[all-commits] [llvm/llvm-project] ed3e49: [X86] Fold PACK(*_EXTEND_VECTOR_INREG, UNDEF) -> *...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: ed3e4917b36f2530703115066700daeb2b45b4f0
      https://github.com/llvm/llvm-project/commit/ed3e4917b36f2530703115066700daeb2b45b4f0
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-09-26 (Sun, 26 Sep 2021)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/pmaddubsw.ll

  Log Message:
  -----------
  [X86] Fold PACK(*_EXTEND_VECTOR_INREG, UNDEF) -> *_EXTEND_VECTOR_INREG

For 128-bit vectors, we can remove a PACK of a EXTEND_VECTOR_INREG node and just create a smaller extension to the result/packed type.


  Commit: c0eff50fc5a48990ad9ebfcb7e81c6ab6fea79c5
      https://github.com/llvm/llvm-project/commit/c0eff50fc5a48990ad9ebfcb7e81c6ab6fea79c5
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-09-26 (Sun, 26 Sep 2021)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/madd.ll

  Log Message:
  -----------
  [X86][SSE] combineMulToPMADDWD - enable sext_extend_vector_inreg(vXi16) -> zext_extend_vector_inreg(vXi16) fold

The plan is to allow combineMulToPMADDWD to match illegal vector types (as long as they're still pow2), which should allow us to start removing the 128-bit limit on more of the PMADDWD combines.


Compare: https://github.com/llvm/llvm-project/compare/f40685138ba1...c0eff50fc5a4