PyTorch eager
MinGPTCausalAttention · n_embd = 768 · seq_len = 512 · batch_size = 128 · fp32 · run 01a03f0e-f842…
●passedReported evidence
Not re-observed recently.
Primary measurement
19.7ms±0.11 · mean of 100
Rank 2 in its comparison group · source-native comparison · observed 2026-03-05
max ms 20.8 · min ms 18.7 · std ms 0.584 · mean ms 19.7
Identity
implementationPyTorch eager
projectPyTorch
revisionunknown
workloadn_embd = 768 · seq_len = 512 · batch_size = 128 · fp32
comparison keysha256:c830006d0cf643de…
sourceKernelBench baseline timings
external idH100_PCIe_LambdaLabs/torch/level3/43_MinGPTCausalAttention.py
sha256:7713106dc872e306bf717e0…
Correctness
Marked passed by the source; the correctness policy was not published.
Workload
n_embd768
seq_len512
batch_size128
input_1fp32 [128, 512, 768]
definition comparatornot_asserted
Measurements
latency · max20.8 ms · n=100
latency · mean19.7 ms · n=100
latency · min18.7 ms · n=100
latency · std584.0 µs
Protocol
harnessKernelBench timing scripts
timercuda_events
primaryStatisticmean
comparabilityFamilykernelbench_baseline_timing
Environment
gpuNVIDIA H100 (sm_90)
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": {
"samples": 100,
"latencyNs": {
"mean": 19700000,
"maximum": 20800000,
"minimum": 18700000,
"confidence95": [
19585536,
19814464
]
},
"primaryStatistic": "mean"
},
"observedAt": "2026-03-05T08:38:15.000Z",
"measurements": [
{
"unit": "ns",
"value": 584000,
"metric": "latency",
"statistic": "std"
}
],
"sourceNative": {
"source": "kernelbench",
"metrics": {
"max_ms": 20.8,
"min_ms": 18.7,
"std_ms": 0.584,
"mean_ms": 19.7
},
"benchmark": "H100_PCIe_LambdaLabs/baseline_time_torch.json",
"externalId": "H100_PCIe_LambdaLabs/torch/level3/43_MinGPTCausalAttention.py"
},
"protocolDigest": "sha256:547725b033a9911eae0c5ac362344a85aed824894ed4ab476af945b9bb491ea6",
"workloadDigest": "sha256:6a79750947048c57a6ddd061ec26b81fb4bb31c13c6dd22bbba3a8993577a877",
"environmentDigest": "sha256:89f9b28b06758f22a1df331fea58bfde2c9cc0f3627d1829509743ea538f0011",
"implementationDigest": "sha256:8342088fda08f4acb1d5f0f14a8517e416cfe395beb8d50d811e6f9cea7a281c"
},
"metadata": {
"name": "kernelbench-h100-pcie-lambdalabs-torch-l3-43-mingptcausalattention",
"title": "MinGPTCausalAttention · PyTorch eager · Lambda Labs"
},
"apiVersion": "kernelindex.dev/v1alpha1"
},
"protocol": {
"kind": "BenchmarkProtocol",
"spec": {
"harness": {
"name": "KernelBench timing scripts",
"repository": "https://github.com/ScalingIntelligence/KernelBench"
},
"measurement": {
"timer": "cuda_events",
"primaryStatistic": "mean"
},
"comparability": {
"notes": "Mean of 100 timed forward passes (CUDA events, warm-up excluded) of the reference module on fixed inputs; the 95% interval is the mean's, from the reported standard deviation. The upstream JSON records no torch or CUDA version. Comparable only within one problem and one timing host.",
"family": "kernelbench_baseline_timing"
}
},
"metadata": {
"name": "kernelbench-timing-v1",
"title": "KernelBench baseline timing"
},
"apiVersion": "kernelindex.dev/v1alpha1"
},
"environment": {
"kind": "ExecutionEnvironment",
"spec": {
"hardware": {
"vendor": "nvidia",
"product": "NVIDIA H100",
"formFactor": "PCIe",
"architecture": "sm_90"
},
"software": {}
},
"metadata": {
"name": "kernelbench-h100-pcie-lambdalabs",
"title": "KernelBench host · H100 PCIe (Lambda Labs)"
},
"apiVersion": "kernelindex.dev/v1alpha1"
}
}Cite this record (permalink, digest, access date)
Report an issue with this run
Published 2026-08-26 · KernelBench baseline timings · MITAll results for MinGPTCausalAttention →JSON