[llvm] [X86] Prefer sub-of-not for vector inc-of-add (PR #207695)

[Rohan Shenoy via llvm-commits]

================
@@ -60622,6 +60629,12 @@ static SDValue combineSub(SDNode *N, SelectionDAG &DAG,
     return DAG.getNode(ISD::SUB, DL, VT, ADC.getValue(0), Op1.getOperand(0));
   }
 
+  // Fold SUB(~A,~B) -> SUB(B,A)
+  if (isBitwiseNot(Op0) && Op0.hasOneUse() && isBitwiseNot(Op1) &&
+      Op1.hasOneUse()) {
----------------
222rohan wrote:

AFAIK as of now it only affects x86 path for this particular patch. Unsure if generic path would benefit, I will have to check

https://github.com/llvm/llvm-project/pull/207695