diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h b/llvm/include/llvm/Analysis/TargetTransformInfo.h
--- a/llvm/include/llvm/Analysis/TargetTransformInfo.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h
@@ -1193,7 +1193,8 @@
   /// case is to provision the cost of vectorization/scalarization in
   /// vectorizer passes.
   InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index = -1) const;
+                                     unsigned Index = -1, Value *Op0 = nullptr,
+                                     Value *Op1 = nullptr) const;
 
   /// \return The expected cost of vector Insert and Extract.
   /// This is used when instruction is available, and implementation
@@ -1786,7 +1787,8 @@
                                              TTI::TargetCostKind CostKind,
                                              const Instruction *I) = 0;
   virtual InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                             unsigned Index) = 0;
+                                             unsigned Index, Value *Op0,
+                                             Value *Op1) = 0;
   virtual InstructionCost getVectorInstrCost(const Instruction &I, Type *Val,
                                              unsigned Index) = 0;
 
@@ -2358,9 +2360,9 @@
                                      const Instruction *I) override {
     return Impl.getCmpSelInstrCost(Opcode, ValTy, CondTy, VecPred, CostKind, I);
   }
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index) override {
-    return Impl.getVectorInstrCost(Opcode, Val, Index);
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1) override {
+    return Impl.getVectorInstrCost(Opcode, Val, Index, Op0, Op1);
   }
   InstructionCost getVectorInstrCost(const Instruction &I, Type *Val,
                                      unsigned Index) override {
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
--- a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
@@ -585,8 +585,8 @@
     return 1;
   }
 
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index) const {
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1) const {
     return 1;
   }
 
diff --git a/llvm/include/llvm/CodeGen/BasicTTIImpl.h b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
--- a/llvm/include/llvm/CodeGen/BasicTTIImpl.h
+++ b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
@@ -90,10 +90,12 @@
     InstructionCost Cost = 0;
     // Broadcast cost is equal to the cost of extracting the zero'th element
     // plus the cost of inserting it into every element of the result vector.
-    Cost += thisT()->getVectorInstrCost(Instruction::ExtractElement, VTy, 0);
+    Cost += thisT()->getVectorInstrCost(Instruction::ExtractElement, VTy, 0,
+                                        nullptr, nullptr);
 
     for (int i = 0, e = VTy->getNumElements(); i < e; ++i) {
-      Cost += thisT()->getVectorInstrCost(Instruction::InsertElement, VTy, i);
+      Cost += thisT()->getVectorInstrCost(Instruction::InsertElement, VTy, i,
+                                          nullptr, nullptr);
     }
     return Cost;
   }
@@ -110,8 +112,10 @@
     // vector and finally index 3 of second vector and insert them at index
     // <0,1,2,3> of result vector.
     for (int i = 0, e = VTy->getNumElements(); i < e; ++i) {
-      Cost += thisT()->getVectorInstrCost(Instruction::InsertElement, VTy, i);
-      Cost += thisT()->getVectorInstrCost(Instruction::ExtractElement, VTy, i);
+      Cost += thisT()->getVectorInstrCost(Instruction::InsertElement, VTy, i,
+                                          nullptr, nullptr);
+      Cost += thisT()->getVectorInstrCost(Instruction::ExtractElement, VTy, i,
+                                          nullptr, nullptr);
     }
     return Cost;
   }
@@ -134,9 +138,9 @@
     // type.
     for (int i = 0; i != NumSubElts; ++i) {
       Cost += thisT()->getVectorInstrCost(Instruction::ExtractElement, VTy,
-                                          i + Index);
-      Cost +=
-          thisT()->getVectorInstrCost(Instruction::InsertElement, SubVTy, i);
+                                          i + Index, nullptr, nullptr);
+      Cost += thisT()->getVectorInstrCost(Instruction::InsertElement, SubVTy, i,
+                                          nullptr, nullptr);
     }
     return Cost;
   }
@@ -158,10 +162,10 @@
     // the source type plus the cost of inserting them into the result vector
     // type.
     for (int i = 0; i != NumSubElts; ++i) {
-      Cost +=
-          thisT()->getVectorInstrCost(Instruction::ExtractElement, SubVTy, i);
+      Cost += thisT()->getVectorInstrCost(Instruction::ExtractElement, SubVTy,
+                                          i, nullptr, nullptr);
       Cost += thisT()->getVectorInstrCost(Instruction::InsertElement, VTy,
-                                          i + Index);
+                                          i + Index, nullptr, nullptr);
     }
     return Cost;
   }
@@ -212,7 +216,7 @@
                                  FixedVectorType::get(
                                      PointerType::get(VT->getElementType(), 0),
                                      VT->getNumElements()),
-                                 -1)
+                                 -1, nullptr, nullptr)
             : 0;
     InstructionCost LoadCost =
         VT->getNumElements() *
@@ -237,7 +241,7 @@
                Instruction::ExtractElement,
                FixedVectorType::get(Type::getInt1Ty(DataTy->getContext()),
                                     VT->getNumElements()),
-               -1) +
+               -1, nullptr, nullptr) +
            getCFInstrCost(Instruction::Br, CostKind) +
            getCFInstrCost(Instruction::PHI, CostKind));
     }
@@ -722,9 +726,11 @@
       if (!DemandedElts[i])
         continue;
       if (Insert)
-        Cost += thisT()->getVectorInstrCost(Instruction::InsertElement, Ty, i);
+        Cost += thisT()->getVectorInstrCost(Instruction::InsertElement, Ty, i,
+                                            nullptr, nullptr);
       if (Extract)
-        Cost += thisT()->getVectorInstrCost(Instruction::ExtractElement, Ty, i);
+        Cost += thisT()->getVectorInstrCost(Instruction::ExtractElement, Ty, i,
+                                            nullptr, nullptr);
     }
 
     return Cost;
@@ -1123,7 +1129,7 @@
   InstructionCost getExtractWithExtendCost(unsigned Opcode, Type *Dst,
                                            VectorType *VecTy, unsigned Index) {
     return thisT()->getVectorInstrCost(Instruction::ExtractElement, VecTy,
-                                       Index) +
+                                       Index, nullptr, nullptr) +
            thisT()->getCastInstrCost(Opcode, Dst, VecTy->getElementType(),
                                      TTI::CastContextHint::None,
                                      TTI::TCK_RecipThroughput);
@@ -1184,14 +1190,20 @@
     return 1;
   }
 
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index) {
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1) {
     return getRegUsageForType(Val->getScalarType());
   }
 
   InstructionCost getVectorInstrCost(const Instruction &I, Type *Val,
                                      unsigned Index) {
-    return thisT()->getVectorInstrCost(I.getOpcode(), Val, Index);
+    Value *Op0 = nullptr;
+    Value *Op1 = nullptr;
+    if (auto *IE = dyn_cast<InsertElementInst>(&I)) {
+      Op0 = IE->getOperand(0);
+      Op1 = IE->getOperand(1);
+    }
+    return thisT()->getVectorInstrCost(I.getOpcode(), Val, Index, Op0, Op1);
   }
 
   InstructionCost getReplicationShuffleCost(Type *EltTy, int ReplicationFactor,
@@ -2246,7 +2258,8 @@
     ArithCost +=
         NumReduxLevels * thisT()->getArithmeticInstrCost(Opcode, Ty, CostKind);
     return ShuffleCost + ArithCost +
-           thisT()->getVectorInstrCost(Instruction::ExtractElement, Ty, 0);
+           thisT()->getVectorInstrCost(Instruction::ExtractElement, Ty, 0,
+                                       nullptr, nullptr);
   }
 
   /// Try to calculate the cost of performing strict (in-order) reductions,
@@ -2353,7 +2366,8 @@
     // The last min/max should be in vector registers and we counted it above.
     // So just need a single extractelement.
     return ShuffleCost + MinMaxCost +
-           thisT()->getVectorInstrCost(Instruction::ExtractElement, Ty, 0);
+           thisT()->getVectorInstrCost(Instruction::ExtractElement, Ty, 0,
+                                       nullptr, nullptr);
   }
 
   InstructionCost getExtendedReductionCost(unsigned Opcode, bool IsUnsigned,
diff --git a/llvm/lib/Analysis/TargetTransformInfo.cpp b/llvm/lib/Analysis/TargetTransformInfo.cpp
--- a/llvm/lib/Analysis/TargetTransformInfo.cpp
+++ b/llvm/lib/Analysis/TargetTransformInfo.cpp
@@ -897,13 +897,13 @@
   return Cost;
 }
 
-InstructionCost TargetTransformInfo::getVectorInstrCost(unsigned Opcode,
-                                                        Type *Val,
-                                                        unsigned Index) const {
+InstructionCost TargetTransformInfo::getVectorInstrCost(
+    unsigned Opcode, Type *Val, unsigned Index, Value *Op0, Value *Op1) const {
   // FIXME: Assert that Opcode is either InsertElement or ExtractElement.
   // This is mentioned in the interface description and respected by all
   // callers, but never asserted upon.
-  InstructionCost Cost = TTIImpl->getVectorInstrCost(Opcode, Val, Index);
+  InstructionCost Cost =
+      TTIImpl->getVectorInstrCost(Opcode, Val, Index, Op0, Op1);
   assert(Cost >= 0 && "TTI should not produce negative costs!");
   return Cost;
 }
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
@@ -169,8 +169,8 @@
   InstructionCost getCFInstrCost(unsigned Opcode, TTI::TargetCostKind CostKind,
                                  const Instruction *I = nullptr);
 
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index);
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1);
   InstructionCost getVectorInstrCost(const Instruction &I, Type *Val,
                                      unsigned Index);
 
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -2034,8 +2034,8 @@
 
   // Get the cost for the extract. We compute the cost (if any) for the extend
   // below.
-  InstructionCost Cost =
-      getVectorInstrCost(Instruction::ExtractElement, VecTy, Index);
+  InstructionCost Cost = getVectorInstrCost(Instruction::ExtractElement, VecTy,
+                                            Index, nullptr, nullptr);
 
   // Legalize the types.
   auto VecLT = getTypeLegalizationCost(VecTy);
@@ -2128,7 +2128,8 @@
 }
 
 InstructionCost AArch64TTIImpl::getVectorInstrCost(unsigned Opcode, Type *Val,
-                                                   unsigned Index) {
+                                                   unsigned Index, Value *Op0,
+                                                   Value *Op1) {
   return getVectorInstrCostHelper(Val, Index, false /* HasRealUse */);
 }
 
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
@@ -162,7 +162,7 @@
 
   using BaseT::getVectorInstrCost;
   InstructionCost getVectorInstrCost(unsigned Opcode, Type *ValTy,
-                                     unsigned Index);
+                                     unsigned Index, Value *Op0, Value *Op1);
 
   bool isReadRegisterSourceOfDivergence(const IntrinsicInst *ReadReg) const;
   bool isSourceOfDivergence(const Value *V) const;
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
@@ -790,7 +790,8 @@
 }
 
 InstructionCost GCNTTIImpl::getVectorInstrCost(unsigned Opcode, Type *ValTy,
-                                               unsigned Index) {
+                                               unsigned Index, Value *Op0,
+                                               Value *Op1) {
   switch (Opcode) {
   case Instruction::ExtractElement:
   case Instruction::InsertElement: {
@@ -799,7 +800,7 @@
     if (EltSize < 32) {
       if (EltSize == 16 && Index == 0 && ST->has16BitInsts())
         return 0;
-      return BaseT::getVectorInstrCost(Opcode, ValTy, Index);
+      return BaseT::getVectorInstrCost(Opcode, ValTy, Index, Op0, Op1);
     }
 
