[llvm] add power function to APInt (PR #122788)

[Craig Topper via llvm-commits]

================
@@ -3108,3 +3108,21 @@ APInt APIntOps::mulhu(const APInt &C1, const APInt &C2) {
   APInt C2Ext = C2.zext(FullWidth);
   return (C1Ext * C2Ext).extractBits(C1.getBitWidth(), C1.getBitWidth());
 }
+
+APInt APIntOps::pow(const APInt &X, int64_t N) {
+  assert(N >= 0 && "negative exponents not supported.");
+  APInt Acc = APInt(X.getBitWidth(), 1);
+  if (N == 0)
+    return Acc;
+  APInt Base = X;
+  int64_t RemainingExponent = N;
+  while (RemainingExponent > 0) {
+    while (RemainingExponent % 2 == 0) {
+      Base = Base * Base;
----------------
topperc wrote:

`Base *= Base` might give the APInt the opportunity to update `Base` in place for large bit widths. Though I don't think that optimization exists for multiply. It does exist for add/sub.

https://github.com/llvm/llvm-project/pull/122788