Skip to content
KernelIndex
Search⌘K

submission 29478

KernelBot user 258 · python · License unknown

Use it

Vendorable · source mirrored · license unknownView source →

No package. Vendor the mirrored source: 1 lines, June 9 Researcher Reciprocity License v1.0.

submission.py
curl "https://kernelindex.com/api/v1/implementations/kernelbot-amd-mla-decode-29478?include=source"
interfacepython
Compatibility
measured onAMD Instinct MI300X
declared hardwareAMD Instinct MI300X
architecturesgfx942
dtypesbf16

Benchmark evidence

1 measurement across 1 GPU, fastest first.

Operation / workload
Hardware
Latency
Rank
Observed
AMD MLA decodebf16 · [128, 1, 7168]
AMD Instinct MI300X
21.1ms
#46 of 79
2025-05-31

Reported · How evidence levels are derived →

Source and license

sourceavailable
revision digestsha256:dab24bf4781434a4a2f9926de7f5cde57761c9ad0da5cf27d5ff3c1640b1eec2
license declaredunknown
license concludedunknown
imported2026-08-15

Kernel source

submission.py1 lines
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

Source code from GPU Mode and the KernelBot dataset · June 9 Researcher Reciprocity License v1.0

Best evidence level for this revision: reported

JSON