Run 01a00b1d · KernelIndex

Primary measurement

16.4msmean

Rank 1 in its comparison group · source-native comparison · observed 2026-03-31

2.11× reference

Identity

implementationflashinfer_wrapper_1b7890

projectFlashInfer-Bench baselines

revisionda915083d4c7c5e61aa3005e3d17ae488e0fc71c

workloadnum_pages = 19713 · batch_size = 16 · len_indptr = 17 · num_kv_indices = 16593 · bf16

comparison keysha256:3d98083e8ee8052a…

sourceFlashInfer-Bench

external idtrace/flashinfer_wrapper_1b7890/6c8a7a2b-8656-4808-a33b-7a6ab285e935

sha256:b28f765e3b32dd34d163006…

Correctness

comparatorflashinfer_bench_eval

max absolute error0.0001220703125

max relative error629.3129272460938

resultpassed

Workload

num_pages19713

batch_size16

len_indptr17

num_kv_indices16593

qbf16 [16, 24, 128]

k_cachebf16 [19713, 64, 4, 128]

v_cachebf16 [19713, 64, 4, 128]

kv_indptrint32 [17]

kv_indicesint32 [16593]

kv_last_page_lenint32 [16]

definition comparatorsol_execbench_eval

Measurements

latency · mean16.4 ms

Protocol

harnessflashinfer-bench

timerharness_reported

primaryStatisticmean

correctnessReferencedefinition reference implementation

comparabilityFamilyflashinfer_bench_trace

Environment

gpuNVIDIA B200 (unknown)

cudaToolkit12.8

frameworkpytorch 2.9.1+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": 16377816
        },
        "primaryStatistic": "mean"
      },
      "observedAt": "2026-03-31T14:30:44.604Z",
      "correctness": {
        "comparator": "sol_execbench_eval",
        "maximumAbsoluteError": 0.0001220703125,
        "maximumRelativeError": 629.3129272460938
      },
      "sourceNative": {
        "source": "flashinfer-bench",
        "metrics": {
          "latency_ms": 16.377816166666666,
          "speedup_factor": 2.1087538054651307,
          "reference_latency_ms": 34.53678216666667
        },
        "benchmark": "gqa_paged_decode_h24_kv4_d128_ps64"
      },
      "protocolDigest": "sha256:7d44d206ad086b4bece35ce31f6531ebb11f07e0d9dcfb74c75be86d3ec02095",
      "workloadDigest": "sha256:3048531e8c93be8d187ec796fd3b604ae58015c907eb592c3da5f44b08f80a3e",
      "environmentDigest": "sha256:af7cb68a0bb0eb9671092a620668c2600f0431301fc46a71807ea75fb7de8bce",
      "implementationDigest": "sha256:f93c7e9cdbce77603e6624341de86495d8c3315e276228537ef3fd79596f0cbc"
    },
    "metadata": {
      "name": "sol-trace-gqa-paged-decode-h24-kv4-d128-ps64-6c8a7a2b",
      "title": "flashinfer_wrapper_1b7890 · gqa_paged_decode_h24_kv4_d128_ps64"
    },
    "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": "2.9.1+cu128"
        },
        "libraries": {
          "triton": "3.5.1"
        },
        "cudaToolkit": "12.8"
      }
    },
    "metadata": {
      "name": "sol-trace-nvidia-b200-2-9-1-cu128-3-5-1-12-8",
      "title": "NVIDIA B200 (trace-reported software)"
    },
    "apiVersion": "kernelindex.dev/v1alpha1"
  }
}

Cite this record (permalink, digest, access date)

Report an issue with this run

Published 2026-08-16 · FlashInfer-Bench · Apache-2.0All results for GQA paged decode h24 kv4 d128 ps64 →JSON

flashinfer / wrapper1b7890