[llvm] [AArch64][LV][SLP] Vectorizers use getFRemInstrCost for frem costs (PR #82488)

[Paul Walker via llvm-commits]

================
@@ -6899,25 +6899,15 @@ LoopVectorizationCostModel::getInstructionCost(Instruction *I, ElementCount VF,
       Op2Info.Kind = TargetTransformInfo::OK_UniformValue;
 
     SmallVector<const Value *, 4> Operands(I->operand_values());
-    auto InstrCost = TTI.getArithmeticInstrCost(
-        I->getOpcode(), VectorTy, CostKind,
-        {TargetTransformInfo::OK_AnyValue, TargetTransformInfo::OP_None},
-        Op2Info, Operands, I);
-
-    // Some targets can replace frem with vector library calls.
-    InstructionCost VecCallCost = InstructionCost::getInvalid();
-    if (I->getOpcode() == Instruction::FRem) {
-      LibFunc Func;
-      if (TLI->getLibFunc(I->getOpcode(), I->getType(), Func) &&
-          TLI->isFunctionVectorizable(TLI->getName(Func), VF)) {
-        SmallVector<Type *, 4> OpTypes;
-        for (auto &Op : I->operands())
-          OpTypes.push_back(Op->getType());
-        VecCallCost =
-            TTI.getCallInstrCost(nullptr, VectorTy, OpTypes, CostKind);
-      }
-    }
-    return std::min(InstrCost, VecCallCost);
+    TTI::OperandValueInfo Op1Info{TargetTransformInfo::OK_AnyValue,
+                                  TargetTransformInfo::OP_None};
+    // Some targets replace frem with vector library calls.
----------------
paulwalker-arm wrote:

This comment is not longer relevant.  We're simply using a more specific costing function, which is clear from the code.

https://github.com/llvm/llvm-project/pull/82488