     // Extracts are just reads of a subregister, so are free. Inserts are
@@ -810,7 +811,7 @@
     return Index == ~0u ? 2 : 0;
   }
   default:
-    return BaseT::getVectorInstrCost(Opcode, ValTy, Index);
+    return BaseT::getVectorInstrCost(Opcode, ValTy, Index, Op0, Op1);
   }
 }
 
diff --git a/llvm/lib/Target/AMDGPU/R600TargetTransformInfo.h b/llvm/lib/Target/AMDGPU/R600TargetTransformInfo.h
--- a/llvm/lib/Target/AMDGPU/R600TargetTransformInfo.h
+++ b/llvm/lib/Target/AMDGPU/R600TargetTransformInfo.h
@@ -62,7 +62,7 @@
                                  const Instruction *I = nullptr);
   using BaseT::getVectorInstrCost;
   InstructionCost getVectorInstrCost(unsigned Opcode, Type *ValTy,
-                                     unsigned Index);
+                                     unsigned Index, Value *Op0, Value *Op1);
 };
 
 } // end namespace llvm
diff --git a/llvm/lib/Target/AMDGPU/R600TargetTransformInfo.cpp b/llvm/lib/Target/AMDGPU/R600TargetTransformInfo.cpp
--- a/llvm/lib/Target/AMDGPU/R600TargetTransformInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/R600TargetTransformInfo.cpp
@@ -108,14 +108,15 @@
 }
 
 InstructionCost R600TTIImpl::getVectorInstrCost(unsigned Opcode, Type *ValTy,
-                                                unsigned Index) {
+                                                unsigned Index, Value *Op0,
+                                                Value *Op1) {
   switch (Opcode) {
   case Instruction::ExtractElement:
   case Instruction::InsertElement: {
     unsigned EltSize =
         DL.getTypeSizeInBits(cast<VectorType>(ValTy)->getElementType());
     if (EltSize < 32) {
-      return BaseT::getVectorInstrCost(Opcode, ValTy, Index);
+      return BaseT::getVectorInstrCost(Opcode, ValTy, Index, Op0, Op1);
     }
 
     // Extracts are just reads of a subregister, so are free. Inserts are
@@ -126,7 +127,7 @@
     return Index == ~0u ? 2 : 0;
   }
   default:
-    return BaseT::getVectorInstrCost(Opcode, ValTy, Index);
+    return BaseT::getVectorInstrCost(Opcode, ValTy, Index, Op0, Op1);
   }
 }
 
diff --git a/llvm/lib/Target/ARM/ARMTargetTransformInfo.h b/llvm/lib/Target/ARM/ARMTargetTransformInfo.h
--- a/llvm/lib/Target/ARM/ARMTargetTransformInfo.h
+++ b/llvm/lib/Target/ARM/ARMTargetTransformInfo.h
@@ -240,8 +240,8 @@
                                      const Instruction *I = nullptr);
 
   using BaseT::getVectorInstrCost;
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index);
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1);
 
   InstructionCost getAddressComputationCost(Type *Val, ScalarEvolution *SE,
                                             const SCEV *Ptr);
diff --git a/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp b/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp
--- a/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp
+++ b/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp
@@ -874,7 +874,8 @@
 }
 
 InstructionCost ARMTTIImpl::getVectorInstrCost(unsigned Opcode, Type *ValTy,
-                                               unsigned Index) {
+                                               unsigned Index, Value *Op0,
+                                               Value *Op1) {
   // Penalize inserting into an D-subregister. We end up with a three times
   // lower estimated throughput on swift.
   if (ST->hasSlowLoadDSubregister() && Opcode == Instruction::InsertElement &&
@@ -893,7 +894,7 @@
     if (ValTy->isVectorTy() &&
         ValTy->getScalarSizeInBits() <= 32)
       return std::max<InstructionCost>(
-          BaseT::getVectorInstrCost(Opcode, ValTy, Index), 2U);
+          BaseT::getVectorInstrCost(Opcode, ValTy, Index, Op0, Op1), 2U);
   }
 
   if (ST->hasMVEIntegerOps() && (Opcode == Instruction::InsertElement ||
@@ -906,7 +907,7 @@
     return LT.first * (ValTy->getScalarType()->isIntegerTy() ? 4 : 1);
   }
 
-  return BaseT::getVectorInstrCost(Opcode, ValTy, Index);
+  return BaseT::getVectorInstrCost(Opcode, ValTy, Index, Op0, Op1);
 }
 
 InstructionCost ARMTTIImpl::getCmpSelInstrCost(unsigned Opcode, Type *ValTy,
diff --git a/llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.h b/llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.h
--- a/llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.h
+++ b/llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.h
@@ -154,8 +154,8 @@
                                    TTI::TargetCostKind CostKind,
                                    const Instruction *I = nullptr);
   using BaseT::getVectorInstrCost;
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index);
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1);
 
   InstructionCost getCFInstrCost(unsigned Opcode, TTI::TargetCostKind CostKind,
                                  const Instruction *I = nullptr) {
diff --git a/llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.cpp b/llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.cpp
--- a/llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.cpp
+++ b/llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.cpp
@@ -329,7 +329,8 @@
 }
 
 InstructionCost HexagonTTIImpl::getVectorInstrCost(unsigned Opcode, Type *Val,
-                                                   unsigned Index) {
+                                                   unsigned Index, Value *Op0,
+                                                   Value *Op1) {
   Type *ElemTy = Val->isVectorTy() ? cast<VectorType>(Val)->getElementType()
                                    : Val;
   if (Opcode == Instruction::InsertElement) {
@@ -338,7 +339,8 @@
     if (ElemTy->isIntegerTy(32))
       return Cost;
     // If it's not a 32-bit value, there will need to be an extract.
-    return Cost + getVectorInstrCost(Instruction::ExtractElement, Val, Index);
+    return Cost + getVectorInstrCost(Instruction::ExtractElement, Val, Index,
+                                     Op0, Op1);
   }
 
   if (Opcode == Instruction::ExtractElement)
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
@@ -126,8 +126,8 @@
                                      TTI::TargetCostKind CostKind,
                                      const Instruction *I = nullptr);
   using BaseT::getVectorInstrCost;
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index);
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1);
   InstructionCost
   getMemoryOpCost(unsigned Opcode, Type *Src, MaybeAlign Alignment,
                   unsigned AddressSpace, TTI::TargetCostKind CostKind,
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
@@ -717,7 +717,8 @@
 }
 
 InstructionCost PPCTTIImpl::getVectorInstrCost(unsigned Opcode, Type *Val,
-                                               unsigned Index) {
+                                               unsigned Index, Value *Op0,
+                                               Value *Op1) {
   assert(Val->isVectorTy() && "This must be a vector type");
 
   int ISD = TLI->InstructionOpcodeToISD(Opcode);
@@ -727,7 +728,8 @@
   if (!CostFactor.isValid())
     return InstructionCost::getMax();
 
-  InstructionCost Cost = BaseT::getVectorInstrCost(Opcode, Val, Index);
+  InstructionCost Cost =
+      BaseT::getVectorInstrCost(Opcode, Val, Index, Op0, Op1);
   Cost *= CostFactor;
 
   if (ST->hasVSX() && Val->getScalarType()->isDoubleTy()) {
@@ -869,7 +871,8 @@
   if (Src->isVectorTy() && Opcode == Instruction::Store)
     for (int i = 0, e = cast<FixedVectorType>(Src)->getNumElements(); i < e;
          ++i)
-      Cost += getVectorInstrCost(Instruction::ExtractElement, Src, i);
+      Cost += getVectorInstrCost(Instruction::ExtractElement, Src, i, nullptr,
+                                 nullptr);
 
   return Cost;
 }
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
@@ -157,8 +157,8 @@
                                      const Instruction *I = nullptr);
 
   using BaseT::getVectorInstrCost;
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index);
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1);
 
   InstructionCost getArithmeticInstrCost(
       unsigned Opcode, Type *Ty, TTI::TargetCostKind CostKind,
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -1131,12 +1131,13 @@
 }
 
 InstructionCost RISCVTTIImpl::getVectorInstrCost(unsigned Opcode, Type *Val,
-                                                 unsigned Index) {
+                                                 unsigned Index, Value *Op0,
+                                                 Value *Op1) {
   assert(Val->isVectorTy() && "This must be a vector type");
 
   if (Opcode != Instruction::ExtractElement &&
       Opcode != Instruction::InsertElement)
-    return BaseT::getVectorInstrCost(Opcode, Val, Index);
+    return BaseT::getVectorInstrCost(Opcode, Val, Index, Op0, Op1);
 
   // Legalize the type.
   std::pair<InstructionCost, MVT> LT = getTypeLegalizationCost(Val);
@@ -1150,7 +1151,7 @@
     return LT.first;
 
   if (!isTypeLegal(Val))
-    return BaseT::getVectorInstrCost(Opcode, Val, Index);
+    return BaseT::getVectorInstrCost(Opcode, Val, Index, Op0, Op1);
 
   // In RVV, we could use vslidedown + vmv.x.s to extract element from vector
   // and vslideup + vmv.s.x to insert element to vector.
diff --git a/llvm/lib/Target/SystemZ/SystemZTargetTransformInfo.h b/llvm/lib/Target/SystemZ/SystemZTargetTransformInfo.h
--- a/llvm/lib/Target/SystemZ/SystemZTargetTransformInfo.h
+++ b/llvm/lib/Target/SystemZ/SystemZTargetTransformInfo.h
@@ -107,8 +107,8 @@
                                      TTI::TargetCostKind CostKind,
                                      const Instruction *I = nullptr);
   using BaseT::getVectorInstrCost;
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index);
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1);
   bool isFoldableLoad(const LoadInst *Ld, const Instruction *&FoldedValue);
   InstructionCost
   getMemoryOpCost(unsigned Opcode, Type *Src, MaybeAlign Alignment,
diff --git a/llvm/lib/Target/SystemZ/SystemZTargetTransformInfo.cpp b/llvm/lib/Target/SystemZ/SystemZTargetTransformInfo.cpp
--- a/llvm/lib/Target/SystemZ/SystemZTargetTransformInfo.cpp
+++ b/llvm/lib/Target/SystemZ/SystemZTargetTransformInfo.cpp
@@ -996,7 +996,8 @@
 }
 
 InstructionCost SystemZTTIImpl::getVectorInstrCost(unsigned Opcode, Type *Val,
-                                                   unsigned Index) {
+                                                   unsigned Index, Value *Op0,
+                                                   Value *Op1) {
   // vlvgp will insert two grs into a vector register, so only count half the
   // number of instructions.
   if (Opcode == Instruction::InsertElement && Val->isIntOrIntVectorTy(64))
@@ -1012,7 +1013,7 @@
     return Cost;
   }
 
-  return BaseT::getVectorInstrCost(Opcode, Val, Index);
+  return BaseT::getVectorInstrCost(Opcode, Val, Index, Op0, Op1);
 }
 
 // Check if a load may be folded as a memory operand in its user.
diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyTargetTransformInfo.h b/llvm/lib/Target/WebAssembly/WebAssemblyTargetTransformInfo.h
--- a/llvm/lib/Target/WebAssembly/WebAssemblyTargetTransformInfo.h
+++ b/llvm/lib/Target/WebAssembly/WebAssemblyTargetTransformInfo.h
@@ -66,8 +66,8 @@
       ArrayRef<const Value *> Args = ArrayRef<const Value *>(),
       const Instruction *CxtI = nullptr);
   using BaseT::getVectorInstrCost;
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index);
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1);
 
   /// @}
 
diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyTargetTransformInfo.cpp b/llvm/lib/Target/WebAssembly/WebAssemblyTargetTransformInfo.cpp
--- a/llvm/lib/Target/WebAssembly/WebAssemblyTargetTransformInfo.cpp
+++ b/llvm/lib/Target/WebAssembly/WebAssemblyTargetTransformInfo.cpp
@@ -82,9 +82,10 @@
 
 InstructionCost WebAssemblyTTIImpl::getVectorInstrCost(unsigned Opcode,
                                                        Type *Val,
-                                                       unsigned Index) {
+                                                       unsigned Index,
+                                                       Value *Op0, Value *Op1) {
   InstructionCost Cost =
-      BasicTTIImplBase::getVectorInstrCost(Opcode, Val, Index);
+      BasicTTIImplBase::getVectorInstrCost(Opcode, Val, Index, Op0, Op1);
 
   // SIMD128's insert/extract currently only take constant indices.
   if (Index == -1u)
diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.h b/llvm/lib/Target/X86/X86TargetTransformInfo.h
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.h
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.h
@@ -147,8 +147,8 @@
                                      TTI::TargetCostKind CostKind,
                                      const Instruction *I = nullptr);
   using BaseT::getVectorInstrCost;
-  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
-                                     unsigned Index);
+  InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index,
+                                     Value *Op0, Value *Op1);
   InstructionCost getScalarizationOverhead(VectorType *Ty,
                                            const APInt &DemandedElts,
                                            bool Insert, bool Extract);
diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -4257,7 +4257,8 @@
 }
 
 InstructionCost X86TTIImpl::getVectorInstrCost(unsigned Opcode, Type *Val,
-                                               unsigned Index) {
+                                               unsigned Index, Value *Op0,
+                                               Value *Op1) {
   static const CostTblEntry SLMCostTbl[] = {
      { ISD::EXTRACT_VECTOR_ELT,       MVT::i8,      4 },
      { ISD::EXTRACT_VECTOR_ELT,       MVT::i16,     4 },
@@ -4337,6 +4338,18 @@
       if (ScalarType->isFloatingPointTy())
         return RegisterFileMoveCost;
 
+      if (Opcode == Instruction::InsertElement &&
+          isa_and_nonnull<UndefValue>(Op0)) {
+        // Consider the gather cost to be cheap.
+        if (isa_and_nonnull<LoadInst>(Op1))
+          return RegisterFileMoveCost;
+        // mov constang-to-GPR + movd/movq GPR -> XMM.
+        if (Op1->getType()->isIntegerTy() && isa_and_nonnull<Constant>(Op1))
+          return 2 + RegisterFileMoveCost;
+        // Assume movd/movq GPR -> XMM is relatively cheap on all targets.
+        return 1 + RegisterFileMoveCost;
+      }
+
       // Assume movd/movq XMM -> GPR is relatively cheap on all targets.
       if (ScalarType->isIntegerTy() && Opcode == Instruction::ExtractElement)
         return 1 + RegisterFileMoveCost;
@@ -4383,7 +4396,8 @@
   if (Opcode == Instruction::ExtractElement && ScalarType->isPointerTy())
     RegisterFileMoveCost += 1;
 
-  return BaseT::getVectorInstrCost(Opcode, Val, Index) + RegisterFileMoveCost;
+  return BaseT::getVectorInstrCost(Opcode, Val, Index, Op0, Op1) +
+         RegisterFileMoveCost;
 }
 
 InstructionCost X86TTIImpl::getScalarizationOverhead(VectorType *Ty,
@@ -5155,7 +5169,8 @@
   }
 
   // Add the final extract element to the cost.
-  return ReductionCost + getVectorInstrCost(Instruction::ExtractElement, Ty, 0);
+  return ReductionCost + getVectorInstrCost(Instruction::ExtractElement, Ty, 0,
+                                            nullptr, nullptr);
 }
 
 InstructionCost X86TTIImpl::getMinMaxCost(Type *Ty, Type *CondTy,
@@ -5455,7 +5470,8 @@
   }
 
   // Add the final extract element to the cost.
-  return MinMaxCost + getVectorInstrCost(Instruction::ExtractElement, Ty, 0);
+  return MinMaxCost + getVectorInstrCost(Instruction::ExtractElement, Ty, 0,
+                                         nullptr, nullptr);
 }
 
 /// Calculate the cost of materializing a 64-bit value. This helper
diff --git a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
--- a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -6724,9 +6724,24 @@
       // broadcast.
       assert(VecTy == FinalVecTy &&
              "No reused scalars expected for broadcast.");
-      return TTI->getShuffleCost(TargetTransformInfo::SK_Broadcast, VecTy,
-                                 /*Mask=*/std::nullopt, CostKind, /*Index=*/0,
-                                 /*SubTp=*/nullptr, /*Args=*/VL[0]);
+      const auto *It =
+          find_if(VL, [](Value *V) { return !isa<UndefValue>(V); });
+      // If all values are undefs - consider cost free.
+      if (It == VL.end())
+        return TTI::TCC_Free;
+      // Add broadcast for non-identity shuffle only.
+      bool NeedShuffle =
+          VL.front() != *It || !all_of(VL.drop_front(), UndefValue::classof);
+      InstructionCost InsertCost =
+          TTI->getVectorInstrCost(Instruction::InsertElement, VecTy,
+                                  /*Index=*/0, PoisonValue::get(VecTy), *It);
+      return InsertCost + (NeedShuffle
+                               ? TTI->getShuffleCost(
+                                     TargetTransformInfo::SK_Broadcast, VecTy,
+                                     /*Mask=*/std::nullopt, CostKind,
+                                     /*Index=*/0,
+                                     /*SubTp=*/nullptr, /*Args=*/VL[0])
+                               : TTI::TCC_Free);
     }
     InstructionCost ReuseShuffleCost = 0;
     if (NeedToShuffleReuses)
diff --git a/llvm/test/Analysis/CostModel/X86/loop_v2-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/loop_v2-inseltpoison.ll
--- a/llvm/test/Analysis/CostModel/X86/loop_v2-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/loop_v2-inseltpoison.ll
@@ -20,7 +20,7 @@
   %5 = extractelement <2 x i64> %2, i32 1
   %6 = getelementptr inbounds i32, ptr %A, i64 %5
   %7 = load i32, ptr %4, align 4
-  ;CHECK: cost of 1 {{.*}} insert
+  ;CHECK: cost of 0 {{.*}} insert
   %8 = insertelement <2 x i32> poison, i32 %7, i32 0
   %9 = load i32, ptr %6, align 4
   ;CHECK: cost of 1 {{.*}} insert
diff --git a/llvm/test/Analysis/CostModel/X86/loop_v2.ll b/llvm/test/Analysis/CostModel/X86/loop_v2.ll
--- a/llvm/test/Analysis/CostModel/X86/loop_v2.ll
+++ b/llvm/test/Analysis/CostModel/X86/loop_v2.ll
@@ -20,7 +20,7 @@
   %5 = extractelement <2 x i64> %2, i32 1
   %6 = getelementptr inbounds i32, ptr %A, i64 %5
   %7 = load i32, ptr %4, align 4
-  ;CHECK: cost of 1 {{.*}} insert
+  ;CHECK: cost of 0 {{.*}} insert
   %8 = insertelement <2 x i32> undef, i32 %7, i32 0
   %9 = load i32, ptr %6, align 4
   ;CHECK: cost of 1 {{.*}} insert
diff --git a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll
--- a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll
@@ -1907,7 +1907,7 @@
 
 define <16 x float> @test_gather_16f32_const_mask2(ptr %base, <16 x i32> %ind) {
 ; SSE2-LABEL: 'test_gather_16f32_const_mask2'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %broadcast.splatinsert = insertelement <16 x ptr> poison, ptr %base, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x ptr> poison, ptr %base, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x ptr> %broadcast.splatinsert, <16 x ptr> poison, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr float, <16 x ptr> %broadcast.splat, <16 x i64> %sext_ind
@@ -1966,7 +1966,7 @@
 
 define void @test_scatter_16i32(ptr %base, <16 x i32> %ind, i16 %mask, <16 x i32>%val) {
 ; SSE2-LABEL: 'test_scatter_16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %broadcast.splatinsert = insertelement <16 x ptr> poison, ptr %base, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x ptr> poison, ptr %base, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x ptr> %broadcast.splatinsert, <16 x ptr> poison, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr i32, <16 x ptr> %broadcast.splat, <16 x i32> %ind
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %imask = bitcast i16 %mask to <16 x i1>
diff --git a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll
--- a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll
@@ -1907,7 +1907,7 @@
 
 define <16 x float> @test_gather_16f32_const_mask2(ptr %base, <16 x i32> %ind) {
 ; SSE2-LABEL: 'test_gather_16f32_const_mask2'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %broadcast.splatinsert = insertelement <16 x ptr> undef, ptr %base, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x ptr> undef, ptr %base, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x ptr> %broadcast.splatinsert, <16 x ptr> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr float, <16 x ptr> %broadcast.splat, <16 x i64> %sext_ind
@@ -1966,7 +1966,7 @@
 
