[x86] scalarize extract element 0 of FP cmp
authorSanjay Patel <spatel@rotateright.com>
Fri, 8 Mar 2019 21:54:41 +0000 (21:54 +0000)
committerSanjay Patel <spatel@rotateright.com>
Fri, 8 Mar 2019 21:54:41 +0000 (21:54 +0000)
commitf84083b4dbb1ddb6d2783400f11121f490cdb5a8
tree84e1602ac6f67df3901c30cb45cda83ee62eabc4
parenta8b3eb46b57d2a6e50e1352b7c590d2338c5c2e8
[x86] scalarize extract element 0 of FP cmp

An extension of D58282 noted in PR39665:
https://bugs.llvm.org/show_bug.cgi?id=39665

This doesn't answer the request to use movmsk, but that's an
independent problem. We need this and probably still need
scalarization of FP selects because we can't do that as a
target-independent transform (although it seems likely that
targets besides x86 should have this transform).

llvm-svn: 355741
llvm/lib/Target/X86/X86ISelLowering.cpp
llvm/test/CodeGen/X86/extractelement-fp.ll
llvm/test/CodeGen/X86/vec_floor.ll