PyTorch eager
Mamba2ReturnFinalState · d_head = 64 · n_heads = 8 · batch_size = 2048 · seq_length = 128 · fp32 · run 01a03f0e-f6dd…
●passedReported evidence
Not re-observed recently.
Primary measurement
5.94ms±0.00 · mean of 100
Rank 2 in its comparison group · source-native comparison · observed 2026-03-05
max ms 5.97 · min ms 5.92 · std ms 0.0103 · mean ms 5.94
Identity
implementationPyTorch eager
projectPyTorch
revisionunknown
workloadd_head = 64 · n_heads = 8 · batch_size = 2048 · seq_length = 128 · fp32
comparison keysha256:fa62813eb4ab2602…
sourceKernelBench baseline timings
external idH100_Modal/torch/level3/49_Mamba2ReturnFinalState.py
sha256:0b55b5f4fc1cab30c89fb0e…
Correctness
Marked passed by the source; the correctness policy was not published.
Workload
d_head64
n_heads8
batch_size2048
seq_length128
input_1fp32 [2048, 128, 8, 64]
definition comparatornot_asserted
Measurements
latency · max5.97 ms · n=100
latency · mean5.94 ms · n=100
latency · min5.92 ms · n=100
latency · std10.3 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": 5940000,
"maximum": 5970000,
"minimum": 5920000,
"confidence95": [
5937981,
5942019
]
},
"primaryStatistic": "mean"
},
"observedAt": "2026-03-05T08:38:15.000Z",
"measurements": [
{
"unit": "ns",
"value": 10300,
"metric": "latency",
"statistic": "std"
}
],
"sourceNative": {
"source": "kernelbench",
"metrics": {
"max_ms": 5.97,
"min_ms": 5.92,
"std_ms": 0.0103,
"mean_ms": 5.94
},
"benchmark": "H100_Modal/baseline_time_torch.json",
"externalId": "H100_Modal/torch/level3/49_Mamba2ReturnFinalState.py"
},
"protocolDigest": "sha256:547725b033a9911eae0c5ac362344a85aed824894ed4ab476af945b9bb491ea6",
"workloadDigest": "sha256:bab4ef5c86b034838b74468e3b6801f7d9c9d53660a76d51dda57cdea33526c7",
"environmentDigest": "sha256:2750ae381d6de582f1eb6b9008e0987463d19ede8af127d97d76c5ae09121393",
"implementationDigest": "sha256:68d837fe8aae6933eebb13607729d7f065d93f6361b8a6957be8960bcdc04e83"
},
"metadata": {
"name": "kernelbench-h100-modal-torch-l3-49-mamba2returnfinalstate",
"title": "Mamba2ReturnFinalState · PyTorch eager · Modal"
},
"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": "80GB HBM3",
"architecture": "sm_90"
},
"software": {}
},
"metadata": {
"name": "kernelbench-h100-modal",
"title": "KernelBench host · H100 80GB HBM3 (Modal)"
},
"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 Mamba2ReturnFinalState →JSON