 define void @test_scatter_16i32(ptr %base, <16 x i32> %ind, i16 %mask, <16 x i32>%val) {
 ; SSE2-LABEL: 'test_scatter_16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %broadcast.splatinsert = insertelement <16 x ptr> undef, ptr %base, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x ptr> undef, ptr %base, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x ptr> %broadcast.splatinsert, <16 x ptr> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr i32, <16 x ptr> %broadcast.splat, <16 x i32> %ind
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %imask = bitcast i16 %mask to <16 x i1>
diff --git a/llvm/test/Analysis/CostModel/X86/vector-insert-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/vector-insert-inseltpoison.ll
--- a/llvm/test/Analysis/CostModel/X86/vector-insert-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/vector-insert-inseltpoison.ll
@@ -294,71 +294,71 @@
 ;
 ; SSE4-LABEL: 'insert_i64'
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i64_a = insertelement <2 x i64> poison, i64 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = insertelement <2 x i64> poison, i64 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i64_0 = insertelement <2 x i64> poison, i64 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = insertelement <2 x i64> poison, i64 undef, i32 1
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i64_a = insertelement <4 x i64> poison, i64 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = insertelement <4 x i64> poison, i64 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64_0 = insertelement <4 x i64> poison, i64 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_3 = insertelement <4 x i64> poison, i64 undef, i32 3
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v8i64_a = insertelement <8 x i64> poison, i64 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = insertelement <8 x i64> poison, i64 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_0 = insertelement <8 x i64> poison, i64 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_3 = insertelement <8 x i64> poison, i64 undef, i32 3
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_4 = insertelement <8 x i64> poison, i64 undef, i32 4
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_4 = insertelement <8 x i64> poison, i64 undef, i32 4
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_7 = insertelement <8 x i64> poison, i64 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX-LABEL: 'insert_i64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i64_a = insertelement <2 x i64> poison, i64 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = insertelement <2 x i64> poison, i64 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i64_0 = insertelement <2 x i64> poison, i64 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = insertelement <2 x i64> poison, i64 undef, i32 1
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i64_a = insertelement <4 x i64> poison, i64 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = insertelement <4 x i64> poison, i64 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64_0 = insertelement <4 x i64> poison, i64 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i64_3 = insertelement <4 x i64> poison, i64 undef, i32 3
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i64_a = insertelement <8 x i64> poison, i64 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = insertelement <8 x i64> poison, i64 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_0 = insertelement <8 x i64> poison, i64 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_3 = insertelement <8 x i64> poison, i64 undef, i32 3
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_4 = insertelement <8 x i64> poison, i64 undef, i32 4
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_4 = insertelement <8 x i64> poison, i64 undef, i32 4
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_7 = insertelement <8 x i64> poison, i64 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'insert_i64'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i64_a = insertelement <2 x i64> poison, i64 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = insertelement <2 x i64> poison, i64 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i64_0 = insertelement <2 x i64> poison, i64 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = insertelement <2 x i64> poison, i64 undef, i32 1
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i64_a = insertelement <4 x i64> poison, i64 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = insertelement <4 x i64> poison, i64 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64_0 = insertelement <4 x i64> poison, i64 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i64_3 = insertelement <4 x i64> poison, i64 undef, i32 3
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_a = insertelement <8 x i64> poison, i64 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = insertelement <8 x i64> poison, i64 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_0 = insertelement <8 x i64> poison, i64 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_3 = insertelement <8 x i64> poison, i64 undef, i32 3
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_4 = insertelement <8 x i64> poison, i64 undef, i32 4
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i64_4 = insertelement <8 x i64> poison, i64 undef, i32 4
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_7 = insertelement <8 x i64> poison, i64 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SLM-LABEL: 'insert_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i64_a = insertelement <2 x i64> poison, i64 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = insertelement <2 x i64> poison, i64 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i64_0 = insertelement <2 x i64> poison, i64 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = insertelement <2 x i64> poison, i64 undef, i32 1
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i64_a = insertelement <4 x i64> poison, i64 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = insertelement <4 x i64> poison, i64 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64_0 = insertelement <4 x i64> poison, i64 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_3 = insertelement <4 x i64> poison, i64 undef, i32 3
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v8i64_a = insertelement <8 x i64> poison, i64 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = insertelement <8 x i64> poison, i64 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_0 = insertelement <8 x i64> poison, i64 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_3 = insertelement <8 x i64> poison, i64 undef, i32 3
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_4 = insertelement <8 x i64> poison, i64 undef, i32 4
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_4 = insertelement <8 x i64> poison, i64 undef, i32 4
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_7 = insertelement <8 x i64> poison, i64 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'insert_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i64_a = insertelement <2 x i64> poison, i64 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = insertelement <2 x i64> poison, i64 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i64_0 = insertelement <2 x i64> poison, i64 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = insertelement <2 x i64> poison, i64 undef, i32 1
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i64_a = insertelement <4 x i64> poison, i64 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = insertelement <4 x i64> poison, i64 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64_0 = insertelement <4 x i64> poison, i64 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_3 = insertelement <4 x i64> poison, i64 undef, i32 3
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v8i64_a = insertelement <8 x i64> poison, i64 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = insertelement <8 x i64> poison, i64 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_0 = insertelement <8 x i64> poison, i64 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_3 = insertelement <8 x i64> poison, i64 undef, i32 3
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_4 = insertelement <8 x i64> poison, i64 undef, i32 4
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_4 = insertelement <8 x i64> poison, i64 undef, i32 4
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_7 = insertelement <8 x i64> poison, i64 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -382,153 +382,153 @@
 define i32 @insert_i32(i32 %arg) {
 ; SSE2-LABEL: 'insert_i32'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> poison, i32 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_1 = insertelement <2 x i32> poison, i32 undef, i32 1
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> poison, i32 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_3 = insertelement <4 x i32> poison, i32 undef, i32 3
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> poison, i32 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_3 = insertelement <8 x i32> poison, i32 undef, i32 3
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_7 = insertelement <8 x i32> poison, i32 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> poison, i32 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_3 = insertelement <16 x i32> poison, i32 undef, i32 3
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_15 = insertelement <16 x i32> poison, i32 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE3-LABEL: 'insert_i32'
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> poison, i32 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_1 = insertelement <2 x i32> poison, i32 undef, i32 1
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> poison, i32 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_3 = insertelement <4 x i32> poison, i32 undef, i32 3
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> poison, i32 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_3 = insertelement <8 x i32> poison, i32 undef, i32 3
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_7 = insertelement <8 x i32> poison, i32 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> poison, i32 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_3 = insertelement <16 x i32> poison, i32 undef, i32 3
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_15 = insertelement <16 x i32> poison, i32 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'insert_i32'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> poison, i32 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_1 = insertelement <2 x i32> poison, i32 undef, i32 1
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> poison, i32 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_3 = insertelement <4 x i32> poison, i32 undef, i32 3
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> poison, i32 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_3 = insertelement <8 x i32> poison, i32 undef, i32 3
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_7 = insertelement <8 x i32> poison, i32 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> poison, i32 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_3 = insertelement <16 x i32> poison, i32 undef, i32 3
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_15 = insertelement <16 x i32> poison, i32 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'insert_i32'
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> poison, i32 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = insertelement <2 x i32> poison, i32 undef, i32 1
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> poison, i32 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_3 = insertelement <4 x i32> poison, i32 undef, i32 3
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> poison, i32 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_3 = insertelement <8 x i32> poison, i32 undef, i32 3
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_7 = insertelement <8 x i32> poison, i32 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> poison, i32 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_3 = insertelement <16 x i32> poison, i32 undef, i32 3
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_15 = insertelement <16 x i32> poison, i32 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX-LABEL: 'insert_i32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> poison, i32 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = insertelement <2 x i32> poison, i32 undef, i32 1
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> poison, i32 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_3 = insertelement <4 x i32> poison, i32 undef, i32 3
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_a = insertelement <8 x i32> poison, i32 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_3 = insertelement <8 x i32> poison, i32 undef, i32 3
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_7 = insertelement <8 x i32> poison, i32 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v16i32_a = insertelement <16 x i32> poison, i32 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_3 = insertelement <16 x i32> poison, i32 undef, i32 3
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_15 = insertelement <16 x i32> poison, i32 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'insert_i32'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> poison, i32 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = insertelement <2 x i32> poison, i32 undef, i32 1
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> poison, i32 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_3 = insertelement <4 x i32> poison, i32 undef, i32 3
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_a = insertelement <8 x i32> poison, i32 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_3 = insertelement <8 x i32> poison, i32 undef, i32 3
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_7 = insertelement <8 x i32> poison, i32 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_a = insertelement <16 x i32> poison, i32 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_3 = insertelement <16 x i32> poison, i32 undef, i32 3
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_15 = insertelement <16 x i32> poison, i32 undef, i32 15
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SLM-LABEL: 'insert_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> poison, i32 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = insertelement <2 x i32> poison, i32 undef, i32 1
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> poison, i32 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_3 = insertelement <4 x i32> poison, i32 undef, i32 3
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> poison, i32 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_3 = insertelement <8 x i32> poison, i32 undef, i32 3
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_7 = insertelement <8 x i32> poison, i32 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> poison, i32 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_3 = insertelement <16 x i32> poison, i32 undef, i32 3
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_15 = insertelement <16 x i32> poison, i32 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'insert_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> poison, i32 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> poison, i32 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = insertelement <2 x i32> poison, i32 undef, i32 1
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> poison, i32 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> poison, i32 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_3 = insertelement <4 x i32> poison, i32 undef, i32 3
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> poison, i32 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> poison, i32 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_3 = insertelement <8 x i32> poison, i32 undef, i32 3
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> poison, i32 undef, i32 4
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_7 = insertelement <8 x i32> poison, i32 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> poison, i32 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> poison, i32 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_3 = insertelement <16 x i32> poison, i32 undef, i32 3
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> poison, i32 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_15 = insertelement <16 x i32> poison, i32 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -558,76 +558,76 @@
 define i32 @insert_i16(i32 %arg) {
 ; SSE-LABEL: 'insert_i16'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i16_a = insertelement <2 x i16> poison, i16 undef, i32 %arg
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_0 = insertelement <2 x i16> poison, i16 undef, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i16_0 = insertelement <2 x i16> poison, i16 undef, i32 0
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_1 = insertelement <2 x i16> poison, i16 undef, i32 1
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i16_a = insertelement <4 x i16> poison, i16 undef, i32 %arg
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_0 = insertelement <4 x i16> poison, i16 undef, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i16_0 = insertelement <4 x i16> poison, i16 undef, i32 0
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_3 = insertelement <4 x i16> poison, i16 undef, i32 3
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i16_a = insertelement <8 x i16> poison, i16 undef, i32 %arg
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_0 = insertelement <8 x i16> poison, i16 undef, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i16_0 = insertelement <8 x i16> poison, i16 undef, i32 0
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_7 = insertelement <8 x i16> poison, i16 undef, i32 7
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v16i16_a = insertelement <16 x i16> poison, i16 undef, i32 %arg
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_0 = insertelement <16 x i16> poison, i16 undef, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_0 = insertelement <16 x i16> poison, i16 undef, i32 0
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_7 = insertelement <16 x i16> poison, i16 undef, i32 7
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_8 = insertelement <16 x i16> poison, i16 undef, i32 8
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_8 = insertelement <16 x i16> poison, i16 undef, i32 8
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_15 = insertelement <16 x i16> poison, i16 undef, i32 15
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v32i16_a = insertelement <32 x i16> poison, i16 undef, i32 %arg
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_0 = insertelement <32 x i16> poison, i16 undef, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_0 = insertelement <32 x i16> poison, i16 undef, i32 0
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_7 = insertelement <32 x i16> poison, i16 undef, i32 7
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_8 = insertelement <32 x i16> poison, i16 undef, i32 8
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_8 = insertelement <32 x i16> poison, i16 undef, i32 8
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_15 = insertelement <32 x i16> poison, i16 undef, i32 15
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_16 = insertelement <32 x i16> poison, i16 undef, i32 16
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_24 = insertelement <32 x i16> poison, i16 undef, i32 24
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_16 = insertelement <32 x i16> poison, i16 undef, i32 16
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_24 = insertelement <32 x i16> poison, i16 undef, i32 24
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_31 = insertelement <32 x i16> poison, i16 undef, i32 31
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX-LABEL: 'insert_i16'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i16_a = insertelement <2 x i16> poison, i16 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_0 = insertelement <2 x i16> poison, i16 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i16_0 = insertelement <2 x i16> poison, i16 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_1 = insertelement <2 x i16> poison, i16 undef, i32 1
