gpt-o3 / triton4c17a1
MLA paged decode h16 ckv512 kpe64 ps1 · num_pages = 989669 · batch_size = 1 · len_indptr = 2 · num_kv_indices = 708 · bf16 · run 01a02032-5e3d…
●passedReported evidence
Not re-observed recently.
Primary measurement
129.3µsmean
Rank 2 in its comparison group · source-native comparison · observed 2025-10-16
13.40× reference
Identity
implementationgpt-o3_triton_4c17a1
projectgpt-o3
revisionda915083d4c7c5e61aa3005e3d17ae488e0fc71c
workloadnum_pages = 989669 · batch_size = 1 · len_indptr = 2 · num_kv_indices = 708 · bf16
comparison keysha256:4664497a5ee4aba4…
sourceFlashInfer-Bench
external idtrace/gpt-o3_triton_4c17a1/d232cde2-c7ff-4f74-aac4-f6c86d59c0b6
sha256:785bddbff51f1590e8abad9…
Correctness
comparatorflashinfer_bench_eval
max absolute error0.00390625
max relative error0.012263626791536808
resultpassed
Workload
num_pages989669
batch_size1
len_indptr2
num_kv_indices708
q_pebf16 [1, 16, 64]
q_nopebf16 [1, 16, 512]
ckv_cachebf16 [989669, 1, 512]
kpe_cachebf16 [989669, 1, 64]
kv_indptrint32 [2]
kv_indicesint32 [708]
definition comparatorsol_execbench_eval
Measurements
latency · mean129.3 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": 129279
},
"primaryStatistic": "mean"
},
"observedAt": "2025-10-16T11:05:26.358Z",
"correctness": {
"comparator": "sol_execbench_eval",
"maximumAbsoluteError": 0.00390625,
"maximumRelativeError": 0.012263626791536808
},
"sourceNative": {
"source": "flashinfer-bench",
"metrics": {
"latency_ms": 0.1292791808821686,
"speedup_factor": 13.400235150813963,
"reference_latency_ms": 1.7323714239256722
},
"benchmark": "mla_paged_decode_h16_ckv512_kpe64_ps1"
},
"protocolDigest": "sha256:7d44d206ad086b4bece35ce31f6531ebb11f07e0d9dcfb74c75be86d3ec02095",
"workloadDigest": "sha256:1d42537624e9d1e8809c10c24db7ec282b521437e4cbdb5d2e4e9b8937f68d1f",
"environmentDigest": "sha256:f4292a8be127b7230cd3995c1d605fb6c644e2c2f94566e1d28cc1be766ca5d5",
"implementationDigest": "sha256:3da822530f4925b23d07b281c971e733bdc1067be986982bc2e1bdbdbd7a38f0"
},
"metadata": {
"name": "sol-trace-mla-paged-decode-h16-ckv512-kpe64-ps1-d232cde2",
"title": "gpt-o3_triton_4c17a1 · mla_paged_decode_h16_ckv512_kpe64_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 MLA paged decode h16 ckv512 kpe64 ps1 →JSON