[all-commits] [llvm/llvm-project] 6ada70: [X86][SSE] LowerUINT_TO_FP_i64 - only use HADDPD f...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: 6ada70d1b55b9efb8b0603575a0b86e14f049553
      https://github.com/llvm/llvm-project/commit/6ada70d1b55b9efb8b0603575a0b86e14f049553
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2019-10-19 (Sat, 19 Oct 2019)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/avx512-intrinsics-fast-isel.ll
    M llvm/test/CodeGen/X86/haddsub-3.ll
    M llvm/test/CodeGen/X86/haddsub-broadcast.ll
    M llvm/test/CodeGen/X86/scalar-int-to-fp.ll

  Log Message:
  -----------
  [X86][SSE] LowerUINT_TO_FP_i64 - only use HADDPD for size/fast-hops

We were always generating a single source HADDPD, but really we should only do this if shouldUseHorizontalOp says its a good idea.

Differential Revision: https://reviews.llvm.org/D69175

llvm-svn: 375341