[all-commits] [llvm/llvm-project] 6072e4: [MLIR][XeGPU] Preserve leading unit dimension duri...

[Jianhui Li via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 6072e4006fb1c66331372fd11205e034dafe0ca3
      https://github.com/llvm/llvm-project/commit/6072e4006fb1c66331372fd11205e034dafe0ca3
  Author: Jianhui Li <jian.hui.li at intel.com>
  Date:   2026-02-12 (Thu, 12 Feb 2026)

  Changed paths:
    M mlir/lib/Dialect/XeGPU/Transforms/XeGPUBlocking.cpp
    M mlir/lib/Dialect/XeGPU/Transforms/XeGPUSubgroupDistribute.cpp
    M mlir/test/Dialect/XeGPU/subgroup-distribute-unit.mlir
    M mlir/test/Dialect/XeGPU/xegpu-blocking.mlir

  Log Message:
  -----------
  [MLIR][XeGPU] Preserve leading unit dimension during blocking (#180884)

This PR preserve leading dimension during blocking. This ensures the
blocking process avoid generating unnecessary
insert/extract_strided_slice, which under certain condition becomes
difficult to be canceled, and creates extra burden in lane layout
propagation and subgroup distribution.
This PR also extended subgroup distribution so load and store can
support payload/mask/offsets with leading unit dimension. The
distributed load/store works on 1d only, but shapecast is inserted to
remove and add the leading dimension for the input/output vectors.
Comparing to the insert/extract inserted at subgroup level, the
shapecast inserted at lane level handling leading unit dimension is
essentially a nop and can be processed lightly.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications