[llvm] [AMDGPU] Implement IR variant of isFMAFasterThanFMulAndFAdd (PR #121465)

[Matt Arsenault via llvm-commits]

================
@@ -16942,6 +16969,37 @@ bool SITargetLowering::checkForPhysRegDependency(
   return false;
 }
 
+/// Check if it is profitable to hoist instruction in then/else to if.
+/// Not profitable if I and it's user can form a FMA instruction
+/// because we prefer FMSUB/FMADD.
+bool SITargetLowering::isProfitableToHoist(Instruction *I) const {
+  if (!I->hasOneUse())
+    return true;
+
+  Instruction *User = I->user_back();
+  // TODO: Add more patterns that are not profitable to hoist
+  switch (I->getOpcode()) {
+  case Instruction::FMul: {
+    if (User->getOpcode() != Instruction::FSub &&
+        User->getOpcode() != Instruction::FAdd)
+      return true;
+
+    const TargetOptions &Options = getTargetMachine().Options;
+    const Function *F = I->getFunction();
+    const DataLayout &DL = F->getDataLayout();
+    Type *Ty = User->getOperand(0)->getType();
+
+    return !isOperationLegalOrCustom(ISD::FMA, getValueType(DL, Ty)) ||
----------------
arsenm wrote:

We probably don't need the legality check. It's legal for every type that matters. Maybe we care about the number of pieces in the wide vector cases? 

https://github.com/llvm/llvm-project/pull/121465