[llvm] [AArch64][SDAG] Enable +sve2p2/+sme2p2 lowerings for VECTOR_COMPRESS (PR #215219)

Benjamin Maxwell via llvm-commits llvm-commits at lists.llvm.org
Tue Aug 11 06:06:32 PDT 2026


================
@@ -2166,6 +2166,36 @@ AArch64TargetLowering::AArch64TargetLowering(const TargetMachine &TM,
     setOperationAction(ISD::CLMUL, MVT::nxv2i64, Custom);
   }
 
+  if (Subtarget->isSVEAvailable() ||
+      (Subtarget->isSVEorStreamingSVEAvailable() && Subtarget->hasSME2p2())) {
----------------
MacDue wrote:

These predicates are a bit tricky, but that change is not equivalent. It disables the lowering in streaming compatible functions where both +sme2p2 and +sve2p2 are available. I'll add a test. 

https://github.com/llvm/llvm-project/pull/215219


More information about the llvm-commits mailing list