.017237332959969837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,65535,0.01937599976857503
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,65535,0.0355679988861084
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,32767,0.01923199991385142
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,131071,0.05494399865468343
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,131071,0.03748800108830134
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,fp8,15,0.009375999992092451
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,64,1,1,1,128,1,float16,fp8,65535,0.16205333669980368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,1,0.01073066641887029
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,1,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,3,0.010794666906197866
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,127,0.009306666751702627
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,3,0.009125333279371262
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,7,0.01062400018175443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,15,0.008943999807039896
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,15,0.010634666929642359
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,31,0.010656000425418219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,63,0.010768000036478043
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,31,0.009653333574533463
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,63,0.009114666531483332
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,127,0.010832000523805618
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,127,0.0107893335322539
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,8191,0.014981333166360855
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,255,0.010640000303586325
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,255,0.009008000294367472
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,4,1,1,1,128,1,float16,float16,16383,0.017093333105246227
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,511,0.010709332923094431
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,1023,0.012762666990359625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,1023,0.012997332960367203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,2047,0.029285334050655365
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,2047,0.014917333920796713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,8191,0.08469866712888081
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,8191,0.04970133304595947
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,4095,0.030053332448005676
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,7,0.00980266680320104
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,32767,0.3081173300743103
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,16383,0.15964266657829285
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,16383,0.08714666962623596
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,511,0.010847999403874079
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,32767,0.1629866659641266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,1,0.011301333705584208
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,1,0.011071999867757162
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,3,0.011285333583752314
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,4095,0.04794133206208547
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,fp8,65535,0.30959467093149823
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,7,0.010965333630641302
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,3,0.010922666639089584
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,7,0.011007999380429586
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,15,0.010949333508809408
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,128,1,1,1,128,1,float16,float16,65535,0.6047146717707316
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,31,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,63,0.010863999525705973
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,63,0.010816000401973724
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,127,0.011178666104873022
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,15,0.010970667004585266
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,255,0.011018666128317514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,255,0.011152000476916632
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,511,0.01452800010641416
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,511,0.012997332960367203
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,31,0.010751999914646149
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,1023,0.029824001093705494
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,1023,0.014917333920796713
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,2047,0.048250665267308555
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,2047,0.031178665657838184
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,8191,0.15984533230463663
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,127,0.011039999624093374
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,8191,0.08820799986521403
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,16383,0.3088373343149821
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,4095,0.05146666864554087
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,float16,4095,0.08593599994977315
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,3,0.013610667238632837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,256,1,1,1,128,1,float16,fp8,16383,0.1669600009918213
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,1,0.014794666320085526
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,3,0.013061333447694778
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,7,0.014805333067973455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,15,0.01293333371480306
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,31,0.012730666746695837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,63,0.014208000153303146
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,7,0.013002666334311167
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,31,0.013776000589132309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,15,0.01360000049074491
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,1,0.012863999853531519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,63,0.012901333471139273
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,127,0.013178666432698568
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,127,0.013088000317414602
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,255,0.014127999544143677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,511,0.02938666691382726
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,511,0.017071999609470367
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,1023,0.04844266672929128
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,1023,0.03140799949566523
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,1,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,2047,0.05018133421738943
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,1,0.00878399983048439
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,255,0.012991999586423239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,2047,0.0860693355401357
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,3,0.008767999708652496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,15,0.009253333633144697
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,15,0.008832000195980072
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,31,0.008789333204428354
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,7,0.008842666943868002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,31,0.008752000207702318
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,63,0.00895999992887179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,63,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,127,0.008912000184257826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,127,0.008954666554927826
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,255,0.009130666653315226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,255,0.008933333059151968
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,8191,0.31565866867701214
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,511,0.009183999771873156
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,511,0.009861333295702934
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,8191,0.16250133514404297
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,1023,0.008826666822036108
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,2047,0.00898133342464765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,2047,0.010735999792814255
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,4095,0.012991999586423239
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,8191,0.015077333897352219
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,8191,0.01332266628742218
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,float16,4095,0.16038399934768677
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,16383,0.01709866647919019
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,512,1,1,1,128,1,float16,fp8,4095,0.08711999654769897
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,16383,0.015226667126019796
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,1023,0.008965333302815756
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,32767,0.018986667195955913
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,3,0.00891733355820179
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,4095,0.012762666990359625
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,7,0.008762666955590248
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,131071,0.05362666646639506
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,131071,0.08777599533398946
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,32767,0.033285332222779594
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,float16,65535,0.05179200073083242
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,1,0.01921066641807556
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,8,1,1,1,128,1,float16,fp8,65535,0.033733333150545754
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,3,0.019215999792019527
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,3,0.017029333859682083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,7,0.01911466692884763
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,7,0.017029333859682083
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,15,0.018858666221300762
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,1,0.017162666966517765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,31,0.01926400015751521
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,15,0.01704000060757001
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,31,0.01717866708834966
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,63,0.019082666685183842
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,63,0.01708799973130226
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,127,0.0189280000825723
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,127,0.01714133347074191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,255,0.029792000850041706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,255,0.017103999853134155
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,511,0.04859200119972229
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,511,0.03149333347876867
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,1023,0.04997866849104563
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,1023,0.08518399794896443
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,2047,0.08648000160853068
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,1,0.03035733352104823
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,1,0.02738133321205775
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,2047,0.15919466813405356
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,float16,4095,0.30566932757695514
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,1024,1,1,1,128,1,float16,fp8,4095,0.16406400005022684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,3,0.029733332494894665
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,3,0.027162666122118633
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,7,0.02956799914439519
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,7,0.027450665831565857
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,15,0.0295413335164388
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,63,0.03025600065787633
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,15,0.027242665489514668
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,63,0.027274665733178455
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,127,0.03455466777086258
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,255,0.05051200091838837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,127,0.027130665878454845
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,31,0.029792000850041706
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,255,0.03515200068553289
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,1,0.00901333304742972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,31,0.027386667827765148
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,1,0.008976000050703684
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,3,0.008885333314538002
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,3,0.00892800030608972
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,7,0.008997333546479544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,7,0.008879999940594038
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,15,0.009098666409651438
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,15,0.009039999917149544
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,31,0.00877333308259646
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,511,0.08841066559155782
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,31,0.008837333569924036
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,63,0.009029333169261614
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,63,0.009109333157539368
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,127,0.009002666920423508
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,255,0.009136000027259191
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,255,0.00902399979531765
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,511,0.00903466654320558
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,511,0.010666667173306147
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,511,0.052239999175071716
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,1023,0.009397333487868309
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,1023,0.01097600037852923
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,2047,0.010992000500361124
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,2047,0.011749333391586939
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,8191,0.01586666703224182
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,4095,0.013050666699806849
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,8191,0.014736000448465347
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,16383,0.03160000095764796
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,16383,0.017152000218629837
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,127,0.00890666681031386
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,32767,0.05016533533732096
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,32767,0.032560000816980995
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,65535,0.08704533179601033
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,65535,0.052058666944503784
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,131071,0.16248533129692078
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,fp8,131071,0.08878933389981587
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,16,1,1,1,128,1,float16,float16,4095,0.013104000439246496
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,1023,0.16103466351826987
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,1023,0.09079999725023906
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,float16,2047,0.3086026708285014
VLLM,0.14.1.dev1+gd68209402,NVIDIA B200,generation_attention,vllm_flashinfer,2048,1,1,1,128,1,float16,fp8,2047,0.16401066382726034
