[PATCH] D153879: [AMDGPU] Handle Additional Cases in tryFoldPhiAGPR

[Pierre van Houtryve via Phabricator via llvm-commits]

Pierre-vh created this revision.
Pierre-vh added reviewers: AMDGPU, arsenm.
Herald added subscribers: foad, kerbowa, hiraditya, tpr, dstuttard, yaxunl, jvesely, kzhuravl.
Herald added a project: All.
Pierre-vh requested review of this revision.
Herald added subscribers: llvm-commits, wdng.
Herald added a project: LLVM.

Sometimes PHI have different incoming values, such as:

  %1:vgpr_256 = COPY %0:agpr_256
  %2:vgpr_32 = COPY %1:vgpr_256.sub0

Those weren't handled, which could lead to massive performance issues if break-large-PHIs kicked in + AGPRs were used (MFMA)

Fixes SWDEV-407986


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D153879

Files:
  llvm/lib/Target/AMDGPU/SIFoldOperands.cpp
  llvm/test/CodeGen/AMDGPU/fold-agpr-phis.mir

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D153879.534988.patch
Type: text/x-patch
Size: 9340 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230627/e2d63a92/attachment.bin>