gpt-5 / cuda21ea96
GQA paged prefill causal h32 kv8 d128 ps1 · total_q = 1 · num_pages = 51 · len_indptr = 2 · num_kv_indices = 34 · bf16 · run 01a02032-594f…
●passedReported evidence
Not re-observed recently.
Primary measurement
100.7µsmean
Rank 1 in its comparison group · source-native comparison · observed 2025-10-21
337.09× reference
Identity
implementationgpt-5_cuda_21ea96
projectgpt-5-2025-08-07
revisionda915083d4c7c5e61aa3005e3d17ae488e0fc71c
workloadtotal_q = 1 · num_pages = 51 · len_indptr = 2 · num_kv_indices = 34 · bf16
comparison keysha256:536ee9af822e2a3c…
sourceFlashInfer-Bench
external idtrace/gpt-5_cuda_21ea96/25f2945c-c282-47a9-ac12-bbc702a218f4
sha256:73ac4584842b0dbec9dc910…
Correctness
comparatorflashinfer_bench_eval
max absolute error9.5367431640625e-7
max relative error1.8185296823958197e-7
resultpassed
Workload
total_q1
num_pages51
len_indptr2
num_kv_indices34
qbf16 [1, 32, 128]
k_cachebf16 [51, 1, 8, 128]
v_cachebf16 [51, 1, 8, 128]
kv_indptrint32 [2]
qo_indptrint32 [2]
kv_indicesint32 [34]
definition comparatorsol_execbench_eval
Measurements
latency · mean100.7 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 manifestHide manifest
{
"run": {
"kind": "BenchmarkRun",
"spec": {
"status": "passed",
"timing": {
"latencyNs": {
"mean": 100676
},
"primaryStatistic": "mean"
},
"observedAt": "2025-10-21T13:55:22.722Z",
"correctness": {
"comparator": "sol_execbench_eval",
"maximumAbsoluteError": 9.5367431640625e-7,
"maximumRelativeError": 1.8185296823958197e-7
},
"sourceNative": {
"source": "flashinfer-bench",
"metrics": {
"latency_ms": 0.10067559422485872,
"speedup_factor": 337.0853367442845,
"reference_latency_ms": 33.936266581217446
},
"benchmark": "gqa_paged_prefill_causal_h32_kv8_d128_ps1"
},
"protocolDigest": "sha256:7d44d206ad086b4bece35ce31f6531ebb11f07e0d9dcfb74c75be86d3ec02095",
"workloadDigest": "sha256:b730c1e1fe38f60090c1736917527e9afde346fbd0a9c83b959fc033f68082e2",
"environmentDigest": "sha256:f4292a8be127b7230cd3995c1d605fb6c644e2c2f94566e1d28cc1be766ca5d5",
"implementationDigest": "sha256:5500be4af9df644c8dba7c0f20575cc4c96f5794df6e41c033b76276dc5353a7"
},
"metadata": {
"name": "sol-trace-gqa-paged-prefill-causal-h32-kv8-d128-ps1-25f2945c",
"title": "gpt-5_cuda_21ea96 · gqa_paged_prefill_causal_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 prefill causal h32 kv8 d128 ps1 →JSON