[all-commits] [llvm/llvm-project] 1725c5: [mlir][acc] Add attribute for redundant execution ...

[Razvan Lupusoru via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 1725c53752c84cc19c656413b0e75e1dbe9e154b
      https://github.com/llvm/llvm-project/commit/1725c53752c84cc19c656413b0e75e1dbe9e154b
  Author: Razvan Lupusoru <razvan.lupusoru at gmail.com>
  Date:   2026-07-15 (Wed, 15 Jul 2026)

  Changed paths:
    M mlir/include/mlir/Dialect/OpenACC/OpenACCCGAttributes.td
    M mlir/include/mlir/Dialect/OpenACC/OpenACCUtilsCG.h
    M mlir/lib/Dialect/OpenACC/Transforms/ACCCGToGPU.cpp
    M mlir/lib/Dialect/OpenACC/Utils/OpenACCUtilsCG.cpp
    A mlir/test/Dialect/OpenACC/acc-cg-to-gpu-block-redundant.mlir

  Log Message:
  -----------
  [mlir][acc] Add attribute for redundant execution marking (#209784)

Adds `acc.gpu_block_redundant` attribute for correct gang-redundant
lowering.

OpenACC gang-redundant regions execute on every gang rather than
partitioning work across gangs. When lowering to the GPU dialect, such
regions must not be assigned block-level par dims; only thread-level
parallelism applies. This marks loops with this attribute to ensure
proper execution and to avoid incorrect predication.

Co-authored-by: Delaram Talaashrafi <dtalaashrafi at nvidia.com>



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