[PATCH] D129073: [AMDGPU] Combine s_or_saveexec, s_xor instructions.

[Thomas Symalla via Phabricator via llvm-commits]

tsymalla created this revision.
tsymalla added reviewers: foad, critson, nhaehnle.
Herald added subscribers: kosarev, jsilvanus, kerbowa, hiraditya, t-tye, tpr, dstuttard, yaxunl, jvesely, kzhuravl, arsenm, MatzeB.
Herald added a project: All.
tsymalla requested review of this revision.
Herald added subscribers: llvm-commits, wdng.
Herald added a project: LLVM.

This patch merges a consecutive sequence of

s_or_saveexec s_i, s_i
s_xor exec, exec, s_i

into a single

s_andn2_saveexec s_i, s_i instruction.
This patch also cleans up the SIOptimizeExecMasking pass a bit.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D129073

Files:
  llvm/lib/Target/AMDGPU/SIOptimizeExecMasking.cpp
  llvm/test/CodeGen/AMDGPU/GlobalISel/localizer.ll
  llvm/test/CodeGen/AMDGPU/GlobalISel/sdiv.i64.ll
  llvm/test/CodeGen/AMDGPU/GlobalISel/udiv.i64.ll
  llvm/test/CodeGen/AMDGPU/branch-relaxation.ll
  llvm/test/CodeGen/AMDGPU/collapse-endcf.ll
  llvm/test/CodeGen/AMDGPU/else.ll
  llvm/test/CodeGen/AMDGPU/llvm.amdgcn.softwqm.ll
  llvm/test/CodeGen/AMDGPU/s_or_saveexec_xor_combine.mir
  llvm/test/CodeGen/AMDGPU/set-inactive-wwm-overwrite.ll
  llvm/test/CodeGen/AMDGPU/transform-block-with-return-to-epilog.ll
  llvm/test/CodeGen/AMDGPU/valu-i1.ll
  llvm/test/CodeGen/AMDGPU/vcmp-saveexec-to-vcmpx.ll
  llvm/test/CodeGen/AMDGPU/vgpr-liverange.ll
  llvm/test/CodeGen/AMDGPU/wqm.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D129073.442065.patch
Type: text/x-patch
Size: 52800 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220704/27b98d05/attachment.bin>