[all-commits] [llvm/llvm-project] 6f4a52: [mlir][memref] Use dedicated ops in `AtomicRMWOpCo...

[Daniil Dudkin via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 6f4a5286981dd57826751834f7a434be1a463339
      https://github.com/llvm/llvm-project/commit/6f4a5286981dd57826751834f7a434be1a463339
  Author: Daniil Dudkin <39276703+unterumarmung at users.noreply.github.com>
  Date:   2023-09-15 (Fri, 15 Sep 2023)

  Changed paths:
    M mlir/lib/Dialect/MemRef/Transforms/ExpandOps.cpp
    M mlir/test/Dialect/MemRef/expand-ops.mlir

  Log Message:
  -----------
  [mlir][memref] Use dedicated ops in `AtomicRMWOpConverter` (#66437)

This patch refactors the `AtomicRMWOpConverter` class to use
the dedicated operations from Arith dialect instead of using
`cmpf` + `select` pattern.
Also, a test for `minimumf` kind of `atomic_rmw` has been added.