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i16_a = insertelement <4 x i16> poison, i16 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_0 = insertelement <4 x i16> poison, i16 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i16_0 = insertelement <4 x i16> poison, i16 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_3 = insertelement <4 x i16> poison, i16 undef, i32 3
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i16_a = insertelement <8 x i16> poison, i16 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_0 = insertelement <8 x i16> poison, i16 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i16_0 = insertelement <8 x i16> poison, i16 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_7 = insertelement <8 x i16> poison, i16 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_a = insertelement <16 x i16> poison, i16 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_0 = insertelement <16 x i16> poison, i16 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_0 = insertelement <16 x i16> poison, i16 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_7 = insertelement <16 x i16> poison, i16 undef, i32 7
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_8 = insertelement <16 x i16> poison, i16 undef, i32 8
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i16_8 = insertelement <16 x i16> poison, i16 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_15 = insertelement <16 x i16> poison, i16 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i16_a = insertelement <32 x i16> poison, i16 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_0 = insertelement <32 x i16> poison, i16 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_0 = insertelement <32 x i16> poison, i16 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_7 = insertelement <32 x i16> poison, i16 undef, i32 7
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_8 = insertelement <32 x i16> poison, i16 undef, i32 8
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i16_8 = insertelement <32 x i16> poison, i16 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_15 = insertelement <32 x i16> poison, i16 undef, i32 15
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_16 = insertelement <32 x i16> poison, i16 undef, i32 16
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_24 = insertelement <32 x i16> poison, i16 undef, i32 24
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_16 = insertelement <32 x i16> poison, i16 undef, i32 16
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i16_24 = insertelement <32 x i16> poison, i16 undef, i32 24
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_31 = insertelement <32 x i16> poison, i16 undef, i32 31
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'insert_i16'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i16_a = insertelement <2 x i16> poison, i16 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_0 = insertelement <2 x i16> poison, i16 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i16_0 = insertelement <2 x i16> poison, i16 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_1 = insertelement <2 x i16> poison, i16 undef, i32 1
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i16_a = insertelement <4 x i16> poison, i16 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_0 = insertelement <4 x i16> poison, i16 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i16_0 = insertelement <4 x i16> poison, i16 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_3 = insertelement <4 x i16> poison, i16 undef, i32 3
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i16_a = insertelement <8 x i16> poison, i16 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_0 = insertelement <8 x i16> poison, i16 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i16_0 = insertelement <8 x i16> poison, i16 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_7 = insertelement <8 x i16> poison, i16 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_a = insertelement <16 x i16> poison, i16 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_0 = insertelement <16 x i16> poison, i16 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_0 = insertelement <16 x i16> poison, i16 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_7 = insertelement <16 x i16> poison, i16 undef, i32 7
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_8 = insertelement <16 x i16> poison, i16 undef, i32 8
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i16_8 = insertelement <16 x i16> poison, i16 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_15 = insertelement <16 x i16> poison, i16 undef, i32 15
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_a = insertelement <32 x i16> poison, i16 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_0 = insertelement <32 x i16> poison, i16 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_0 = insertelement <32 x i16> poison, i16 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_7 = insertelement <32 x i16> poison, i16 undef, i32 7
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_8 = insertelement <32 x i16> poison, i16 undef, i32 8
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i16_8 = insertelement <32 x i16> poison, i16 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_15 = insertelement <32 x i16> poison, i16 undef, i32 15
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_16 = insertelement <32 x i16> poison, i16 undef, i32 16
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_24 = insertelement <32 x i16> poison, i16 undef, i32 24
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i16_16 = insertelement <32 x i16> poison, i16 undef, i32 16
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i16_24 = insertelement <32 x i16> poison, i16 undef, i32 24
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_31 = insertelement <32 x i16> poison, i16 undef, i32 31
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -664,265 +664,265 @@
 define i32 @insert_i8(i32 %arg) {
 ; SSE2-LABEL: 'insert_i8'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> poison, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_3 = insertelement <2 x i8> poison, i8 undef, i32 1
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> poison, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i8_3 = insertelement <4 x i8> poison, i8 undef, i32 3
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> poison, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i8_7 = insertelement <8 x i8> poison, i8 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> poison, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_8 = insertelement <16 x i8> poison, i8 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_15 = insertelement <16 x i8> poison, i8 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> poison, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_7 = insertelement <32 x i8> poison, i8 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_8 = insertelement <32 x i8> poison, i8 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_15 = insertelement <32 x i8> poison, i8 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_24 = insertelement <32 x i8> poison, i8 undef, i32 24
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_31 = insertelement <32 x i8> poison, i8 undef, i32 31
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> poison, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_7 = insertelement <64 x i8> poison, i8 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_8 = insertelement <64 x i8> poison, i8 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_15 = insertelement <64 x i8> poison, i8 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_24 = insertelement <64 x i8> poison, i8 undef, i32 24
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_31 = insertelement <64 x i8> poison, i8 undef, i32 31
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_63 = insertelement <64 x i8> poison, i8 undef, i32 63
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE3-LABEL: 'insert_i8'
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> poison, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_3 = insertelement <2 x i8> poison, i8 undef, i32 1
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> poison, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i8_3 = insertelement <4 x i8> poison, i8 undef, i32 3
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> poison, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i8_7 = insertelement <8 x i8> poison, i8 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> poison, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_8 = insertelement <16 x i8> poison, i8 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_15 = insertelement <16 x i8> poison, i8 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> poison, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_7 = insertelement <32 x i8> poison, i8 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_8 = insertelement <32 x i8> poison, i8 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_15 = insertelement <32 x i8> poison, i8 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_24 = insertelement <32 x i8> poison, i8 undef, i32 24
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_31 = insertelement <32 x i8> poison, i8 undef, i32 31
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> poison, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_7 = insertelement <64 x i8> poison, i8 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_8 = insertelement <64 x i8> poison, i8 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_15 = insertelement <64 x i8> poison, i8 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_24 = insertelement <64 x i8> poison, i8 undef, i32 24
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_31 = insertelement <64 x i8> poison, i8 undef, i32 31
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_63 = insertelement <64 x i8> poison, i8 undef, i32 63
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'insert_i8'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> poison, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v2i8_3 = insertelement <2 x i8> poison, i8 undef, i32 1
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> poison, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v4i8_3 = insertelement <4 x i8> poison, i8 undef, i32 3
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> poison, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i8_7 = insertelement <8 x i8> poison, i8 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> poison, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i8_8 = insertelement <16 x i8> poison, i8 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i8_15 = insertelement <16 x i8> poison, i8 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> poison, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_7 = insertelement <32 x i8> poison, i8 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_8 = insertelement <32 x i8> poison, i8 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_15 = insertelement <32 x i8> poison, i8 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_24 = insertelement <32 x i8> poison, i8 undef, i32 24
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_31 = insertelement <32 x i8> poison, i8 undef, i32 31
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> poison, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_7 = insertelement <64 x i8> poison, i8 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_8 = insertelement <64 x i8> poison, i8 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_15 = insertelement <64 x i8> poison, i8 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_24 = insertelement <64 x i8> poison, i8 undef, i32 24
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_31 = insertelement <64 x i8> poison, i8 undef, i32 31
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_63 = insertelement <64 x i8> poison, i8 undef, i32 63
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'insert_i8'
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> poison, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_3 = insertelement <2 x i8> poison, i8 undef, i32 1
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> poison, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_3 = insertelement <4 x i8> poison, i8 undef, i32 3
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> poison, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_7 = insertelement <8 x i8> poison, i8 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> poison, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_8 = insertelement <16 x i8> poison, i8 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_15 = insertelement <16 x i8> poison, i8 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> poison, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_7 = insertelement <32 x i8> poison, i8 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_8 = insertelement <32 x i8> poison, i8 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_15 = insertelement <32 x i8> poison, i8 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_24 = insertelement <32 x i8> poison, i8 undef, i32 24
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_31 = insertelement <32 x i8> poison, i8 undef, i32 31
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> poison, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_7 = insertelement <64 x i8> poison, i8 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_8 = insertelement <64 x i8> poison, i8 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_15 = insertelement <64 x i8> poison, i8 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_24 = insertelement <64 x i8> poison, i8 undef, i32 24
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_31 = insertelement <64 x i8> poison, i8 undef, i32 31
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_63 = insertelement <64 x i8> poison, i8 undef, i32 63
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX-LABEL: 'insert_i8'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> poison, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_3 = insertelement <2 x i8> poison, i8 undef, i32 1
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> poison, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_3 = insertelement <4 x i8> poison, i8 undef, i32 3
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> poison, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_7 = insertelement <8 x i8> poison, i8 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> poison, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_8 = insertelement <16 x i8> poison, i8 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_15 = insertelement <16 x i8> poison, i8 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_a = insertelement <32 x i8> poison, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_7 = insertelement <32 x i8> poison, i8 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_8 = insertelement <32 x i8> poison, i8 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_15 = insertelement <32 x i8> poison, i8 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_24 = insertelement <32 x i8> poison, i8 undef, i32 24
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_31 = insertelement <32 x i8> poison, i8 undef, i32 31
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v64i8_a = insertelement <64 x i8> poison, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_7 = insertelement <64 x i8> poison, i8 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_8 = insertelement <64 x i8> poison, i8 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_15 = insertelement <64 x i8> poison, i8 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_24 = insertelement <64 x i8> poison, i8 undef, i32 24
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_31 = insertelement <64 x i8> poison, i8 undef, i32 31
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_63 = insertelement <64 x i8> poison, i8 undef, i32 63
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'insert_i8'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> poison, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_3 = insertelement <2 x i8> poison, i8 undef, i32 1
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> poison, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_3 = insertelement <4 x i8> poison, i8 undef, i32 3
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> poison, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_7 = insertelement <8 x i8> poison, i8 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> poison, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_8 = insertelement <16 x i8> poison, i8 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_15 = insertelement <16 x i8> poison, i8 undef, i32 15
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_a = insertelement <32 x i8> poison, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_7 = insertelement <32 x i8> poison, i8 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_8 = insertelement <32 x i8> poison, i8 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_15 = insertelement <32 x i8> poison, i8 undef, i32 15
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_24 = insertelement <32 x i8> poison, i8 undef, i32 24
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_31 = insertelement <32 x i8> poison, i8 undef, i32 31
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_a = insertelement <64 x i8> poison, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_7 = insertelement <64 x i8> poison, i8 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_8 = insertelement <64 x i8> poison, i8 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_15 = insertelement <64 x i8> poison, i8 undef, i32 15
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_24 = insertelement <64 x i8> poison, i8 undef, i32 24
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_31 = insertelement <64 x i8> poison, i8 undef, i32 31
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_63 = insertelement <64 x i8> poison, i8 undef, i32 63
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SLM-LABEL: 'insert_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> poison, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_3 = insertelement <2 x i8> poison, i8 undef, i32 1
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> poison, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_3 = insertelement <4 x i8> poison, i8 undef, i32 3
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> poison, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_7 = insertelement <8 x i8> poison, i8 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> poison, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_8 = insertelement <16 x i8> poison, i8 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_15 = insertelement <16 x i8> poison, i8 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> poison, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_7 = insertelement <32 x i8> poison, i8 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_8 = insertelement <32 x i8> poison, i8 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_15 = insertelement <32 x i8> poison, i8 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_24 = insertelement <32 x i8> poison, i8 undef, i32 24
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_31 = insertelement <32 x i8> poison, i8 undef, i32 31
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> poison, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_7 = insertelement <64 x i8> poison, i8 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_8 = insertelement <64 x i8> poison, i8 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_15 = insertelement <64 x i8> poison, i8 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_24 = insertelement <64 x i8> poison, i8 undef, i32 24
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_31 = insertelement <64 x i8> poison, i8 undef, i32 31
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_63 = insertelement <64 x i8> poison, i8 undef, i32 63
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'insert_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> poison, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> poison, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_3 = insertelement <2 x i8> poison, i8 undef, i32 1
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> poison, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> poison, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_3 = insertelement <4 x i8> poison, i8 undef, i32 3
