[PATCH] D130622: [AMDGPU][SIFoldOperands] Clear kills when folding COPY

[Jay Foad via Phabricator via llvm-commits]

foad added a comment.

>> See: https://reviews.llvm.org/D52577#1248697 which introduced this code, and the comment: "Why is this necessary? PeepholeOpt runs first and should have eliminated redundant copies"
>
> In my specific case these extra COPYs come from SILoadStoreOptimizer.

Maybe SILoadStoreOptimizer should not insert copies at all. If it replaces:

  %old0:vgpr_32 = BUFFER_LOAD_DWORD ...
  %old1:vgpr_32 = BUFFER_LOAD_DWORD ...

with:

  %new:vgpr_64 = BUFFER_LOAD_DWORDX2 ...

then perhaps it could just replace all uses of %old0 with %new.sub0?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D130622/new/

https://reviews.llvm.org/D130622