{
  "experiment": {
    "candidate": "experiments/033-c1-deep-cluster-qkv.py",
    "mode": "submit",
    "job_id": "54eb1a07e4014863",
    "status": "done",
    "passed": true,
    "speedup": 1.2887917482340643,
    "speedup_ci": [
      1.2694598720105532,
      1.3081236244575751
    ],
    "number": 33,
    "problem": "muse_bf16_proj"
  },
  "result": {
    "passed": true,
    "speedup": 1.2887917482340643,
    "speedup_ci": [
      1.2694598720105532,
      1.3081236244575751
    ],
    "baseline_us": 4830.8,
    "score_us": 3748.3169849739393,
    "gpu": "H200",
    "records": [
      {
        "error": null,
        "event": "setup",
        "prepare_s": 32.635212009,
        "status": "pass"
      },
      {
        "case": {
          "layer": "qkv_gate",
          "m": 1,
          "seed": 700595995010753488
        },
        "event": "test",
        "index": 0,
        "status": "pass"
      },
      {
        "case": {
          "layer": "pooler",
          "m": 7,
          "seed": 700644173195089702
        },
        "event": "test",
        "index": 1,
        "status": "pass"
      },
      {
        "case": {
          "layer": "o",
          "m": 33,
          "seed": 700692351379425917
        },
        "event": "test",
        "index": 2,
        "status": "pass"
      },
      {
        "case": {
          "layer": "unpooler",
          "m": 130,
          "seed": 700740529563762133
        },
        "event": "test",
        "index": 3,
        "status": "pass"
      },
      {
        "case": {
          "layer": "unpooler",
          "m": 5,
          "seed": 700981420485443228
        },
        "event": "test",
        "index": 4,
        "status": "pass"
      },
      {
        "case": {
          "layer": "router",
          "m": 8,
          "seed": 700788707748098350
        },
        "event": "test",
        "index": 5,
        "status": "pass"
      },
      {
        "case": {
          "dist": "outliers",
          "layer": "router",
          "m": 9,
          "seed": 700836885932434568
        },
        "event": "test",
        "index": 6,
        "status": "pass"
      },
      {
        "case": {
          "dist": "outliers",
          "layer": "qkv_gate",
          "m": 512,
          "seed": 700885064116770787
        },
        "event": "test",
        "index": 7,
        "status": "pass"
      },
      {
        "case": {
          "dist": "outliers",
          "layer": "unpooler",
          "m": 1000,
          "seed": 700933242301107007
        },
        "event": "test",
        "index": 8,
        "status": "pass"
      },
      {
        "baseline_ns": 14495.999086648226,
        "case": {
          "layer": "qkv_gate",
          "load": "decode 1",
          "m": 1,
          "seed": 705413813444379838,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 0,
        "median_ns": 10706.522161133036,
        "paired_ratio": 1.3539409780770642,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 6912.000011652708,
        "case": {
          "layer": "o",
          "load": "decode 1",
          "m": 1,
          "seed": 705461991628716152,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 1,
        "median_ns": 5426.291549184719,
        "paired_ratio": 1.2737981269530592,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 21584.000904113054,
        "case": {
          "layer": "pooler",
          "load": "decode 1",
          "m": 1,
          "seed": 705510169813052467,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 2,
        "median_ns": 17204.833880636983,
        "paired_ratio": 1.2545312005833757,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 7040.000054985285,
        "case": {
          "layer": "unpooler",
          "load": "decode 1",
          "m": 1,
          "seed": 705558347997388783,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 3,
        "median_ns": 3859.582306933666,
        "paired_ratio": 1.8240315907599791,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 10224.000085145235,
        "case": {
          "layer": "router",
          "load": "decode 1",
          "m": 1,
          "seed": 705606526181725100,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 4,
        "median_ns": 3529.84237739632,
        "paired_ratio": 2.896446637565345,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 14063.999988138676,
        "case": {
          "layer": "qkv_gate",
          "load": "decode 16",
          "m": 16,
          "seed": 705654704366061418,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 5,
        "median_ns": 12585.705376649845,
        "paired_ratio": 1.117458224807288,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 6880.0002336502075,
        "case": {
          "layer": "o",
          "load": "decode 16",
          "m": 16,
          "seed": 705702882550397737,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 6,
        "median_ns": 6991.543138138834,
        "paired_ratio": 0.9840460249926571,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 21344.00000795722,
        "case": {
          "layer": "pooler",
          "load": "decode 16",
          "m": 16,
          "seed": 705751060734734057,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 7,
        "median_ns": 18658.392423062916,
        "paired_ratio": 1.1439356362542106,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 18415.99913313985,
        "case": {
          "layer": "unpooler",
          "load": "decode 16",
          "m": 16,
          "seed": 705799238919070378,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 8,
        "median_ns": 17939.88645425968,
        "paired_ratio": 1.0265393362490944,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 15279.99946847558,
        "case": {
          "layer": "router",
          "load": "decode 16",
          "m": 16,
          "seed": 705847417103406700,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 9,
        "median_ns": 5238.738033674665,
        "paired_ratio": 2.9167328792269394,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 14288.000296801329,
        "case": {
          "layer": "qkv_gate",
          "load": "decode 32",
          "m": 32,
          "seed": 705895595287743023,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 10,
        "median_ns": 13860.736761906493,
        "paired_ratio": 1.0308254562678865,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 6927.999202162027,
        "case": {
          "layer": "o",
          "load": "decode 32",
          "m": 32,
