[llvm] [InstCombine][X86] Try to convert BLENDV(X,Y,SHL()) -> SELECT(ICMP_SGT(0,SHL()),Y,X) (PR #173389)
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Wed Dec 24 02:07:35 PST 2025
- Previous message: [llvm] [InstCombine][X86] Try to convert BLENDV(X,Y,SHL()) -> SELECT(ICMP_SGT(0,SHL()),Y,X) (PR #173389)
- Next message: [llvm] [InstCombine][X86] Try to convert BLENDV(X,Y,SHL()) -> SELECT(ICMP_SGT(0,SHL()),Y,X) (PR #173389)
- Messages sorted by:
[ date ]
[ thread ]
[ subject ]
[ author ]
================
@@ -2892,6 +2893,19 @@ X86TTIImpl::instCombineIntrinsic(InstCombiner &IC, IntrinsicInst &II) const {
Mask = InstCombiner::peekThroughBitcast(Mask);
+ // Bitshift upto the signbit can always be converted to an efficient
+ // test+select pattern.
+ if (match(Mask, m_Shl(m_Value(), m_Value()))) {
+ if (auto *MaskTy = dyn_cast<FixedVectorType>(Mask->getType())) {
----------------
RKSimon wrote:
There is the remote possibility that we've peeked through bitcasts to a non-vector type.
https://github.com/llvm/llvm-project/pull/173389
- Previous message: [llvm] [InstCombine][X86] Try to convert BLENDV(X,Y,SHL()) -> SELECT(ICMP_SGT(0,SHL()),Y,X) (PR #173389)
- Next message: [llvm] [InstCombine][X86] Try to convert BLENDV(X,Y,SHL()) -> SELECT(ICMP_SGT(0,SHL()),Y,X) (PR #173389)
- Messages sorted by:
[ date ]
[ thread ]
[ subject ]
[ author ]
More information about the llvm-commits
mailing list