Skip to content
KernelIndex
Search⌘K

submission 10955

KernelBot user 7 · 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.

amd-fp8-mm.py
curl "https://kernelindex.com/api/v1/implementations/kernelbot-amd-fp8-mm-10955?include=source"
interfacepython
Compatibility
measured onAMD Instinct MI300X
declared hardwareAMD Instinct MI300X
architecturesgfx942
dtypesfp32, fp8_e4m3fnuz

Benchmark evidence

18 measurements across 1 GPU, fastest first.

Operation / workload
Hardware
Latency
Rank
Observed
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [1024, 512]
AMD Instinct MI300X
229.8µs
#95 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [1024, 256]
AMD Instinct MI300X
242.7µs
#96 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [1024, 1536]
AMD Instinct MI300X
309.4µs
#95 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [1024, 7168] · n=512 · seed=6563
AMD Instinct MI300X
352.5µs
#86 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [1024, 7168] · n=576 · seed=12346
AMD Instinct MI300X
387.1µs
#89 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [6144, 512]
AMD Instinct MI300X
554.7µs
#96 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [6144, 256]
AMD Instinct MI300X
556.0µs
#96 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [1024, 2048]
AMD Instinct MI300X
680.7µs
#94 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [1024, 2304]
AMD Instinct MI300X
739.9µs
#94 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [1024, 7168] · n=1536 · seed=8135
AMD Instinct MI300X
744.6µs
#92 of 98
2025-04-28
Show all 18 measurements ›
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [6144, 1536]
AMD Instinct MI300X
1.01ms
#94 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [6144, 7168] · n=512 · seed=12341
AMD Instinct MI300X
1.33ms
#92 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [6144, 7168] · n=576 · seed=9863
AMD Instinct MI300X
1.49ms
#92 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [1024, 7168] · n=4608 · seed=7531
AMD Instinct MI300X
1.56ms
#94 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [6144, 7168] · n=1536 · seed=6543
AMD Instinct MI300X
2.53ms
#90 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [6144, 2048]
AMD Instinct MI300X
2.70ms
#89 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [6144, 2304]
AMD Instinct MI300X
3.05ms
#91 of 98
2025-04-28
AMD FP8 blockwise GEMMfp32/fp8_e4m3fnuz · [6144, 7168] · n=4608 · seed=65436
AMD Instinct MI300X
6.03ms
#88 of 98
2025-04-28

Reported · How evidence levels are derived →

Source and license

sourceavailable
revision digestsha256:2ffa434aa9624d4317cd3d1edaaf919964c32c896099415d824aec8483a8ae95
license declaredunknown
license concludedunknown
imported2026-08-15

Kernel source

amd-fp8-mm.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