submission 527935
DizzleRama · python · License unknown
Use it
Vendorable · source mirrored · license unknownView source →
No package. Vendor the mirrored source: 25 lines, June 9 Researcher Reciprocity License v1.0.
tester_v2.py
curl "https://kernelindex.com/api/v1/implementations/kernelbot-amd-mxfp4-mm-527935?include=source"interfacepython
Compatibility
measured onAMD Instinct MI355X
declared hardwareAMD Instinct MI355X
architecturesgfx950
dtypesbf16, mxfp4
Benchmark evidence
1 measurement across 1 GPU, fastest first.
Operation / workload
Hardware
Latency
Rank
Observed
Reported · How evidence levels are derived →
Source and license
sourceavailable
revision digestsha256:f12c7f6377dc856959004cb653900bd6ce81693faa64672b49e8e80c47a3d09a
license declaredunknown
license concludedunknown
authorsDizzleRama
imported2026-08-26
Techniques
Extracted from the mirrored source by pattern, never inferred. Each row cites its line.
fp4
MXFP4 GEMM: bf16 A, MXFP4 B -> MXFP4 per-1x32 quant A -> gemm_a4w4 -> bf16 C.Kernel source
tester_v2.py25 lines
"""
MXFP4 GEMM: bf16 A, MXFP4 B -> MXFP4 per-1x32 quant A -> gemm_a4w4 -> bf16 C.
"""
import aiter
from aiter import QuantType, dtypes
from task import input_t, output_t
QKV_DTYPE = "mxfp4"
_quant_func = aiter.get_triton_quant(QuantType.per_1x32)
def custom_kernel(data: input_t) -> output_t:
A, B, B_q, B_shuffle, B_scale_sh = data
A_q, A_scale_sh = _quant_func(A.contiguous(), shuffle=True)
return aiter.gemm_a4w4(
A_q,
B_shuffle,
A_scale_sh,
B_scale_sh,
dtype=dtypes.bf16,
bpreshuffle=True,
)Source code from GPU Mode and the KernelBot dataset · June 9 Researcher Reciprocity License v1.0
Best evidence level for this revision: reported
JSON