[PATCH] D131560: AMDGPU: Improve atomicrmw fadd selection

[Petar Avramovic via Phabricator via llvm-commits]

Petar.Avramovic updated this revision to Diff 462025.
Petar.Avramovic added a comment.

Treat fadd rmw atomics on gfx908, gfx940 and gfx11+ the same way as for gfx90a.
Regression for gfx940 which now expands rmw in some cases that it did not used to.
Added tests to transforms for interesting cases that don't expand rmw.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131560/new/

https://reviews.llvm.org/D131560

Files:
  llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
  llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp
  llvm/lib/Target/AMDGPU/SIISelLowering.cpp
  llvm/test/CodeGen/AMDGPU/GlobalISel/flat-atomic-fadd.f32.ll
  llvm/test/CodeGen/AMDGPU/GlobalISel/fp-atomics-gfx940.ll
  llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.global.atomic.fadd-with-ret.ll
  llvm/test/CodeGen/AMDGPU/fp-atomics-gfx940.ll
  llvm/test/CodeGen/AMDGPU/global-atomics-fp.ll
  llvm/test/Transforms/AtomicExpand/AMDGPU/expand-atomic-rmw-fadd.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D131560.462025.patch
Type: text/x-patch
Size: 51142 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220921/9b25a08c/attachment.bin>