[llvm-branch-commits] [llvm] TTI: Check legalization cost of mul overflow ISD nodes (PR #100519)

[Matt Arsenault via llvm-branch-commits]

https://github.com/arsenm updated https://github.com/llvm/llvm-project/pull/100519

>From fd7e3162be6e57d62942759b10e9a3e192af2d18 Mon Sep 17 00:00:00 2001
From: Matt Arsenault <Matthew.Arsenault at amd.com>
Date: Thu, 25 Jul 2024 10:27:54 +0400
Subject: [PATCH] TTI: Check legalization cost of mul overflow ISD nodes

---
 llvm/include/llvm/CodeGen/BasicTTIImpl.h | 67 +++++++++++++-----------
 1 file changed, 36 insertions(+), 31 deletions(-)

diff --git a/llvm/include/llvm/CodeGen/BasicTTIImpl.h b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
index ac3968de5d672..bdf53abe9533d 100644
--- a/llvm/include/llvm/CodeGen/BasicTTIImpl.h
+++ b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
@@ -2196,37 +2196,11 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
       ISD = ISD::USUBO;
       break;
     case Intrinsic::smul_with_overflow:
-    case Intrinsic::umul_with_overflow: {
-      Type *MulTy = RetTy->getContainedType(0);
-      Type *OverflowTy = RetTy->getContainedType(1);
-      unsigned ExtSize = MulTy->getScalarSizeInBits() * 2;
-      Type *ExtTy = MulTy->getWithNewBitWidth(ExtSize);
-      bool IsSigned = IID == Intrinsic::smul_with_overflow;
-
-      unsigned ExtOp = IsSigned ? Instruction::SExt : Instruction::ZExt;
-      TTI::CastContextHint CCH = TTI::CastContextHint::None;
-
-      InstructionCost Cost = 0;
-      Cost += 2 * thisT()->getCastInstrCost(ExtOp, ExtTy, MulTy, CCH, CostKind);
-      Cost +=
-          thisT()->getArithmeticInstrCost(Instruction::Mul, ExtTy, CostKind);
-      Cost += 2 * thisT()->getCastInstrCost(Instruction::Trunc, MulTy, ExtTy,
-                                            CCH, CostKind);
-      Cost += thisT()->getArithmeticInstrCost(Instruction::LShr, ExtTy,
-                                              CostKind,
-                                              {TTI::OK_AnyValue, TTI::OP_None},
-                                              {TTI::OK_UniformConstantValue, TTI::OP_None});
-
-      if (IsSigned)
-        Cost += thisT()->getArithmeticInstrCost(Instruction::AShr, MulTy,
-                                                CostKind,
-                                                {TTI::OK_AnyValue, TTI::OP_None},
-                                                {TTI::OK_UniformConstantValue, TTI::OP_None});
-
-      Cost += thisT()->getCmpSelInstrCost(
-          BinaryOperator::ICmp, MulTy, OverflowTy, CmpInst::ICMP_NE, CostKind);
-      return Cost;
-    }
+      ISD = ISD::SMULO;
+      break;
+    case Intrinsic::umul_with_overflow:
+      ISD = ISD::UMULO;
+      break;
     case Intrinsic::fptosi_sat:
     case Intrinsic::fptoui_sat: {
       if (Tys.empty())
@@ -2371,6 +2345,37 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
                                           OverflowTy, Pred, CostKind);
       return Cost;
     }
+    case Intrinsic::smul_with_overflow:
+    case Intrinsic::umul_with_overflow: {
+      Type *MulTy = RetTy->getContainedType(0);
+      Type *OverflowTy = RetTy->getContainedType(1);
+      unsigned ExtSize = MulTy->getScalarSizeInBits() * 2;
+      Type *ExtTy = MulTy->getWithNewBitWidth(ExtSize);
+      bool IsSigned = IID == Intrinsic::smul_with_overflow;
+
+      unsigned ExtOp = IsSigned ? Instruction::SExt : Instruction::ZExt;
+      TTI::CastContextHint CCH = TTI::CastContextHint::None;
+
+      InstructionCost Cost = 0;
+      Cost += 2 * thisT()->getCastInstrCost(ExtOp, ExtTy, MulTy, CCH, CostKind);
+      Cost +=
+          thisT()->getArithmeticInstrCost(Instruction::Mul, ExtTy, CostKind);
+      Cost += 2 * thisT()->getCastInstrCost(Instruction::Trunc, MulTy, ExtTy,
+                                            CCH, CostKind);
+      Cost += thisT()->getArithmeticInstrCost(
+          Instruction::LShr, ExtTy, CostKind, {TTI::OK_AnyValue, TTI::OP_None},
+          {TTI::OK_UniformConstantValue, TTI::OP_None});
+
+      if (IsSigned)
+        Cost += thisT()->getArithmeticInstrCost(
+            Instruction::AShr, MulTy, CostKind,
+            {TTI::OK_AnyValue, TTI::OP_None},
+            {TTI::OK_UniformConstantValue, TTI::OP_None});
+
+      Cost += thisT()->getCmpSelInstrCost(
+          BinaryOperator::ICmp, MulTy, OverflowTy, CmpInst::ICMP_NE, CostKind);
+      return Cost;
+    }
     case Intrinsic::sadd_sat:
     case Intrinsic::ssub_sat: {
       // Assume a default expansion.