[X86][SSE] canReduceVMulWidth - use ComputeNumSignBits/SignBitIsZero directly
authorSimon Pilgrim <llvm-dev@redking.me.uk>
Wed, 26 Sep 2018 11:48:52 +0000 (11:48 +0000)
committerSimon Pilgrim <llvm-dev@redking.me.uk>
Wed, 26 Sep 2018 11:48:52 +0000 (11:48 +0000)
commitebabd79f43be3360ea2f83e0169396e72162b547
tree2042af87a33d36f3d9bec1d0bf4132387f1e8a0a
parent5d6f3d9f45b2222f97515b2b07fa214dbd016a53
[X86][SSE] canReduceVMulWidth - use ComputeNumSignBits/SignBitIsZero directly

Don't reinvent the wheel for BUILD_VECTOR/ZERO_EXTEND - its only the ANY_EXTEND special case that needs handling.

llvm-svn: 343096
llvm/lib/Target/X86/X86ISelLowering.cpp