[all-commits] [llvm/llvm-project] 0b665c: [mlir][scf] Implement conversion from scf.forall t...

[Spenser Bauman via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 0b665c3dd206abb3deab4c3af9dfeb43000ab0c7
      https://github.com/llvm/llvm-project/commit/0b665c3dd206abb3deab4c3af9dfeb43000ab0c7
  Author: Spenser Bauman <sabauma at fastmail.com>
  Date:   2024-06-04 (Tue, 04 Jun 2024)

  Changed paths:
    M mlir/include/mlir/Dialect/SCF/TransformOps/SCFTransformOps.td
    M mlir/include/mlir/Dialect/SCF/Transforms/Passes.h
    M mlir/include/mlir/Dialect/SCF/Transforms/Passes.td
    M mlir/include/mlir/Dialect/SCF/Transforms/Transforms.h
    M mlir/lib/Conversion/SCFToControlFlow/CMakeLists.txt
    M mlir/lib/Conversion/SCFToControlFlow/SCFToControlFlow.cpp
    M mlir/lib/Dialect/SCF/TransformOps/SCFTransformOps.cpp
    M mlir/lib/Dialect/SCF/Transforms/CMakeLists.txt
    A mlir/lib/Dialect/SCF/Transforms/ForallToParallel.cpp
    A mlir/test/Dialect/SCF/forall-to-parallel.mlir
    A mlir/test/Dialect/SCF/transform-op-forall-to-parallel.mlir

  Log Message:
  -----------
  [mlir][scf] Implement conversion from scf.forall to scf.parallel (#94109)

There is currently no path to lower scf.forall to scf.parallel with the
goal of targeting the OpenMP dialect.

In the SCF->ControlFlow conversion, scf.forall is briefly converted to
scf.parallel, but the scf.parallel is lowered directly to a sequential
loop. This makes experimenting with scf.forall for CPU execution
difficult.

This change factors out the rewrite in the SCF->ControlFlow pass into a
utility function that can then be used in the SCF->ControlFlow lowering
and via a separate -scf-forall-to-parallel pass.

---------

Co-authored-by: Spenser Bauman <sabauma at fastmail>



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