[llvm] [AMDGPU] Add folding ISD::SELECT from vXiY into vZi32 with X * Y = Z * 32 (PR #173328)

[Matt Arsenault via llvm-commits]

================
@@ -6985,6 +6985,72 @@ static SDValue combineSelectAsExtAnd(SDValue Cond, SDValue T, SDValue F,
   return DAG.getNode(ISD::AND, DL, OpVT, CondMask, T.getOperand(0));
 }
 
+// Widden int vector is not power of 2 in length.
+static SDValue widdenIntVectorSelect(SDNode *N, SelectionDAG &DAG,
----------------
arsenm wrote:

Typo widden 

https://github.com/llvm/llvm-project/pull/173328