List view

[ROCm] Enable MXFP8/MXFP4 emulation tests on ROCm (MI300+)

ciflow/rocm

ciflow/rocm-mi300

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

module: trainingquantize_ api training flowquantize_ api training flow

topic: rocm

Status: Open (in progress).

pytorch/aonumber 4041

#4041 In pytorch/ao;

· brucechanglongxu opened on Mar 10, 2026

Triton + Symmetric Memory fused all2all dispatch + token group padding kernel

moe

mx

Status: Open.

#4050 In pytorch/ao;

· danielvegamyhre opened on Mar 11, 2026

CUDA -> CuteDSL dim1 quantization kernel with RCEIL scaling that writes scales directly to ((32,4),4) layout for tcgen05 mma

mx

Status: Open.

#4053 In pytorch/ao;

· danielvegamyhre opened on Mar 11, 2026

[mxfp8 moe training] fused all2all dispatch + token group padding + expert major kernel

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

module: trainingquantize_ api training flowquantize_ api training flow

Status: Open (in progress).

pytorch/aonumber 4066

#4066 In pytorch/ao;

· danielvegamyhre opened on Mar 12, 2026

Fuse expert grouping and padding into the MXFP8 EP dispatch path

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

module: trainingquantize_ api training flowquantize_ api training flow

moe

mx

Status: Open (in progress).

pytorch/aonumber 4086

#4086 In pytorch/ao;

· MagellaX opened on Mar 14, 2026

[MXFP8 MoE] mx_block_rearrange_2d_M_groups_cuda fails for models with > 32 experts

module: trainingquantize_ api training flowquantize_ api training flow

moe

mx

Status: Open.

#4163 In pytorch/ao;

· yzhautouskay opened on Mar 24, 2026

fuse dynamic per group padding into cutedsl 2d MXFP8 quantization kernel

moe

mx

Status: Open.

#4184 In pytorch/ao;

· danielvegamyhre opened on Mar 26, 2026

Support 32x32 scaling for weights in MXFP8 weight quantization kernel

moe

mx

Status: Open.

#4185 In pytorch/ao;

· danielvegamyhre opened on Mar 26, 2026