[all-commits] [llvm/llvm-project] d5c0c1: [AMDGPU] Select flat atomic fmin/fmax

[Mirko Brkušanin via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: d5c0c1b6f0fa7c0bdd9336da10428f27f7d8b411
      https://github.com/llvm/llvm-project/commit/d5c0c1b6f0fa7c0bdd9336da10428f27f7d8b411
  Author: Mirko Brkusanin <Mirko.Brkusanin at amd.com>
  Date:   2023-03-16 (Thu, 16 Mar 2023)

  Changed paths:
    M llvm/lib/Target/AMDGPU/FLATInstructions.td
    A llvm/test/CodeGen/AMDGPU/fp-min-max-flat-atomics-f64.ll
    A llvm/test/CodeGen/AMDGPU/fp-min-max-flat-atomics.ll
    A llvm/test/CodeGen/AMDGPU/fp-min-max-global-atomics-f64.ll
    R llvm/test/CodeGen/AMDGPU/fp-min-max-global-atomics-gfx10.ll
    A llvm/test/CodeGen/AMDGPU/fp-min-max-global-atomics.ll

  Log Message:
  -----------
  [AMDGPU] Select flat atomic fmin/fmax

Also disables global atomic fmin/fmax x2 patterns on gfx11

Differential Revision: https://reviews.llvm.org/D146137