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> poison, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> poison, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_7 = insertelement <8 x i8> poison, i8 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> poison, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> poison, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_8 = insertelement <16 x i8> poison, i8 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_15 = insertelement <16 x i8> poison, i8 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> poison, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> poison, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_7 = insertelement <32 x i8> poison, i8 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_8 = insertelement <32 x i8> poison, i8 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_15 = insertelement <32 x i8> poison, i8 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_24 = insertelement <32 x i8> poison, i8 undef, i32 24
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_31 = insertelement <32 x i8> poison, i8 undef, i32 31
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> poison, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> poison, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_7 = insertelement <64 x i8> poison, i8 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_8 = insertelement <64 x i8> poison, i8 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_15 = insertelement <64 x i8> poison, i8 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_24 = insertelement <64 x i8> poison, i8 undef, i32 24
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_31 = insertelement <64 x i8> poison, i8 undef, i32 31
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> poison, i8 undef, i32 32
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> poison, i8 undef, i32 48
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_63 = insertelement <64 x i8> poison, i8 undef, i32 63
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -971,31 +971,31 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_1 = insertelement <2 x i1> poison, i1 undef, i32 1
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> poison, i1 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_2 = insertelement <4 x i1> poison, i1 undef, i32 2
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> poison, i1 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> poison, i1 undef, i32 4
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> poison, i1 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_8 = insertelement <16 x i1> poison, i1 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_15 = insertelement <16 x i1> poison, i1 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> poison, i1 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_7 = insertelement <32 x i1> poison, i1 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_8 = insertelement <32 x i1> poison, i1 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_15 = insertelement <32 x i1> poison, i1 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_24 = insertelement <32 x i1> poison, i1 undef, i32 24
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_31 = insertelement <32 x i1> poison, i1 undef, i32 31
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> poison, i1 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_7 = insertelement <64 x i1> poison, i1 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_8 = insertelement <64 x i1> poison, i1 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_15 = insertelement <64 x i1> poison, i1 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_24 = insertelement <64 x i1> poison, i1 undef, i32 24
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_31 = insertelement <64 x i1> poison, i1 undef, i32 31
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_63 = insertelement <64 x i1> poison, i1 undef, i32 63
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -1004,31 +1004,31 @@
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_1 = insertelement <2 x i1> poison, i1 undef, i32 1
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> poison, i1 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_2 = insertelement <4 x i1> poison, i1 undef, i32 2
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> poison, i1 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> poison, i1 undef, i32 4
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> poison, i1 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_8 = insertelement <16 x i1> poison, i1 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_15 = insertelement <16 x i1> poison, i1 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> poison, i1 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_7 = insertelement <32 x i1> poison, i1 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_8 = insertelement <32 x i1> poison, i1 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_15 = insertelement <32 x i1> poison, i1 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_24 = insertelement <32 x i1> poison, i1 undef, i32 24
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_31 = insertelement <32 x i1> poison, i1 undef, i32 31
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> poison, i1 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_7 = insertelement <64 x i1> poison, i1 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_8 = insertelement <64 x i1> poison, i1 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_15 = insertelement <64 x i1> poison, i1 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_24 = insertelement <64 x i1> poison, i1 undef, i32 24
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_31 = insertelement <64 x i1> poison, i1 undef, i32 31
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_63 = insertelement <64 x i1> poison, i1 undef, i32 63
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -1037,156 +1037,156 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_1 = insertelement <2 x i1> poison, i1 undef, i32 1
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> poison, i1 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_2 = insertelement <4 x i1> poison, i1 undef, i32 2
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> poison, i1 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> poison, i1 undef, i32 4
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> poison, i1 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i1_8 = insertelement <16 x i1> poison, i1 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i1_15 = insertelement <16 x i1> poison, i1 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> poison, i1 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_7 = insertelement <32 x i1> poison, i1 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_8 = insertelement <32 x i1> poison, i1 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_15 = insertelement <32 x i1> poison, i1 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_24 = insertelement <32 x i1> poison, i1 undef, i32 24
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_31 = insertelement <32 x i1> poison, i1 undef, i32 31
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> poison, i1 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_7 = insertelement <64 x i1> poison, i1 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_8 = insertelement <64 x i1> poison, i1 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_15 = insertelement <64 x i1> poison, i1 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_24 = insertelement <64 x i1> poison, i1 undef, i32 24
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_31 = insertelement <64 x i1> poison, i1 undef, i32 31
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_63 = insertelement <64 x i1> poison, i1 undef, i32 63
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'insert_i1'
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> poison, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> poison, i1 undef, i32 1
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> poison, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> poison, i1 undef, i32 2
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> poison, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> poison, i1 undef, i32 4
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> poison, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> poison, i1 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> poison, i1 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> poison, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> poison, i1 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> poison, i1 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> poison, i1 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_24 = insertelement <32 x i1> poison, i1 undef, i32 24
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_31 = insertelement <32 x i1> poison, i1 undef, i32 31
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> poison, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> poison, i1 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> poison, i1 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> poison, i1 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_24 = insertelement <64 x i1> poison, i1 undef, i32 24
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_31 = insertelement <64 x i1> poison, i1 undef, i32 31
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_63 = insertelement <64 x i1> poison, i1 undef, i32 63
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX-LABEL: 'insert_i1'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> poison, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> poison, i1 undef, i32 1
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> poison, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> poison, i1 undef, i32 2
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> poison, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> poison, i1 undef, i32 4
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> poison, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> poison, i1 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> poison, i1 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> poison, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> poison, i1 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> poison, i1 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> poison, i1 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_24 = insertelement <32 x i1> poison, i1 undef, i32 24
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_31 = insertelement <32 x i1> poison, i1 undef, i32 31
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> poison, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> poison, i1 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> poison, i1 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> poison, i1 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_24 = insertelement <64 x i1> poison, i1 undef, i32 24
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_31 = insertelement <64 x i1> poison, i1 undef, i32 31
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_63 = insertelement <64 x i1> poison, i1 undef, i32 63
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'insert_i1'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> poison, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> poison, i1 undef, i32 1
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> poison, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> poison, i1 undef, i32 2
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> poison, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> poison, i1 undef, i32 4
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> poison, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> poison, i1 undef, i32 8
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> poison, i1 undef, i32 15
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i1_a = insertelement <32 x i1> poison, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> poison, i1 undef, i32 7
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> poison, i1 undef, i32 8
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> poison, i1 undef, i32 15
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_24 = insertelement <32 x i1> poison, i1 undef, i32 24
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_31 = insertelement <32 x i1> poison, i1 undef, i32 31
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i1_a = insertelement <64 x i1> poison, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> poison, i1 undef, i32 7
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> poison, i1 undef, i32 8
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> poison, i1 undef, i32 15
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_24 = insertelement <64 x i1> poison, i1 undef, i32 24
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_31 = insertelement <64 x i1> poison, i1 undef, i32 31
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_63 = insertelement <64 x i1> poison, i1 undef, i32 63
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'insert_i1'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> poison, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> poison, i1 undef, i32 1
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> poison, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> poison, i1 undef, i32 2
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> poison, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> poison, i1 undef, i32 4
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> poison, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> poison, i1 undef, i32 8
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> poison, i1 undef, i32 15
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> poison, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> poison, i1 undef, i32 7
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> poison, i1 undef, i32 8
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> poison, i1 undef, i32 15
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_24 = insertelement <32 x i1> poison, i1 undef, i32 24
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_31 = insertelement <32 x i1> poison, i1 undef, i32 31
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> poison, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> poison, i1 undef, i32 7
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> poison, i1 undef, i32 8
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> poison, i1 undef, i32 15
@@ -1199,67 +1199,67 @@
 ;
 ; SLM-LABEL: 'insert_i1'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> poison, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> poison, i1 undef, i32 1
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> poison, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> poison, i1 undef, i32 2
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> poison, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> poison, i1 undef, i32 4
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> poison, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> poison, i1 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> poison, i1 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> poison, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> poison, i1 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> poison, i1 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> poison, i1 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_24 = insertelement <32 x i1> poison, i1 undef, i32 24
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_31 = insertelement <32 x i1> poison, i1 undef, i32 31
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> poison, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> poison, i1 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> poison, i1 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> poison, i1 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_24 = insertelement <64 x i1> poison, i1 undef, i32 24
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_31 = insertelement <64 x i1> poison, i1 undef, i32 31
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_63 = insertelement <64 x i1> poison, i1 undef, i32 63
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'insert_i1'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> poison, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> poison, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> poison, i1 undef, i32 1
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> poison, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> poison, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> poison, i1 undef, i32 2
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> poison, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> poison, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> poison, i1 undef, i32 4
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> poison, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> poison, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> poison, i1 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> poison, i1 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> poison, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> poison, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> poison, i1 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> poison, i1 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> poison, i1 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_24 = insertelement <32 x i1> poison, i1 undef, i32 24
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_31 = insertelement <32 x i1> poison, i1 undef, i32 31
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> poison, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> poison, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> poison, i1 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> poison, i1 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> poison, i1 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_24 = insertelement <64 x i1> poison, i1 undef, i32 24
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_31 = insertelement <64 x i1> poison, i1 undef, i32 31
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> poison, i1 undef, i32 32
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> poison, i1 undef, i32 48
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_63 = insertelement <64 x i1> poison, i1 undef, i32 63
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
diff --git a/llvm/test/Analysis/CostModel/X86/vector-insert.ll b/llvm/test/Analysis/CostModel/X86/vector-insert.ll
--- a/llvm/test/Analysis/CostModel/X86/vector-insert.ll
+++ b/llvm/test/Analysis/CostModel/X86/vector-insert.ll
@@ -294,71 +294,71 @@
 ;
 ; SSE4-LABEL: 'insert_i64'
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i64_a = insertelement <2 x i64> undef, i64 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = insertelement <2 x i64> undef, i64 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i64_0 = insertelement <2 x i64> undef, i64 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = insertelement <2 x i64> undef, i64 undef, i32 1
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i64_a = insertelement <4 x i64> undef, i64 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = insertelement <4 x i64> undef, i64 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64_0 = insertelement <4 x i64> undef, i64 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_3 = insertelement <4 x i64> undef, i64 undef, i32 3
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v8i64_a = insertelement <8 x i64> undef, i64 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = insertelement <8 x i64> undef, i64 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_0 = insertelement <8 x i64> undef, i64 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_3 = insertelement <8 x i64> undef, i64 undef, i32 3
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_4 = insertelement <8 x i64> undef, i64 undef, i32 4
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_4 = insertelement <8 x i64> undef, i64 undef, i32 4
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_7 = insertelement <8 x i64> undef, i64 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX-LABEL: 'insert_i64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i64_a = insertelement <2 x i64> undef, i64 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = insertelement <2 x i64> undef, i64 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i64_0 = insertelement <2 x i64> undef, i64 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = insertelement <2 x i64> undef, i64 undef, i32 1
