[all-commits] [llvm/llvm-project] 4524d8: [X86] combineHorizOpWithShuffle - generalize HOP(S...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 4524d8b7552c1d2dcb941c708620fb1f1b998cd1
      https://github.com/llvm/llvm-project/commit/4524d8b7552c1d2dcb941c708620fb1f1b998cd1
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-05-08 (Sat, 08 May 2021)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/horizontal-sum.ll

  Log Message:
  -----------
  [X86] combineHorizOpWithShuffle - generalize HOP(SHUFFLE(X),SHUFFLE(Y)) -> SHUFFLE(HOP(X,Y)) fold.

For 128-bit types, generalize the fold to recognise duplicate operands in either shuffle.


  Commit: ab5ee342b92b4661cfec3cdd647c9a5c18e346dd
      https://github.com/llvm/llvm-project/commit/ab5ee342b92b4661cfec3cdd647c9a5c18e346dd
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-05-08 (Sat, 08 May 2021)

  Changed paths:
    M llvm/include/llvm/CodeGen/GlobalISel/MachineIRBuilder.h

  Log Message:
  -----------
  [GlobalISel] Ensure MachineIRBuilder::getDebugLoc() returns a const reference. NFCI.

Avoids a lot of unnecessary tracking increments/decrements of the underlying TrackingMDNodeRef.


Compare: https://github.com/llvm/llvm-project/compare/74d096e55879...ab5ee342b92b