Run 01a00b1e · KernelIndex

Primary measurement

41.4µsmean

Rank 1 in its comparison group · source-native comparison · observed 2026-04-12

16.55× reference

Identity

implementationflashinfer_wrapper_fe59e7

projectFlashInfer-Bench baselines

revisionda915083d4c7c5e61aa3005e3d17ae488e0fc71c

workloadtotal_q = 1 · total_kv = 1 · len_indptr = 2 · bf16

comparison keysha256:2759c4f2702f4916…

sourceFlashInfer-Bench

external idtrace/flashinfer_wrapper_fe59e7/6da20eee-a756-4633-97d4-a5922cba2dae

sha256:3376fa5cb2e35e1904daa33…

Correctness

comparatorflashinfer_bench_eval

max absolute error9.5367431640625e-7

max relative error0.000002363636895097443

resultpassed

Workload

total_q1

total_kv1

len_indptr2

kbf16 [1, 8, 128]

qbf16 [1, 24, 128]

vbf16 [1, 8, 128]

kv_indptrint32 [2]

qo_indptrint32 [2]

definition comparatorsol_execbench_eval

Measurements

latency · mean41.4 µs

Protocol

harnessflashinfer-bench

timerharness_reported

primaryStatisticmean

correctnessReferencedefinition reference implementation

comparabilityFamilyflashinfer_bench_trace

Environment

gpuNVIDIA GB200 (unknown)

cudaToolkit13.0

frameworkpytorch 2.9.1+cu130

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": 41371
        },
        "primaryStatistic": "mean"
      },
      "observedAt": "2026-04-12T13:17:39.231Z",
      "correctness": {
        "comparator": "sol_execbench_eval",
        "maximumAbsoluteError": 9.5367431640625e-7,
        "maximumRelativeError": 0.000002363636895097443
      },
      "sourceNative": {
        "source": "flashinfer-bench",
        "metrics": {
          "latency_ms": 0.0413706668963035,
          "speedup_factor": 16.552017470612483,
          "reference_latency_ms": 0.684768001238505
        },
        "benchmark": "gqa_ragged_prefill_causal_h24_kv8_d128"
      },
      "protocolDigest": "sha256:7d44d206ad086b4bece35ce31f6531ebb11f07e0d9dcfb74c75be86d3ec02095",
      "workloadDigest": "sha256:7bf6d33c0b8bd0f094126d194b6dfd1056c3d47d0982b5ccec696723e0ec4f92",
      "environmentDigest": "sha256:1766500d1b1ecc9787a211f3c98e76bf311671f443544a305252cb4d9564be8b",
      "implementationDigest": "sha256:1ae32bdf663d64c4ba87676614658cc570a1530f315eee8d5c4f3faeb9f2b080"
    },
    "metadata": {
      "name": "sol-trace-gqa-ragged-prefill-causal-h24-kv8-d128-6da20eee",
      "title": "flashinfer_wrapper_fe59e7 · gqa_ragged_prefill_causal_h24_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 GB200",
        "architecture": "unknown"
      },
      "software": {
        "framework": {
          "name": "pytorch",
          "version": "2.9.1+cu130"
        },
        "libraries": {
          "triton": "3.3.0"
        },
        "cudaToolkit": "13.0"
      }
    },
    "metadata": {
      "name": "sol-trace-nvidia-gb200-2-9-1-cu130-3-3-0-13-0",
      "title": "NVIDIA GB200 (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 ragged prefill causal h24 kv8 d128 →JSON

flashinfer / wrapperfe59e7