Skip to content
KernelIndex
Search⌘K

submission 607187

ftyghome · python · License unknown

Use it

Vendorable · source mirrored · license unknownView source →

No package. Vendor the mirrored source: 27 lines, June 9 Researcher Reciprocity License v1.0.

submission.py
curl "https://kernelindex.com/api/v1/implementations/kernelbot-amd-mxfp4-mm-607187?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
AMD MXFP4 GEMMsuite of 6 cases
AMD Instinct MI355X
9.37µs
#175 of 1143
2026-03-22

Reported · How evidence levels are derived →

Source and license

sourceavailable
revision digestsha256:c710eeb8dcabcd0f87badd64883e61a83ff829ce32c417e169eceea3f7f5cc5e
license declaredunknown
license concludedunknown
authorsftyghome
imported2026-08-15

Kernel source

submission.py27 lines
#!POPCORN leaderboard amd-mxfp4-mm
# This file is generated by gen_submission.py. DO NOT EDIT.
import os, zlib, base64, torch
os.environ.setdefault("HSA_XNACK", "0")
os.environ["PYTORCH_ROCM_ARCH"] = "gfx950:xnack-"
from torch.utils.cpp_extension import load_inline
from task import input_t, output_t

CUDA_SRC = zlib.decompress(base64.b64decode(b'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')).decode("utf-8")

_mod = load_inline(
    name="mxfp4_gemm_submission",
    cpp_sources=[""],
    cuda_sources=[CUDA_SRC],
    verbose=False,
    extra_cflags=["-O3", "-std=c++17"],
    extra_cuda_cflags=[
        "-std=c++17", "-O3", "-ffast-math", "-ffp-contract=fast",
        "-funroll-loops", "-DSUBMISSION_BUILD", "-DGEMM6_SWIZZLE=0",
        "-U__HIP_NO_HALF_OPERATORS__", "-U__HIP_NO_HALF_CONVERSIONS__",
    ],
)
_quant_gemm = _mod.quant_gemm

def custom_kernel(data: input_t) -> output_t:
    A, _, _, B_shuffle, B_scale_sh = data
    return _quant_gemm(A, B_shuffle.view(torch.uint8), B_scale_sh.view(torch.uint8))

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

Best evidence level for this revision: reported

JSON