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i64_a = insertelement <4 x i64> undef, i64 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = insertelement <4 x i64> undef, i64 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64_0 = insertelement <4 x i64> undef, i64 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i64_3 = insertelement <4 x i64> undef, i64 undef, i32 3
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i64_a = insertelement <8 x i64> undef, i64 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = insertelement <8 x i64> undef, i64 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_0 = insertelement <8 x i64> undef, i64 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_3 = insertelement <8 x i64> undef, i64 undef, i32 3
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_4 = insertelement <8 x i64> undef, i64 undef, i32 4
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_4 = insertelement <8 x i64> undef, i64 undef, i32 4
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_7 = insertelement <8 x i64> undef, i64 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'insert_i64'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i64_a = insertelement <2 x i64> undef, i64 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = insertelement <2 x i64> undef, i64 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i64_0 = insertelement <2 x i64> undef, i64 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = insertelement <2 x i64> undef, i64 undef, i32 1
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i64_a = insertelement <4 x i64> undef, i64 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = insertelement <4 x i64> undef, i64 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64_0 = insertelement <4 x i64> undef, i64 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i64_3 = insertelement <4 x i64> undef, i64 undef, i32 3
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_a = insertelement <8 x i64> undef, i64 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = insertelement <8 x i64> undef, i64 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_0 = insertelement <8 x i64> undef, i64 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_3 = insertelement <8 x i64> undef, i64 undef, i32 3
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_4 = insertelement <8 x i64> undef, i64 undef, i32 4
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i64_4 = insertelement <8 x i64> undef, i64 undef, i32 4
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i64_7 = insertelement <8 x i64> undef, i64 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SLM-LABEL: 'insert_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i64_a = insertelement <2 x i64> undef, i64 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = insertelement <2 x i64> undef, i64 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i64_0 = insertelement <2 x i64> undef, i64 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = insertelement <2 x i64> undef, i64 undef, i32 1
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i64_a = insertelement <4 x i64> undef, i64 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = insertelement <4 x i64> undef, i64 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64_0 = insertelement <4 x i64> undef, i64 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_3 = insertelement <4 x i64> undef, i64 undef, i32 3
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v8i64_a = insertelement <8 x i64> undef, i64 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = insertelement <8 x i64> undef, i64 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_0 = insertelement <8 x i64> undef, i64 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_3 = insertelement <8 x i64> undef, i64 undef, i32 3
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_4 = insertelement <8 x i64> undef, i64 undef, i32 4
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_4 = insertelement <8 x i64> undef, i64 undef, i32 4
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_7 = insertelement <8 x i64> undef, i64 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'insert_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i64_a = insertelement <2 x i64> undef, i64 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = insertelement <2 x i64> undef, i64 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i64_0 = insertelement <2 x i64> undef, i64 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = insertelement <2 x i64> undef, i64 undef, i32 1
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i64_a = insertelement <4 x i64> undef, i64 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = insertelement <4 x i64> undef, i64 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64_0 = insertelement <4 x i64> undef, i64 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i64_3 = insertelement <4 x i64> undef, i64 undef, i32 3
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v8i64_a = insertelement <8 x i64> undef, i64 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = insertelement <8 x i64> undef, i64 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_0 = insertelement <8 x i64> undef, i64 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_3 = insertelement <8 x i64> undef, i64 undef, i32 3
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_4 = insertelement <8 x i64> undef, i64 undef, i32 4
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i64_4 = insertelement <8 x i64> undef, i64 undef, i32 4
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i64_7 = insertelement <8 x i64> undef, i64 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -382,153 +382,153 @@
 define i32 @insert_i32(i32 %arg) {
 ; SSE2-LABEL: 'insert_i32'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> undef, i32 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_1 = insertelement <2 x i32> undef, i32 undef, i32 1
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> undef, i32 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_3 = insertelement <4 x i32> undef, i32 undef, i32 3
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> undef, i32 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_3 = insertelement <8 x i32> undef, i32 undef, i32 3
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_7 = insertelement <8 x i32> undef, i32 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> undef, i32 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_3 = insertelement <16 x i32> undef, i32 undef, i32 3
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_15 = insertelement <16 x i32> undef, i32 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE3-LABEL: 'insert_i32'
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> undef, i32 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_1 = insertelement <2 x i32> undef, i32 undef, i32 1
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> undef, i32 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_3 = insertelement <4 x i32> undef, i32 undef, i32 3
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> undef, i32 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_3 = insertelement <8 x i32> undef, i32 undef, i32 3
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_7 = insertelement <8 x i32> undef, i32 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> undef, i32 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_3 = insertelement <16 x i32> undef, i32 undef, i32 3
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_15 = insertelement <16 x i32> undef, i32 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'insert_i32'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> undef, i32 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_1 = insertelement <2 x i32> undef, i32 undef, i32 1
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> undef, i32 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_3 = insertelement <4 x i32> undef, i32 undef, i32 3
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> undef, i32 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_3 = insertelement <8 x i32> undef, i32 undef, i32 3
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_7 = insertelement <8 x i32> undef, i32 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> undef, i32 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_3 = insertelement <16 x i32> undef, i32 undef, i32 3
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_15 = insertelement <16 x i32> undef, i32 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'insert_i32'
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> undef, i32 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = insertelement <2 x i32> undef, i32 undef, i32 1
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> undef, i32 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_3 = insertelement <4 x i32> undef, i32 undef, i32 3
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> undef, i32 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_3 = insertelement <8 x i32> undef, i32 undef, i32 3
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_7 = insertelement <8 x i32> undef, i32 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> undef, i32 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_3 = insertelement <16 x i32> undef, i32 undef, i32 3
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_15 = insertelement <16 x i32> undef, i32 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX-LABEL: 'insert_i32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> undef, i32 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = insertelement <2 x i32> undef, i32 undef, i32 1
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> undef, i32 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_3 = insertelement <4 x i32> undef, i32 undef, i32 3
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_a = insertelement <8 x i32> undef, i32 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_3 = insertelement <8 x i32> undef, i32 undef, i32 3
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_7 = insertelement <8 x i32> undef, i32 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v16i32_a = insertelement <16 x i32> undef, i32 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_3 = insertelement <16 x i32> undef, i32 undef, i32 3
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_15 = insertelement <16 x i32> undef, i32 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'insert_i32'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> undef, i32 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = insertelement <2 x i32> undef, i32 undef, i32 1
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> undef, i32 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_3 = insertelement <4 x i32> undef, i32 undef, i32 3
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_a = insertelement <8 x i32> undef, i32 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_3 = insertelement <8 x i32> undef, i32 undef, i32 3
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i32_7 = insertelement <8 x i32> undef, i32 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_a = insertelement <16 x i32> undef, i32 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_3 = insertelement <16 x i32> undef, i32 undef, i32 3
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i32_15 = insertelement <16 x i32> undef, i32 undef, i32 15
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SLM-LABEL: 'insert_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> undef, i32 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = insertelement <2 x i32> undef, i32 undef, i32 1
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> undef, i32 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_3 = insertelement <4 x i32> undef, i32 undef, i32 3
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> undef, i32 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_3 = insertelement <8 x i32> undef, i32 undef, i32 3
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_7 = insertelement <8 x i32> undef, i32 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> undef, i32 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_3 = insertelement <16 x i32> undef, i32 undef, i32 3
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_15 = insertelement <16 x i32> undef, i32 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'insert_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i32_a = insertelement <2 x i32> undef, i32 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_0 = insertelement <2 x i32> undef, i32 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = insertelement <2 x i32> undef, i32 undef, i32 1
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_a = insertelement <4 x i32> undef, i32 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32_0 = insertelement <4 x i32> undef, i32 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_3 = insertelement <4 x i32> undef, i32 undef, i32 3
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8i32_a = insertelement <8 x i32> undef, i32 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = insertelement <8 x i32> undef, i32 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_3 = insertelement <8 x i32> undef, i32 undef, i32 3
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_4 = insertelement <8 x i32> undef, i32 undef, i32 4
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i32_7 = insertelement <8 x i32> undef, i32 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16i32_a = insertelement <16 x i32> undef, i32 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_0 = insertelement <16 x i32> undef, i32 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_3 = insertelement <16 x i32> undef, i32 undef, i32 3
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i32_8 = insertelement <16 x i32> undef, i32 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i32_15 = insertelement <16 x i32> undef, i32 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -558,76 +558,76 @@
 define i32 @insert_i16(i32 %arg) {
 ; SSE-LABEL: 'insert_i16'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i16_a = insertelement <2 x i16> undef, i16 undef, i32 %arg
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_0 = insertelement <2 x i16> undef, i16 undef, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i16_0 = insertelement <2 x i16> undef, i16 undef, i32 0
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_1 = insertelement <2 x i16> undef, i16 undef, i32 1
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i16_a = insertelement <4 x i16> undef, i16 undef, i32 %arg
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_0 = insertelement <4 x i16> undef, i16 undef, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i16_0 = insertelement <4 x i16> undef, i16 undef, i32 0
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_3 = insertelement <4 x i16> undef, i16 undef, i32 3
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i16_a = insertelement <8 x i16> undef, i16 undef, i32 %arg
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_0 = insertelement <8 x i16> undef, i16 undef, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i16_0 = insertelement <8 x i16> undef, i16 undef, i32 0
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_7 = insertelement <8 x i16> undef, i16 undef, i32 7
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v16i16_a = insertelement <16 x i16> undef, i16 undef, i32 %arg
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_0 = insertelement <16 x i16> undef, i16 undef, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_0 = insertelement <16 x i16> undef, i16 undef, i32 0
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_7 = insertelement <16 x i16> undef, i16 undef, i32 7
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_8 = insertelement <16 x i16> undef, i16 undef, i32 8
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_8 = insertelement <16 x i16> undef, i16 undef, i32 8
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_15 = insertelement <16 x i16> undef, i16 undef, i32 15
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v32i16_a = insertelement <32 x i16> undef, i16 undef, i32 %arg
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_0 = insertelement <32 x i16> undef, i16 undef, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_0 = insertelement <32 x i16> undef, i16 undef, i32 0
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_7 = insertelement <32 x i16> undef, i16 undef, i32 7
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_8 = insertelement <32 x i16> undef, i16 undef, i32 8
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_8 = insertelement <32 x i16> undef, i16 undef, i32 8
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_15 = insertelement <32 x i16> undef, i16 undef, i32 15
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_16 = insertelement <32 x i16> undef, i16 undef, i32 16
-; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_24 = insertelement <32 x i16> undef, i16 undef, i32 24
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_16 = insertelement <32 x i16> undef, i16 undef, i32 16
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_24 = insertelement <32 x i16> undef, i16 undef, i32 24
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_31 = insertelement <32 x i16> undef, i16 undef, i32 31
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX-LABEL: 'insert_i16'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i16_a = insertelement <2 x i16> undef, i16 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_0 = insertelement <2 x i16> undef, i16 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i16_0 = insertelement <2 x i16> undef, i16 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_1 = insertelement <2 x i16> undef, i16 undef, i32 1
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i16_a = insertelement <4 x i16> undef, i16 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_0 = insertelement <4 x i16> undef, i16 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i16_0 = insertelement <4 x i16> undef, i16 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_3 = insertelement <4 x i16> undef, i16 undef, i32 3
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i16_a = insertelement <8 x i16> undef, i16 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_0 = insertelement <8 x i16> undef, i16 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i16_0 = insertelement <8 x i16> undef, i16 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_7 = insertelement <8 x i16> undef, i16 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_a = insertelement <16 x i16> undef, i16 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_0 = insertelement <16 x i16> undef, i16 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_0 = insertelement <16 x i16> undef, i16 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_7 = insertelement <16 x i16> undef, i16 undef, i32 7
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_8 = insertelement <16 x i16> undef, i16 undef, i32 8
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i16_8 = insertelement <16 x i16> undef, i16 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_15 = insertelement <16 x i16> undef, i16 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i16_a = insertelement <32 x i16> undef, i16 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_0 = insertelement <32 x i16> undef, i16 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_0 = insertelement <32 x i16> undef, i16 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_7 = insertelement <32 x i16> undef, i16 undef, i32 7
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_8 = insertelement <32 x i16> undef, i16 undef, i32 8
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i16_8 = insertelement <32 x i16> undef, i16 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_15 = insertelement <32 x i16> undef, i16 undef, i32 15
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_16 = insertelement <32 x i16> undef, i16 undef, i32 16
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_24 = insertelement <32 x i16> undef, i16 undef, i32 24
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_16 = insertelement <32 x i16> undef, i16 undef, i32 16
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i16_24 = insertelement <32 x i16> undef, i16 undef, i32 24
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_31 = insertelement <32 x i16> undef, i16 undef, i32 31
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'insert_i16'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i16_a = insertelement <2 x i16> undef, i16 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_0 = insertelement <2 x i16> undef, i16 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i16_0 = insertelement <2 x i16> undef, i16 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16_1 = insertelement <2 x i16> undef, i16 undef, i32 1
