[PATCH] D69044: [X86] Allow up to 4 loads per inline memcmp()

[David Zarzycki via Phabricator via llvm-commits]

davezarzycki created this revision.
davezarzycki added reviewers: craig.topper, spatel, RKSimon.
davezarzycki added a project: LLVM.
Herald added subscribers: jsji, MaskRay, kbarton, nemanjai.

This effectively relands r308322 / D35067 <https://reviews.llvm.org/D35067>, but sidesteps the PR33914 regression by only increasing the load count for memcmp() if the user only cares about equality (not which operand is greater or lesser).

      

This patch also generalizes combineVectorSizedSetCCEquality() to handle nontrivial memcmp() expansion pass results.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D69044

Files:
  include/llvm/CodeGen/TargetLowering.h
  lib/CodeGen/TargetLoweringBase.cpp
  lib/Target/AArch64/AArch64ISelLowering.cpp
  lib/Target/AArch64/AArch64TargetTransformInfo.cpp
  lib/Target/PowerPC/PPCISelLowering.cpp
  lib/Target/PowerPC/PPCTargetTransformInfo.cpp
  lib/Target/X86/X86ISelLowering.cpp
  lib/Target/X86/X86TargetTransformInfo.cpp
  test/CodeGen/X86/memcmp.ll
  test/Transforms/ExpandMemCmp/X86/memcmp.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D69044.225232.patch
Type: text/x-patch
Size: 88339 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20191016/c361200b/attachment-0001.bin>