submission 754045
jkman2013 · python · License unknown
Use it
Vendorable · source mirrored · license unknownView source →
No package. Vendor the mirrored source: 33 lines, June 9 Researcher Reciprocity License v1.0.
submission_v2.py
curl "https://kernelindex.com/api/v1/implementations/kernelbot-amd-moe-mxfp4-754045?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:cbd9d9a488b94fe6b1e130b3f9ab80a938aa3e2e85ff6690849ac9ffe6774cfa
license declaredunknown
license concludedunknown
authorsjkman2013
imported2026-08-26
Techniques
Extracted from the mirrored source by pattern, never inferred. Each row cites its line.
fp4
Optimized MXFP4 MoE v2 for AMD MI355X.Kernel source
submission_v2.py33 lines
"""
Optimized MXFP4 MoE v2 for AMD MI355X.
- Module-level imports (avoid per-call overhead)
- torch.inference_mode() to disable autograd overhead
- Minimal Python overhead
"""
from task import input_t, output_t
import torch
from aiter import ActivationType, QuantType
from aiter.fused_moe import fused_moe
torch.set_grad_enabled(False)
@torch.inference_mode()
def custom_kernel(data: input_t) -> output_t:
(hidden_states, _, _, _, _,
gate_up_weight_shuffled, down_weight_shuffled,
gate_up_weight_scale_shuffled, down_weight_scale_shuffled,
topk_weights, topk_ids, config) = data
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=config["d_hidden_pad"] - config["d_hidden"],
intermediate_pad=config["d_expert_pad"] - config["d_expert"],
)
scrolls · 33 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