Run 01a02032 · KernelIndex

Primary measurement

29.0µsmean

Rank 4 in its comparison group · source-native comparison · observed 2025-10-16

Compare with #1 →

901.05× reference

Identity

implementationgemini-2.5-pro_triton_pr9imz

projectgemini-2.5-pro

revisionda915083d4c7c5e61aa3005e3d17ae488e0fc71c

workloadnum_pages = 78 · batch_size = 1 · len_indptr = 2 · num_kv_indices = 61 · bf16

comparison keysha256:8ccfaa6042c2da07…

sourceFlashInfer-Bench

external idtrace/gemini-2.5-pro_triton_pr9imz/2f08e877-5dff-443b-b146-0c7c2b9d5e1e

sha256:ac9a55e456df47526bbc550…

Correctness

comparatorflashinfer_bench_eval

max absolute error0.00048828125

max relative error0.016603540629148483

resultpassed

Workload

num_pages78

batch_size1

len_indptr2

num_kv_indices61

qbf16 [1, 32, 128]

k_cachebf16 [78, 1, 8, 128]

v_cachebf16 [78, 1, 8, 128]

kv_indptrint32 [2]

kv_indicesint32 [61]

definition comparatorsol_execbench_eval

Measurements

latency · mean29.0 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

No artifacts published with this run.

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": 28986
        },
        "primaryStatistic": "mean"
      },
      "observedAt": "2025-10-16T01:52:54.516Z",
      "correctness": {
        "comparator": "sol_execbench_eval",
        "maximumAbsoluteError": 0.00048828125,
        "maximumRelativeError": 0.016603540629148483
      },
      "sourceNative": {
        "source": "flashinfer-bench",
        "metrics": {
          "latency_ms": 0.028985706884264578,
          "speedup_factor": 901.0465346094534,
          "reference_latency_ms": 26.117470741271973
        },
        "benchmark": "gqa_paged_decode_h32_kv8_d128_ps1"
      },
      "protocolDigest": "sha256:7d44d206ad086b4bece35ce31f6531ebb11f07e0d9dcfb74c75be86d3ec02095",
      "workloadDigest": "sha256:a5a319f0ffdb92734cb5550f2ee0c773c7bcb21c391f3568d06d6ba38a9a6b4a",
      "environmentDigest": "sha256:f4292a8be127b7230cd3995c1d605fb6c644e2c2f94566e1d28cc1be766ca5d5",
      "implementationDigest": "sha256:e5d6283af3f6143bf93e3bc93f551732f18d50a9d53538cf7fead3802f18b330"
    },
    "metadata": {
      "name": "sol-trace-gqa-paged-decode-h32-kv8-d128-ps1-2f08e877",
      "title": "gemini-2.5-pro_triton_pr9imz · 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

gemini-2.5-pro / tritonpr9imz