Run 01a02032 · KernelIndex

Primary measurement

96.9µsmean

Rank 5 in its comparison group · source-native comparison · observed 2025-10-19

Compare with #1 →

3.86× reference

Identity

implementationgpt-o3_cuda_025ec1

projectgpt-o3

revisionda915083d4c7c5e61aa3005e3d17ae488e0fc71c

workloadtotal_q = 18 · total_kv = 18 · len_indptr = 2 · bf16

comparison keysha256:ac7ab26a87798d40…

sourceFlashInfer-Bench

external idtrace/gpt-o3_cuda_025ec1/568a6016-6234-4700-ae5e-061593e16640

sha256:7267b0e6cb7d3afaaf053db…

Correctness

comparatorflashinfer_bench_eval

max absolute error0.0009765625

max relative error0.023903118446469307

resultpassed

Workload

total_q18

total_kv18

len_indptr2

kbf16 [18, 8, 128]

qbf16 [18, 32, 128]

vbf16 [18, 8, 128]

kv_indptrint32 [2]

qo_indptrint32 [2]

definition comparatorsol_execbench_eval

Measurements

latency · mean96.9 µs

Protocol

harnessflashinfer-bench

timerharness_reported

primaryStatisticmean

correctnessReferencedefinition reference implementation

comparabilityFamilyflashinfer_bench_trace

Environment

gpuNVIDIA B200 (unknown)

cudaToolkit12.8

frameworkpytorch 2.8.0+cu128

Artifacts

logssha256:1411162689502ae2… · text/plain · unavailable

Replications and notes

No attestations yet.

Community attestations. They never change the evidence level; only a KernelIndex-controlled rerun does.

Add a reproduction or note

Canonical manifest

Show manifest

{
  "run": {
    "kind": "BenchmarkRun",
    "spec": {
      "status": "passed",
      "timing": {
        "latencyNs": {
          "mean": 96917
        },
        "primaryStatistic": "mean"
      },
      "evidence": {
        "logs": {
          "uri": "kernelindex:artifact/sha256:1411162689502ae2d733419f919a077a9f677d4f0accdbdb00f3d8a94c8774a6",
          "digest": "sha256:1411162689502ae2d733419f919a077a9f677d4f0accdbdb00f3d8a94c8774a6"
        }
      },
      "observedAt": "2025-10-19T23:00:56.540Z",
      "correctness": {
        "comparator": "sol_execbench_eval",
        "maximumAbsoluteError": 0.0009765625,
        "maximumRelativeError": 0.023903118446469307
      },
      "sourceNative": {
        "source": "flashinfer-bench",
        "metrics": {
          "latency_ms": 0.09691748749270301,
          "speedup_factor": 3.8562470997220206,
          "reference_latency_ms": 0.3737377800560812
        },
        "benchmark": "gqa_ragged_prefill_causal_h32_kv8_d128"
      },
      "protocolDigest": "sha256:7d44d206ad086b4bece35ce31f6531ebb11f07e0d9dcfb74c75be86d3ec02095",
      "workloadDigest": "sha256:2ca4d099e0a6755a7b016121caefd14ffc8294ae41242bd84292035f7b910f1f",
      "environmentDigest": "sha256:f4292a8be127b7230cd3995c1d605fb6c644e2c2f94566e1d28cc1be766ca5d5",
      "implementationDigest": "sha256:5acecad48c17870791beedfc3d3b0d9a71ca6932288f1c391313a146bb02752d"
    },
    "metadata": {
      "name": "sol-trace-gqa-ragged-prefill-causal-h32-kv8-d128-568a6016",
      "title": "gpt-o3_cuda_025ec1 · gqa_ragged_prefill_causal_h32_kv8_d128"
    },
    "apiVersion": "kernelindex.dev/v1alpha1"
  },
  "protocol": {
    "kind": "BenchmarkProtocol",
    "spec": {
      "harness": {
        "name": "flashinfer-bench",
        "repository": "https://github.com/flashinfer-ai/flashinfer-bench"
      },
      "correctness": {
        "reference": "definition reference implementation",
        "comparator": "flashinfer_bench_eval"
      },
      "measurement": {
        "timer": "harness_reported",
        "primaryStatistic": "mean"
      },
      "comparability": {
        "family": "flashinfer_bench_trace"
      }
    },
    "metadata": {
      "name": "flashinfer-bench-trace",
      "title": "FlashInfer-Bench trace harness"
    },
    "apiVersion": "kernelindex.dev/v1alpha1"
  },
  "environment": {
    "kind": "ExecutionEnvironment",
    "spec": {
      "hardware": {
        "vendor": "nvidia",
        "product": "NVIDIA B200",
        "architecture": "unknown"
      },
      "software": {
        "framework": {
          "name": "pytorch",
          "version": "2.8.0+cu128"
        },
        "libraries": {
          "triton": "3.4.0"
        },
        "cudaToolkit": "12.8"
      }
    },
    "metadata": {
      "name": "sol-trace-nvidia-b200-2-8-0-cu128-3-4-0-12-8",
      "title": "NVIDIA B200 (trace-reported software)"
    },
    "apiVersion": "kernelindex.dev/v1alpha1"
  }
}

Cite this record (permalink, digest, access date)

Report an issue with this run

Published 2026-08-20 · FlashInfer-Bench · Apache-2.0All results for GQA ragged prefill causal h32 kv8 d128 →JSON

gpt-o3 / cuda025ec1