[llvm-branch-commits] [llvm] [X86] Cast atomic vectors in IR to support floats (PR #142320)

[Matt Arsenault via llvm-branch-commits]

================
@@ -32070,6 +32070,13 @@ X86TargetLowering::shouldExpandAtomicRMWInIR(AtomicRMWInst *AI) const {
   }
 }
 
+TargetLowering::AtomicExpansionKind
+X86TargetLowering::shouldCastAtomicLoadInIR(LoadInst *LI) const {
+  if (LI->getType()->getScalarType()->isFloatingPointTy())
+    return AtomicExpansionKind::CastToInteger;
+  return AtomicExpansionKind::None;
+}
+
----------------
arsenm wrote:

We want to get away from using the IR expansion

https://github.com/llvm/llvm-project/pull/142320