[llvm] [AMDGPU][AggressiveInstCombine] Narrow 64 bit math to 32 bit if profitable (PR #130577)

[Yingwei Zheng via llvm-commits]

================
@@ -1224,6 +1224,68 @@ static bool foldLibCalls(Instruction &I, TargetTransformInfo &TTI,
   return false;
 }
 
+static bool tryNarrowMathIfNoOverflow(Instruction &I, TargetTransformInfo &TTI,
+                                      const DataLayout &DL) {
+  unsigned opc = I.getOpcode();
+  Type *OldType = I.getType();
+  if (opc != Instruction::Add && opc != Instruction::Mul &&
+      !OldType->isIntOrIntVectorTy()) {
----------------
dtcxzyw wrote:

Add/Mul only work with integer/intvector types.
```suggestion
      ) {
```

https://github.com/llvm/llvm-project/pull/130577