submission 539742
viani · python · License unknown
Use it
Vendorable · source mirrored · license unknownView source →
No package. Vendor the mirrored source: 16 lines, June 9 Researcher Reciprocity License v1.0.
baseline_ref.py
curl "https://kernelindex.com/api/v1/implementations/kernelbot-amd-moe-mxfp4-539742?include=source"interfacepython
Compatibility
measured onAMD Instinct MI355X
declared hardwareAMD Instinct MI355X
architecturesgfx950
dtypesbf16, fp32, fp8_e8m0, int32, 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 digestsha256:52ec7ef61bf295fe99317099cdd41fdad2dbc680e611fcbc016b93833715b8ae
license declaredunknown
license concludedunknown
authorsviani
imported2026-08-26
Techniques
Extracted from the mirrored source by pattern, never inferred. Each row cites its line.
fp4
MoE MXFP4 baseline: direct ref_kernel call.Kernel source
baseline_ref.py16 lines
#!POPCORN leaderboard amd-moe-mxfp4
#!POPCORN gpu MI355X
"""
MoE MXFP4 baseline: direct ref_kernel call.
Establishes the floor performance — same as aiter fused_moe reference.
RESULT: PENDING
"""
import torch
from task import input_t, output_t
from reference import ref_kernel
def custom_kernel(data: input_t) -> output_t:
return ref_kernel(data)
Source code from GPU Mode and the KernelBot dataset · June 9 Researcher Reciprocity License v1.0
Best evidence level for this revision: reported
JSON