Run 01a02032 · KernelIndex

Primary measurement

3.09msmean

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

3.79× reference

Identity

implementationgpt-5-2025-08-07_cuda_a2d8ca

projectgpt-5-2025-08-07

revisionda915083d4c7c5e61aa3005e3d17ae488e0fc71c

workloadseq_len = 7 · bf16/fp32/fp8_e4m3

comparison keysha256:50b4ca648f11e029…

sourceFlashInfer-Bench

external idtrace/gpt-5-2025-08-07_cuda_a2d8ca/b8f4f012-a32e-4356-b4e1-7665b3d598af

sha256:bed65b694af90d8d515f02c…

Correctness

comparatorflashinfer_bench_eval

max absolute error0

max relative error0

resultpassed

Workload

seq_len7

routing_biasbf16 [256]

gemm1_weightsfp8_e4m3 [32, 4096, 7168]

gemm2_weightsfp8_e4m3 [32, 7168, 2048]

hidden_statesfp8_e4m3 [7, 7168]

routing_logitsfp32 [7, 256]

gemm1_weights_scalefp32 [32, 32, 56]

gemm2_weights_scalefp32 [32, 56, 16]

hidden_states_scalefp32 [56, 7]

definition comparatorsol_execbench_eval

Measurements

latency · mean3.09 ms

Protocol

harnessflashinfer-bench

timerharness_reported

primaryStatisticmean

correctnessReferencedefinition reference implementation

comparabilityFamilyflashinfer_bench_trace

Environment

gpuNVIDIA B200 (unknown)

cudaToolkit12.8

frameworkpytorch 2.8.0+cu128

Artifacts

logssha256:5ec3cfade09445f1… · 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": 3086342
        },
        "primaryStatistic": "mean"
      },
      "evidence": {
        "logs": {
          "uri": "kernelindex:artifact/sha256:5ec3cfade09445f1d1c249f6ca2e1d58752894a9a3961b318404d1a1181632b9",
          "digest": "sha256:5ec3cfade09445f1d1c249f6ca2e1d58752894a9a3961b318404d1a1181632b9"
        }
      },
      "observedAt": "2025-10-16T12:02:39.919Z",
      "correctness": {
        "comparator": "sol_execbench_eval",
        "maximumAbsoluteError": 0,
        "maximumRelativeError": 0
      },
      "sourceNative": {
        "source": "flashinfer-bench",
        "metrics": {
          "latency_ms": 3.086341915177364,
          "speedup_factor": 3.790492969378048,
          "reference_latency_ms": 11.698757330576578
        },
        "benchmark": "moe_fp8_block_scale_ds_routing_topk8_ng8_kg4_e32_h7168_i2048"
      },
      "protocolDigest": "sha256:7d44d206ad086b4bece35ce31f6531ebb11f07e0d9dcfb74c75be86d3ec02095",
      "workloadDigest": "sha256:cbd281eef86a9d9b65e0361b7d5881e427629233d864b3ca5060edb3f85eb3c9",
      "environmentDigest": "sha256:f4292a8be127b7230cd3995c1d605fb6c644e2c2f94566e1d28cc1be766ca5d5",
      "implementationDigest": "sha256:17a59015a553f958aa6ed224d10af90a5700205fe68519dbeaf081a7ebffb832"
    },
    "metadata": {
      "name": "sol-trace-moe-fp8-block-scale-ds-routing-topk8-ng8-kg4-e32-h7168-i2048-b8f4f012",
      "title": "gpt-5-2025-08-07_cuda_a2d8ca · moe_fp8_block_scale_ds_routing_topk8_ng8_kg4_e32_h7168_i2048"
    },
    "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 MoE FP8 block scale ds routing topk8 ng8 kg4 e32 h7168 i2048 →JSON

gpt-5-2025-08-07 / cudaa2d8ca