[llvm] [X86] Fold FMSUBADD into VFCMULC for fp16 complex conjugate multiply (PR #224533)

[Simon Pilgrim via llvm-commits]

================
@@ -58559,18 +58550,37 @@ static SDValue combineFMA(SDNode *N, SelectionDAG &DAG,
   }
 }
 
+// Combine FMSUBADD(A, B, C) -> VFCMULC(A, B) for fp16 (complex)
----------------
RKSimon wrote:

Tweak the fold comment to show the correct pattern

https://github.com/llvm/llvm-project/pull/224533