Skip to content
KernelIndex
Search⌘K

submission 604649

Kaushik J · python · License unknown

Use it

Vendorable · source mirrored · license unknownView source →

No package. Vendor the mirrored source: 54 lines, June 9 Researcher Reciprocity License v1.0.

submission.py
curl "https://kernelindex.com/api/v1/implementations/kernelbot-amd-moe-mxfp4-604649?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
AMD MXFP4 MoEsuite of 7 cases
AMD Instinct MI355X
185.3µs
#617 of 782
2026-03-21

Reported · How evidence levels are derived →

Source and license

sourceavailable
revision digestsha256:44bd275cbaa6e33cc109d6ecc1545d58efc938da1018a427f584a0f6f692ef2b
license declaredunknown
license concludedunknown
authorsKaushik J
imported2026-08-26

Kernel source

submission.py54 lines
# Write your code 
# Write your code here
#!POPCORN leaderboard amd-moe-mxfp4
#!POPCORN gpu MI355X

from task import input_t, output_t
from aiter import ActivationType, QuantType
from aiter.fused_moe import fused_moe


def custom_kernel(data: input_t) -> output_t:
    (
        hidden_states,
        gate_up_weight,
        down_weight,
        gate_up_weight_scale,
        down_weight_scale,
        gate_up_weight_shuffled,
        down_weight_shuffled,
        gate_up_weight_scale_shuffled,
        down_weight_scale_shuffled,
        topk_weights,
        topk_ids,
        config,
    ) = data

    hidden_pad = config["d_hidden_pad"] - config["d_hidden"]
    intermediate_pad = config["d_expert_pad"] - config["d_expert"]

    return fused_moe(
        hidden_states,
        gate_up_weight_shuffled,
        down_weight_shuffled,
        topk_weights,
        topk_ids,
        expert_mask=None,
        activation=ActivationType.Silu,
        quant_type=QuantType.per_1x32,
        doweight_stage1=False,
        w1_scale=gate_up_weight_scale_shuffled,
        w2_scale=down_weight_scale_shuffled,
        a1_scale=None,
        a2_scale=None,
        hidden_pad=hidden_pad,
        intermediate_pad=intermediate_pad,
    )


try:
    from utils import make_match_reference
    from reference import ref_kernel
    check_implementation = make_match_reference(ref_kernel, rtol=2e-2, atol=2e-2)
except Exception:
    check_implementation = None
scrolls · 54 lines total

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

Best evidence level for this revision: reported

JSON