[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

[Alexis Engelke via llvm-commits]

================
@@ -202,13 +202,13 @@ define { <4 x i8>, <4 x i1> } @always_usub_const_vector() nounwind {
 ; SSE-LABEL: always_usub_const_vector:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE-NEXT:    pcmpeqd %xmm1, %xmm1
+; SSE-NEXT:    movdqa %xmm0, %xmm1
----------------
aengelke wrote:

Any evidence why move is preferable over the zero-idiom?

https://github.com/llvm/llvm-project/pull/163047