          "seed": 705943773472079347,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 11,
        "median_ns": 7007.939678761975,
        "paired_ratio": 0.9885928703350269,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 22159.999702125788,
        "case": {
          "layer": "pooler",
          "load": "decode 32",
          "m": 32,
          "seed": 705991951656415672,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 12,
        "median_ns": 20266.850092591234,
        "paired_ratio": 1.0934111418836918,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 18687.999341636896,
        "case": {
          "layer": "unpooler",
          "load": "decode 32",
          "m": 32,
          "seed": 706040129840751998,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 13,
        "median_ns": 18177.77134440086,
        "paired_ratio": 1.028068787288008,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 15007.999259978533,
        "case": {
          "layer": "router",
          "load": "decode 32",
          "m": 32,
          "seed": 706088308025088325,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 14,
        "median_ns": 5434.748765216179,
        "paired_ratio": 2.7614890601813418,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 14431.999996304512,
        "case": {
          "layer": "qkv_gate",
          "load": "verify 1",
          "m": 8,
          "seed": 706136486209424653,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 15,
        "median_ns": 12256.119832058852,
        "paired_ratio": 1.1775341783583184,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 6783.999502658844,
        "case": {
          "layer": "o",
          "load": "verify 1",
          "m": 8,
          "seed": 706184664393760982,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 16,
        "median_ns": 6800.0036294452375,
        "paired_ratio": 0.9976464531993641,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 21503.998897969723,
        "case": {
          "layer": "pooler",
          "load": "verify 1",
          "m": 8,
          "seed": 706232842578097312,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 17,
        "median_ns": 18576.385278003963,
        "paired_ratio": 1.1575986703630823,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 6960.00037714839,
        "case": {
          "layer": "unpooler",
          "load": "verify 1",
          "m": 8,
          "seed": 706281020762433643,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 18,
        "median_ns": 4868.67433748631,
        "paired_ratio": 1.4295473253489015,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 26528.000365942717,
        "case": {
          "layer": "router",
          "load": "verify 1",
          "m": 8,
          "seed": 706329198946769975,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 19,
        "median_ns": 5377.846743669711,
        "paired_ratio": 4.932829370262169,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 17072.000075131655,
        "case": {
          "layer": "qkv_gate",
          "load": "verify 16",
          "m": 128,
          "seed": 706377377131106308,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 20,
        "median_ns": 16918.31527313881,
        "paired_ratio": 1.0090839306108008,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 7327.9994539916515,
        "case": {
          "layer": "o",
          "load": "verify 16",
          "m": 128,
          "seed": 706425555315442642,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 21,
        "median_ns": 7222.764214450696,
        "paired_ratio": 1.0145699397649461,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 26064.00055810809,
        "case": {
          "layer": "pooler",
          "load": "verify 16",
          "m": 128,
          "seed": 706473733499778977,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 22,
        "median_ns": 23196.012451051978,
        "paired_ratio": 1.1236414281596079,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 20191.99961796403,
        "case": {
          "layer": "unpooler",
          "load": "verify 16",
          "m": 128,
          "seed": 706521911684115313,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 23,
        "median_ns": 19309.554212657244,
        "paired_ratio": 1.0456999367043052,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 16128.000803291798,
        "case": {
          "layer": "router",
          "load": "verify 16",
          "m": 128,
          "seed": 706570089868451650,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 24,
        "median_ns": 6595.10283510246,
        "paired_ratio": 2.445450996980737,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 20112.00087144971,
        "case": {
          "layer": "qkv_gate",
          "load": "verify 32",
          "m": 256,
          "seed": 706618268052787988,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 25,
        "median_ns": 19645.070877228107,
        "paired_ratio": 1.0237683028551885,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 9120.000060647726,
        "case": {
          "layer": "o",
          "load": "verify 32",
          "m": 256,
          "seed": 706666446237124327,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 26,
        "median_ns": 9055.810207558552,
        "paired_ratio": 1.0070882507051215,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 29232.00000077486,
        "case": {
          "layer": "pooler",
          "load": "verify 32",
          "m": 256,
          "seed": 706714624421460667,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 27,
        "median_ns": 28992.01775560059,
        "paired_ratio": 1.0082775282216399,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 24287.999141961336,
        "case": {
          "layer": "unpooler",
          "load": "verify 32",
          "m": 256,
          "seed": 706762802605797008,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 28,
        "median_ns": 23852.135795483955,
        "paired_ratio": 1.0182735563060104,
        "runs": 60,
        "status": "pass"
      },
      {
        "baseline_ns": 17344.00074928999,
        "case": {
          "layer": "router",
          "load": "verify 32",
          "m": 256,
          "seed": 706810980790133350,
          "weight": 62
        },
        "event": "benchmark",
        "flags": [],
        "index": 29,
        "median_ns": 8217.759022575923,
        "paired_ratio": 2.1105511492418247,
        "runs": 60,
        "status": "pass"
      }
    ]
  }
}