cd9e59a140a9649186419fd401be7fd881a838d6 - llvm

commit	cd9e59a140a9649186419fd401be7fd881a838d6	[log] [tgz]
author	Simon Pilgrim <llvm-dev@redking.me.uk>	Sun Jan 20 19:27:40 2019 +0000
committer	Simon Pilgrim <llvm-dev@redking.me.uk>	Sun Jan 20 19:27:40 2019 +0000
tree	46d1d51f0bde36001faaa1c96ab97df7facb2d3f
parent	e8d1d04c89079df3c0cf197df2e89e864d0a716c [diff]

[X86] Auto upgrade VPCOM/VPCOMU intrinsics to generic integer comparisons

This causes a couple of changes in the upgrade tests as signed/unsigned eq/ne are equivalent and we constant fold true/false codes, these changes are the same as what we already do for avx512 cmp/ucmp.

Noticed while cleaning up vector integer comparison costs for PR40376.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@351697 91177308-0d34-0410-b5e6-96231b3b80d8

include/llvm/IR/IntrinsicsX86.td[diff]
lib/IR/AutoUpgrade.cpp[diff]
lib/Target/X86/X86IntrinsicsInfo.h[diff]
lib/Transforms/InstCombine/InstCombineCalls.cpp[diff]
test/CodeGen/X86/commute-xop.ll[diff]
test/CodeGen/X86/xop-intrinsics-x86_64-upgrade.ll[diff]
test/CodeGen/X86/xop-intrinsics-x86_64.ll[diff]

7 files changed

tree: 46d1d51f0bde36001faaa1c96ab97df7facb2d3f