mm_fp4_mxfp4_flashinfer_n4096_k4096
FlashInfer-Bench baselines · python · Apache-2.0
Use it
Vendorable · source mirrored · Apache-2.0View source →
No package. Vendor the mirrored source: 14 lines, Apache-2.0, pinned at da91508.
main.py
curl "https://kernelindex.com/api/v1/implementations/flashinfer-mm-fp4-mxfp4-flashinfer-n4096-k4096?include=source"interfacepython
revisionda915083d4c7
symbolrun
pathmain.py
Compatibility
measured onNVIDIA B200
declared hardwareNVIDIA B200
architecturessm_100
dtypesbf16
Benchmark evidence
6 measurements across 1 GPU, fastest first.
Operation / workload
Hardware
Latency
Rank
Observed
Reproduction-ready · How evidence levels are derived →
Source and license
sourcehttps://huggingface.co/datasets/flashinfer-ai/flashinfer-trace
commitda915083d4c7c5e61aa3005e3d17ae488e0fc71c
revision digestsha256:dc3b3259a816febab5f32b6f07a4a36df5cc0ddb6951176e8df22bcdd9e68288
license declaredApache-2.0
license concludedApache-2.0
authorsbaseline
imported2026-08-16
Kernel source
main.py14 lines
import torch
import flashinfer
def run(A, B):
a_fp4, a_sf = flashinfer.mxfp4_quantize(A)
b_fp4, b_sf = flashinfer.mxfp4_quantize(B)
out = torch.empty(A.shape[0], B.shape[0], device=A.device, dtype=torch.bfloat16)
flashinfer.mm_fp4(
a_fp4, b_fp4.T, a_sf, b_sf.T, None, torch.bfloat16, out,
block_size=32, use_8x4_sf_layout=False, backend="auto", use_nvfp4=False,
)
return out
Source code from FlashInfer-Bench (flashinfer-ai/flashinfer-trace) · Apache-2.0
Best evidence level for this revision: reproducible
JSON