Run 01a02032 · KernelIndex

Primary measurement

8.19µsmean

Rank 1 in its comparison group · source-native comparison · observed 2025-10-16

3123.68× reference

Identity

implementationgpt-5_cuda_95c7fe

projectgpt-5-2025-08-07

revisionda915083d4c7c5e61aa3005e3d17ae488e0fc71c

workloadnum_pages = 18 · batch_size = 1 · len_indptr = 2 · num_kv_indices = 2 · bf16

comparison keysha256:dd656cddb52671cc…

sourceFlashInfer-Bench

external idtrace/gpt-5_cuda_95c7fe/be0e9c07-d051-424b-ad4b-eac7cadf8a8e

sha256:3c00a06c060d0251a0fd2e1…

Correctness

comparatorflashinfer_bench_eval

max absolute error0.000003814697265625

max relative error0.006289204582571983

resultpassed

Workload

num_pages18

batch_size1

len_indptr2

num_kv_indices2

qbf16 [1, 32, 128]

k_cachebf16 [18, 1, 8, 128]

v_cachebf16 [18, 1, 8, 128]

kv_indptrint32 [2]

kv_indicesint32 [2]

definition comparatorsol_execbench_eval

Measurements

latency · mean8.19 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

logssha256:93a00f2b5a2c63cd… · 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": 8193
        },
        "primaryStatistic": "mean"
      },
      "evidence": {
        "logs": {
          "uri": "kernelindex:artifact/sha256:93a00f2b5a2c63cda7323a9db728f086d3d56c1ca5619cb8c002b87217a887ec",
          "digest": "sha256:93a00f2b5a2c63cda7323a9db728f086d3d56c1ca5619cb8c002b87217a887ec"
        }
      },
      "observedAt": "2025-10-16T01:51:01.208Z",
      "correctness": {
        "comparator": "sol_execbench_eval",
        "maximumAbsoluteError": 0.000003814697265625,
        "maximumRelativeError": 0.006289204582571983
      },
      "sourceNative": {
        "source": "flashinfer-bench",
        "metrics": {
          "latency_ms": 0.008193386637591662,
          "speedup_factor": 3123.678327651907,
          "reference_latency_ms": 25.593504269917805
        },
        "benchmark": "gqa_paged_decode_h32_kv8_d128_ps1"
      },
      "protocolDigest": "sha256:7d44d206ad086b4bece35ce31f6531ebb11f07e0d9dcfb74c75be86d3ec02095",
      "workloadDigest": "sha256:5cc6697b077422a1f59d6dbdf012434eb91d6755b2f97e293060f3397f8e58ef",
      "environmentDigest": "sha256:f4292a8be127b7230cd3995c1d605fb6c644e2c2f94566e1d28cc1be766ca5d5",
      "implementationDigest": "sha256:14185d3afa8e6a2d18582f2f64a63f0d5843645303df60306ca2020142cba687"
    },
    "metadata": {
      "name": "sol-trace-gqa-paged-decode-h32-kv8-d128-ps1-be0e9c07",
      "title": "gpt-5_cuda_95c7fe · gqa_paged_decode_h32_kv8_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 kv8 d128 ps1 →JSON

gpt-5 / cuda95c7fe