Liger

{
  "run": {
    "kind": "BenchmarkRun",
    "spec": {
      "status": "passed",
      "timing": {
        "latencyNs": {
          "median": 103127937
        },
        "primaryStatistic": "median"
      },
      "observedAt": "2024-10-02T16:21:40.000Z",
      "measurements": [
        {
          "unit": "ns",
          "value": 103127937,
          "metric": "latency",
          "statistic": "p20"
        },
        {
          "unit": "ns",
          "value": 103127937,
          "metric": "latency",
          "statistic": "p80"
        }
      ],
      "sourceNative": {
        "source": "liger-kernel-bench",
        "metrics": {
          "ms_20": 103.12793731689453,
          "ms_50": 103.12793731689453,
          "ms_80": 103.12793731689453
        },
        "benchmark": "jsd/full",
        "externalId": "jsd/full/liger/nvidia-h100-pcie/tokens8192-vocab131072-fp32/2024-10-02-16-21-40"
      },
      "protocolDigest": "sha256:5b02ee9a2140fbc6fcc1c89800f6c177ba22718c8897cb532bf0ba91867c1e58",
      "workloadDigest": "sha256:c70c89a34a2ca6b7547fc467f1d4e196826f67eaf466cdb659ef6a677c78dcd7",
      "environmentDigest": "sha256:89f9b28b06758f22a1df331fea58bfde2c9cc0f3627d1829509743ea538f0011",
      "implementationDigest": "sha256:cfb3edb701b526d49e58d1de743fa6b7ce8e9b2d82736c8a68b45577428f96fc"
    },
    "metadata": {
      "name": "liger-jsd-full-liger-nvidia-h100-pcie-tokens8192-vocab131072-fp32-2024-10-02-16-21-40",
      "title": "Jensen-Shannon divergence loss · liger · full",
      "labels": {
        "liger_version": "0.3.1"
      }
    },
    "apiVersion": "kernelindex.dev/v1alpha1"
  },
  "protocol": {
    "kind": "BenchmarkProtocol",
    "spec": {
      "harness": {
        "name": "Liger-Kernel benchmark scripts",
        "repository": "https://github.com/linkedin/Liger-Kernel"
      },
      "measurement": {
        "timer": "cuda_events",
        "primaryStatistic": "median"
      },
      "comparability": {
        "notes": "Timed: the forward pass plus the backward pass, via triton.testing.do_bench with quantiles 0.5/0.2/0.8 (median with p20/p80 spread). The upstream CSV records no CUDA, driver, or torch version; the Liger release rides each run's labels. Comparable only within one kernel, workload, timed pass, and GPU.",
        "family": "liger_kernel_bench"
      }
    },
    "metadata": {
      "name": "liger-bench-do-bench-full",
      "title": "Liger benchmark harness · full pass"
    },
    "apiVersion": "kernelindex.dev/v1alpha1"
  },
  "environment": {
    "kind": "ExecutionEnvironment",
    "spec": {
      "hardware": {
        "vendor": "nvidia",
        "product": "NVIDIA H100",
        "formFactor": "PCIe",
        "architecture": "sm_90"
      },
      "software": {}
    },
    "metadata": {
      "name": "liger-bench-nvidia-h100-pcie",
      "title": "Liger benchmark host · H100 PCIe"
    },
    "apiVersion": "kernelindex.dev/v1alpha1"
  }
}