kernel_bench_full.json

{
  "device": "cuda",
  "dtype": "torch.float16",
  "rows": [
    {
      "L": 512,
      "D": 1,
      "mha_ms": 0.09538995000184514,
      "mla_ms": 0.3815698000039447
    },
    {
      "L": 512,
      "D": 2,
      "mha_ms": 0.09486645000151839,
      "mla_ms": 0.38125579999359616
    },
    {
      "L": 512,
      "D": 4,
      "mha_ms": 0.0951449500007584,
      "mla_ms": 0.38137284999493204
    },
    {
      "L": 512,
      "D": 8,
      "mha_ms": 0.09517000000869302,
      "mla_ms": 0.3810957999917264
    },
    {
      "L": 512,
      "D": 16,
      "mha_ms": 0.09609395000325094,
      "mla_ms": 0.3797977499971239
    },
    {
      "L": 512,
      "D": 32,
      "mha_ms": 0.09909755000307996,
      "mla_ms": 0.3867594499979532
    },
    {
      "L": 512,
      "D": 64,
      "mha_ms": 0.10272314999610899,
      "mla_ms": 0.4647889500006386
    },
    {
      "L": 2048,
      "D": 1,
      "mha_ms": 0.35331259999793474,
      "mla_ms": 1.1370887499992932
    },
    {
      "L": 2048,
      "D": 2,
      "mha_ms": 0.3535740999950576,
      "mla_ms": 1.1424018999946384
    },
    {
      "L": 2048,
      "D": 4,
      "mha_ms": 0.3533986499974162,
      "mla_ms": 1.1358536999978242
    },
    {
      "L": 2048,
      "D": 8,
      "mha_ms": 0.3538531000003786,
      "mla_ms": 1.1377807999906508
    },
    {
      "L": 2048,
      "D": 16,
      "mha_ms": 0.35495164999019835,
      "mla_ms": 1.1279199999989942
    },
    {
      "L": 2048,
      "D": 32,
      "mha_ms": 0.35747325000556884,
      "mla_ms": 1.1601597999970181
    },
    {
      "L": 2048,
      "D": 64,
      "mha_ms": 0.3610117999983231,
      "mla_ms": 1.4664202000062687
    },
    {
      "L": 8192,
      "D": 1,
      "mha_ms": 1.3871011499986707,
      "mla_ms": 4.100175049995869
    },
    {
      "L": 8192,
      "D": 2,
      "mha_ms": 1.387128200008192,
      "mla_ms": 4.1041406499971345
    },
    {
      "L": 8192,
      "D": 4,
      "mha_ms": 1.387747700005093,
      "mla_ms": 4.101858600006381
    },
    {
      "L": 8192,
      "D": 8,
      "mha_ms": 1.3878852500056382,
      "mla_ms": 4.103569650010286
    },
    {
      "L": 8192,
      "D": 16,
      "mha_ms": 1.3885843000025488,
      "mla_ms": 4.033273849995567
    },
    {
      "L": 8192,
      "D": 32,
      "mha_ms": 1.3916749000031814,
      "mla_ms": 4.174569950009754
    },
    {
      "L": 8192,
      "D": 64,
      "mha_ms": 1.395025900001201,
      "mla_ms": 5.488053800002035
    },
    {
      "L": 32768,
      "D": 1,
      "mha_ms": 5.571495399999549,
      "mla_ms": 15.963458550004361
    },
    {
      "L": 32768,
      "D": 2,
      "mha_ms": 5.5245357999979205,
      "mla_ms": 15.928289600003607
    },
    {
      "L": 32768,
      "D": 4,
      "mha_ms": 5.575976099999025,
      "mla_ms": 15.956968850002795
    },
    {
      "L": 32768,
      "D": 8,
      "mha_ms": 5.523514699996213,
      "mla_ms": 15.936660299996674
    },
    {
      "L": 32768,
      "D": 16,
      "mha_ms": 5.572800050003934,
      "mla_ms": 15.694649099998514
    },
    {
      "L": 32768,
      "D": 32,
      "mha_ms": 5.575721499997144,
      "mla_ms": 16.2844683000003
    },
    {
      "L": 32768,
      "D": 64,
      "mha_ms": 5.578881200005981,
      "mla_ms": 22.003299000004972
    }
  ]
}