[llvm] [LV][AArch64] LoopVectorizer allows scalable frem instructions (PR #76247)

[Paschalis Mpeis via llvm-commits]

================
@@ -6951,10 +6951,26 @@ LoopVectorizationCostModel::getInstructionCost(Instruction *I, ElementCount VF,
       Op2Info.Kind = TargetTransformInfo::OK_UniformValue;
 
     SmallVector<const Value *, 4> Operands(I->operand_values());
-    return TTI.getArithmeticInstrCost(
+    auto InstrCost = TTI.getArithmeticInstrCost(
         I->getOpcode(), VectorTy, CostKind,
         {TargetTransformInfo::OK_AnyValue, TargetTransformInfo::OP_None},
         Op2Info, Operands, I);
+
+    // Some targets replace frem with vector library calls.
+    if (I->getOpcode() == Instruction::FRem) {
+      LibFunc Func;
+      if (TLI->getLibFunc(I->getOpcode(), I->getType(), Func)) {
+        if (TLI->isFunctionVectorizable(TLI->getName(Func))) {
+          SmallVector<Type *, 4> OpTypes;
+          for (auto &Op : I->operands())
+            OpTypes.push_back(Op->getType());
+          auto CallCost =
+              TTI.getCallInstrCost(nullptr, VectorTy, OpTypes, CostKind);
+          return std::min(InstrCost, CallCost);
----------------
paschalis-mpeis wrote:

The below PR will bring such functionality to `replace-with-veclib`. Thank you all for your input!
- #78688 

https://github.com/llvm/llvm-project/pull/76247