gpt-o3 / cuda5548d0
GQA paged decode h32 kv4 d128 ps1 · num_pages = 6510 · batch_size = 16 · len_indptr = 17 · num_kv_indices = 6460 · bf16 · run 01a02032-5c5b…
●passedReported evidence
Not re-observed recently.
Primary measurement
659.9µsmean
Rank 4 in its comparison group · source-native comparison · observed 2025-10-16
757.84× reference
Identity
implementationgpt-o3_cuda_5548d0
projectgpt-o3
revisionda915083d4c7c5e61aa3005e3d17ae488e0fc71c
workloadnum_pages = 6510 · batch_size = 16 · len_indptr = 17 · num_kv_indices = 6460 · bf16
comparison keysha256:16e995f1dc1b3acd…
sourceFlashInfer-Bench
external idtrace/gpt-o3_cuda_5548d0/7c888a78-1515-4418-bc81-49b593947860
sha256:a66e313aad01634f5fbaaf7…
Correctness
comparatorflashinfer_bench_eval
max absolute error0.0009765625
max relative error0.05279775708913803
resultpassed
Workload
num_pages6510
batch_size16
len_indptr17
num_kv_indices6460
qbf16 [16, 32, 128]
k_cachebf16 [6510, 1, 4, 128]
v_cachebf16 [6510, 1, 4, 128]
kv_indptrint32 [17]
kv_indicesint32 [6460]
definition comparatorsol_execbench_eval
Measurements
latency · mean659.9 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": 659868
},
"primaryStatistic": "mean"
},
"observedAt": "2025-10-16T01:47:53.673Z",
"correctness": {
"comparator": "sol_execbench_eval",
"maximumAbsoluteError": 0.0009765625,
"maximumRelativeError": 0.05279775708913803
},
"sourceNative": {
"source": "flashinfer-bench",
"metrics": {
"latency_ms": 0.6598680923242525,
"speedup_factor": 757.8377605672135,
"reference_latency_ms": 500.0729573567708
},
"benchmark": "gqa_paged_decode_h32_kv4_d128_ps1"
},
"protocolDigest": "sha256:7d44d206ad086b4bece35ce31f6531ebb11f07e0d9dcfb74c75be86d3ec02095",
"workloadDigest": "sha256:b7766f6062ab885d11d985ab1f5a32df326306b44e6e7a8dd2e2cdd066762af7",
"environmentDigest": "sha256:f4292a8be127b7230cd3995c1d605fb6c644e2c2f94566e1d28cc1be766ca5d5",
"implementationDigest": "sha256:9065668f65f0c70486702b4d7105d2a2da70e1aebd03314d68ffce9743ed081e"
},
"metadata": {
"name": "sol-trace-gqa-paged-decode-h32-kv4-d128-ps1-7c888a78",
"title": "gpt-o3_cuda_5548d0 · gqa_paged_decode_h32_kv4_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 kv4 d128 ps1 →JSON