Liger
SwiGLU MLP · seq = 8192 · batch = 4 · hidden = 4096 · intermediate = 11008 · bf16 · run 01a016ae-5c60…
●passedReported evidence
Not re-observed recently.
Primary measurement
35.9msmedian
Rank 1 in its comparison group · source-native comparison · observed 2024-09-03
ms 20 35.930015563964844 · ms 50 35.930015563964844 · ms 80 35.930015563964844
Identity
implementationLiger
projectLiger-Kernel
revisionunknown
workloadseq = 8192 · batch = 4 · hidden = 4096 · intermediate = 11008 · bf16
comparison keysha256:9a932637cecd4f27…
sourceLiger-Kernel benchmarks
external idswiglu/forward/liger/nvidia-a100-sxm4-80gb/batch4-hidden4096-intermediate11008-seq8192-bf16/2024-09-03-15-40-24
sha256:1213b586b04baf9f6eefdb9…
Correctness
Marked passed by the source; the correctness policy was not published.
Workload
seq8192
batch4
hidden4096
intermediate11008
xbf16 [4, 8192, 4096]
w_upbf16 [11008, 4096]
w_downbf16 [4096, 11008]
w_gatebf16 [11008, 4096]
definition comparatornot_asserted
Measurements
latency · median35.9 ms
latency · p2035.9 ms
latency · p8035.9 ms
Protocol
harnessLiger-Kernel benchmark scripts
timercuda_events
primaryStatisticmedian
comparabilityFamilyliger_kernel_bench
Environment
gpuNVIDIA A100 (sm_80)
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": {
"median": 35930016
},
"primaryStatistic": "median"
},
"observedAt": "2024-09-03T15:40:24.000Z",
"measurements": [
{
"unit": "ns",
"value": 35930016,
"metric": "latency",
"statistic": "p20"
},
{
"unit": "ns",
"value": 35930016,
"metric": "latency",
"statistic": "p80"
}
],
"sourceNative": {
"source": "liger-kernel-bench",
"metrics": {
"ms_20": 35.930015563964844,
"ms_50": 35.930015563964844,
"ms_80": 35.930015563964844
},
"benchmark": "swiglu/forward",
"externalId": "swiglu/forward/liger/nvidia-a100-sxm4-80gb/batch4-hidden4096-intermediate11008-seq8192-bf16/2024-09-03-15-40-24"
},
"protocolDigest": "sha256:a79eaf4df66c97c6adc1fb2d09df78682155d1ea86652ae0b70199875bde7747",
"workloadDigest": "sha256:a9d199825fe361c51ad63ca08179ab443ae1810f268ddec2ece635bf5454afc9",
"environmentDigest": "sha256:90c9e2764222f592477cfccf3373519add5a1b881563956ad058c4f99e41b6b9",
"implementationDigest": "sha256:70a682a3136a79d191bc46b7ccb18a8047564299ed31429e872c4b24efaa861f"
},
"metadata": {
"name": "liger-swiglu-forward-liger-nvidia-a100-sxm4-80gb-batch4-hidden4096-intermediate11008-seq8192-bf16-20",
"title": "SwiGLU MLP · liger · forward",
"labels": {
"liger_version": "0.2.1"
}
},
"apiVersion": "kernelindex.dev/v1alpha1"
},
"protocol": {
"kind": "BenchmarkProtocol",
"spec": {
"harness": {
"name": "Liger-Kernel benchmark scripts",
"repository": "https://github.com/linkedin/Liger-Kernel"
},
"measurement": {
"timer": "cuda_events",
"primaryStatistic": "median"
},
"comparability": {
"notes": "Timed: the module forward pass, via triton.testing.do_bench with quantiles 0.5/0.2/0.8 (median with p20/p80 spread). The upstream CSV records no CUDA, driver, or torch version; the Liger release rides each run's labels. Comparable only within one kernel, workload, timed pass, and GPU.",
"family": "liger_kernel_bench"
}
},
"metadata": {
"name": "liger-bench-do-bench-forward",
"title": "Liger benchmark harness · forward pass"
},
"apiVersion": "kernelindex.dev/v1alpha1"
},
"environment": {
"kind": "ExecutionEnvironment",
"spec": {
"hardware": {
"vendor": "nvidia",
"product": "NVIDIA A100",
"formFactor": "SXM4-80GB",
"architecture": "sm_80"
},
"software": {}
},
"metadata": {
"name": "liger-bench-nvidia-a100-sxm4-80gb",
"title": "Liger benchmark host · A100-SXM4-80GB"
},
"apiVersion": "kernelindex.dev/v1alpha1"
}
}Cite this record (permalink, digest, access date)
Report an issue with this run
Published 2026-08-18 · Liger-Kernel benchmarks · BSD-2-ClauseAll results for SwiGLU MLP →JSON