Run 01a02032 · KernelIndex

Primary measurement

353.0µsmean

Rank 5 in its comparison group · source-native comparison · observed 2025-10-16

Compare with #1 →

92.38× reference

Identity

implementationgpt-o3_cuda_5548d0

projectgpt-o3

revisionda915083d4c7c5e61aa3005e3d17ae488e0fc71c

workloadnum_pages = 412 · batch_size = 1 · len_indptr = 2 · num_kv_indices = 362 · bf16

comparison keysha256:2b2fc4c48f98b72d…

sourceFlashInfer-Bench

external idtrace/gpt-o3_cuda_5548d0/27ef153a-4d10-402a-998b-1922904e94b2

sha256:215ac4f9a26728e69bfeb46…

Correctness

comparatorflashinfer_bench_eval

max absolute error0.000030517578125

max relative error0.006794965825974941

resultpassed

Workload

num_pages412

batch_size1

len_indptr2

num_kv_indices362

qbf16 [1, 32, 128]

k_cachebf16 [412, 1, 4, 128]

v_cachebf16 [412, 1, 4, 128]

kv_indptrint32 [2]

kv_indicesint32 [362]

definition comparatorsol_execbench_eval

Measurements

latency · mean353.0 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

logssha256:7e90ce0aebbdc3c5… · 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": 353049
        },
        "primaryStatistic": "mean"
      },
      "evidence": {
        "logs": {
          "uri": "kernelindex:artifact/sha256:7e90ce0aebbdc3c556c70bd8bc23755a6ced33f9edacfdd06be74d2c2515a872",
          "digest": "sha256:7e90ce0aebbdc3c556c70bd8bc23755a6ced33f9edacfdd06be74d2c2515a872"
        }
      },
      "observedAt": "2025-10-16T01:25:11.078Z",
      "correctness": {
        "comparator": "sol_execbench_eval",
        "maximumAbsoluteError": 0.000030517578125,
        "maximumRelativeError": 0.006794965825974941
      },
      "sourceNative": {
        "source": "flashinfer-bench",
        "metrics": {
          "latency_ms": 0.35304868927719246,
          "speedup_factor": 92.37747909899709,
          "reference_latency_ms": 32.613747914632164
        },
        "benchmark": "gqa_paged_decode_h32_kv4_d128_ps1"
      },
      "protocolDigest": "sha256:7d44d206ad086b4bece35ce31f6531ebb11f07e0d9dcfb74c75be86d3ec02095",
      "workloadDigest": "sha256:fcf16458148fc0adb1372367ac83b468de3af7a4009854a2980b9096ec1e0088",
      "environmentDigest": "sha256:f4292a8be127b7230cd3995c1d605fb6c644e2c2f94566e1d28cc1be766ca5d5",
      "implementationDigest": "sha256:9065668f65f0c70486702b4d7105d2a2da70e1aebd03314d68ffce9743ed081e"
    },
    "metadata": {
      "name": "sol-trace-gqa-paged-decode-h32-kv4-d128-ps1-27ef153a",
      "title": "gpt-o3_cuda_5548d0 · gqa_paged_decode_h32_kv4_d128_ps1"
    },
    "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 paged decode h32 kv4 d128 ps1 →JSON

gpt-o3 / cuda5548d0