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i16_a = insertelement <4 x i16> undef, i16 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_0 = insertelement <4 x i16> undef, i16 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i16_0 = insertelement <4 x i16> undef, i16 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_3 = insertelement <4 x i16> undef, i16 undef, i32 3
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i16_a = insertelement <8 x i16> undef, i16 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_0 = insertelement <8 x i16> undef, i16 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i16_0 = insertelement <8 x i16> undef, i16 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_7 = insertelement <8 x i16> undef, i16 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_a = insertelement <16 x i16> undef, i16 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_0 = insertelement <16 x i16> undef, i16 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_0 = insertelement <16 x i16> undef, i16 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i16_7 = insertelement <16 x i16> undef, i16 undef, i32 7
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_8 = insertelement <16 x i16> undef, i16 undef, i32 8
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i16_8 = insertelement <16 x i16> undef, i16 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i16_15 = insertelement <16 x i16> undef, i16 undef, i32 15
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_a = insertelement <32 x i16> undef, i16 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_0 = insertelement <32 x i16> undef, i16 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i16_0 = insertelement <32 x i16> undef, i16 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i16_7 = insertelement <32 x i16> undef, i16 undef, i32 7
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_8 = insertelement <32 x i16> undef, i16 undef, i32 8
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i16_8 = insertelement <32 x i16> undef, i16 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_15 = insertelement <32 x i16> undef, i16 undef, i32 15
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_16 = insertelement <32 x i16> undef, i16 undef, i32 16
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_24 = insertelement <32 x i16> undef, i16 undef, i32 24
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i16_16 = insertelement <32 x i16> undef, i16 undef, i32 16
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i16_24 = insertelement <32 x i16> undef, i16 undef, i32 24
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i16_31 = insertelement <32 x i16> undef, i16 undef, i32 31
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -664,265 +664,265 @@
 define i32 @insert_i8(i32 %arg) {
 ; SSE2-LABEL: 'insert_i8'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> undef, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_3 = insertelement <2 x i8> undef, i8 undef, i32 1
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> undef, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i8_3 = insertelement <4 x i8> undef, i8 undef, i32 3
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> undef, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i8_7 = insertelement <8 x i8> undef, i8 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> undef, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_8 = insertelement <16 x i8> undef, i8 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_15 = insertelement <16 x i8> undef, i8 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> undef, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_7 = insertelement <32 x i8> undef, i8 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_8 = insertelement <32 x i8> undef, i8 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_15 = insertelement <32 x i8> undef, i8 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_24 = insertelement <32 x i8> undef, i8 undef, i32 24
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_31 = insertelement <32 x i8> undef, i8 undef, i32 31
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> undef, i8 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_7 = insertelement <64 x i8> undef, i8 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_8 = insertelement <64 x i8> undef, i8 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_15 = insertelement <64 x i8> undef, i8 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_24 = insertelement <64 x i8> undef, i8 undef, i32 24
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_31 = insertelement <64 x i8> undef, i8 undef, i32 31
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_63 = insertelement <64 x i8> undef, i8 undef, i32 63
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE3-LABEL: 'insert_i8'
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> undef, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_3 = insertelement <2 x i8> undef, i8 undef, i32 1
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> undef, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v4i8_3 = insertelement <4 x i8> undef, i8 undef, i32 3
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> undef, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i8_7 = insertelement <8 x i8> undef, i8 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> undef, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_8 = insertelement <16 x i8> undef, i8 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i8_15 = insertelement <16 x i8> undef, i8 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> undef, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_7 = insertelement <32 x i8> undef, i8 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_8 = insertelement <32 x i8> undef, i8 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_15 = insertelement <32 x i8> undef, i8 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_24 = insertelement <32 x i8> undef, i8 undef, i32 24
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i8_31 = insertelement <32 x i8> undef, i8 undef, i32 31
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> undef, i8 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_7 = insertelement <64 x i8> undef, i8 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_8 = insertelement <64 x i8> undef, i8 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_15 = insertelement <64 x i8> undef, i8 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_24 = insertelement <64 x i8> undef, i8 undef, i32 24
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_31 = insertelement <64 x i8> undef, i8 undef, i32 31
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i8_63 = insertelement <64 x i8> undef, i8 undef, i32 63
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'insert_i8'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> undef, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v2i8_3 = insertelement <2 x i8> undef, i8 undef, i32 1
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> undef, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v4i8_3 = insertelement <4 x i8> undef, i8 undef, i32 3
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> undef, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i8_7 = insertelement <8 x i8> undef, i8 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> undef, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i8_8 = insertelement <16 x i8> undef, i8 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i8_15 = insertelement <16 x i8> undef, i8 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> undef, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_7 = insertelement <32 x i8> undef, i8 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_8 = insertelement <32 x i8> undef, i8 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_15 = insertelement <32 x i8> undef, i8 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_24 = insertelement <32 x i8> undef, i8 undef, i32 24
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i8_31 = insertelement <32 x i8> undef, i8 undef, i32 31
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> undef, i8 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_7 = insertelement <64 x i8> undef, i8 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_8 = insertelement <64 x i8> undef, i8 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_15 = insertelement <64 x i8> undef, i8 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_24 = insertelement <64 x i8> undef, i8 undef, i32 24
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_31 = insertelement <64 x i8> undef, i8 undef, i32 31
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_63 = insertelement <64 x i8> undef, i8 undef, i32 63
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'insert_i8'
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> undef, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_3 = insertelement <2 x i8> undef, i8 undef, i32 1
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> undef, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_3 = insertelement <4 x i8> undef, i8 undef, i32 3
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> undef, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_7 = insertelement <8 x i8> undef, i8 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> undef, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_8 = insertelement <16 x i8> undef, i8 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_15 = insertelement <16 x i8> undef, i8 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> undef, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_7 = insertelement <32 x i8> undef, i8 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_8 = insertelement <32 x i8> undef, i8 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_15 = insertelement <32 x i8> undef, i8 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_24 = insertelement <32 x i8> undef, i8 undef, i32 24
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_31 = insertelement <32 x i8> undef, i8 undef, i32 31
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> undef, i8 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_7 = insertelement <64 x i8> undef, i8 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_8 = insertelement <64 x i8> undef, i8 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_15 = insertelement <64 x i8> undef, i8 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_24 = insertelement <64 x i8> undef, i8 undef, i32 24
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_31 = insertelement <64 x i8> undef, i8 undef, i32 31
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_63 = insertelement <64 x i8> undef, i8 undef, i32 63
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX-LABEL: 'insert_i8'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> undef, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_3 = insertelement <2 x i8> undef, i8 undef, i32 1
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> undef, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_3 = insertelement <4 x i8> undef, i8 undef, i32 3
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> undef, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_7 = insertelement <8 x i8> undef, i8 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> undef, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_8 = insertelement <16 x i8> undef, i8 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_15 = insertelement <16 x i8> undef, i8 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_a = insertelement <32 x i8> undef, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_7 = insertelement <32 x i8> undef, i8 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_8 = insertelement <32 x i8> undef, i8 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_15 = insertelement <32 x i8> undef, i8 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_24 = insertelement <32 x i8> undef, i8 undef, i32 24
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_31 = insertelement <32 x i8> undef, i8 undef, i32 31
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v64i8_a = insertelement <64 x i8> undef, i8 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_7 = insertelement <64 x i8> undef, i8 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_8 = insertelement <64 x i8> undef, i8 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_15 = insertelement <64 x i8> undef, i8 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_24 = insertelement <64 x i8> undef, i8 undef, i32 24
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_31 = insertelement <64 x i8> undef, i8 undef, i32 31
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_63 = insertelement <64 x i8> undef, i8 undef, i32 63
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'insert_i8'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> undef, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_3 = insertelement <2 x i8> undef, i8 undef, i32 1
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> undef, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_3 = insertelement <4 x i8> undef, i8 undef, i32 3
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> undef, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_7 = insertelement <8 x i8> undef, i8 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> undef, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_8 = insertelement <16 x i8> undef, i8 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_15 = insertelement <16 x i8> undef, i8 undef, i32 15
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_a = insertelement <32 x i8> undef, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_7 = insertelement <32 x i8> undef, i8 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_8 = insertelement <32 x i8> undef, i8 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_15 = insertelement <32 x i8> undef, i8 undef, i32 15
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_24 = insertelement <32 x i8> undef, i8 undef, i32 24
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i8_31 = insertelement <32 x i8> undef, i8 undef, i32 31
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_a = insertelement <64 x i8> undef, i8 undef, i32 %arg
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_7 = insertelement <64 x i8> undef, i8 undef, i32 7
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_8 = insertelement <64 x i8> undef, i8 undef, i32 8
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_15 = insertelement <64 x i8> undef, i8 undef, i32 15
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_24 = insertelement <64 x i8> undef, i8 undef, i32 24
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_31 = insertelement <64 x i8> undef, i8 undef, i32 31
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i8_63 = insertelement <64 x i8> undef, i8 undef, i32 63
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SLM-LABEL: 'insert_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> undef, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_3 = insertelement <2 x i8> undef, i8 undef, i32 1
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> undef, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_3 = insertelement <4 x i8> undef, i8 undef, i32 3
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> undef, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_7 = insertelement <8 x i8> undef, i8 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> undef, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_8 = insertelement <16 x i8> undef, i8 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_15 = insertelement <16 x i8> undef, i8 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> undef, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_7 = insertelement <32 x i8> undef, i8 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_8 = insertelement <32 x i8> undef, i8 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_15 = insertelement <32 x i8> undef, i8 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_24 = insertelement <32 x i8> undef, i8 undef, i32 24
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_31 = insertelement <32 x i8> undef, i8 undef, i32 31
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> undef, i8 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_7 = insertelement <64 x i8> undef, i8 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_8 = insertelement <64 x i8> undef, i8 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_15 = insertelement <64 x i8> undef, i8 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_24 = insertelement <64 x i8> undef, i8 undef, i32 24
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_31 = insertelement <64 x i8> undef, i8 undef, i32 31
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_63 = insertelement <64 x i8> undef, i8 undef, i32 63
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'insert_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i8_a = insertelement <2 x i8> undef, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i8_0 = insertelement <2 x i8> undef, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8_3 = insertelement <2 x i8> undef, i8 undef, i32 1
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i8_a = insertelement <4 x i8> undef, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i8_0 = insertelement <4 x i8> undef, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8_3 = insertelement <4 x i8> undef, i8 undef, i32 3
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i8_a = insertelement <8 x i8> undef, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8_0 = insertelement <8 x i8> undef, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_7 = insertelement <8 x i8> undef, i8 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i8_a = insertelement <16 x i8> undef, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8_0 = insertelement <16 x i8> undef, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_8 = insertelement <16 x i8> undef, i8 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_15 = insertelement <16 x i8> undef, i8 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i8_a = insertelement <32 x i8> undef, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i8_0 = insertelement <32 x i8> undef, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_7 = insertelement <32 x i8> undef, i8 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_8 = insertelement <32 x i8> undef, i8 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_15 = insertelement <32 x i8> undef, i8 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_24 = insertelement <32 x i8> undef, i8 undef, i32 24
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i8_31 = insertelement <32 x i8> undef, i8 undef, i32 31
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i8_a = insertelement <64 x i8> undef, i8 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_0 = insertelement <64 x i8> undef, i8 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_7 = insertelement <64 x i8> undef, i8 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_8 = insertelement <64 x i8> undef, i8 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_15 = insertelement <64 x i8> undef, i8 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_24 = insertelement <64 x i8> undef, i8 undef, i32 24
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_31 = insertelement <64 x i8> undef, i8 undef, i32 31
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_32 = insertelement <64 x i8> undef, i8 undef, i32 32
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i8_48 = insertelement <64 x i8> undef, i8 undef, i32 48
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i8_63 = insertelement <64 x i8> undef, i8 undef, i32 63
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -971,31 +971,31 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_1 = insertelement <2 x i1> undef, i1 undef, i32 1
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> undef, i1 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_2 = insertelement <4 x i1> undef, i1 undef, i32 2
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> undef, i1 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> undef, i1 undef, i32 4
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> undef, i1 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_8 = insertelement <16 x i1> undef, i1 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_15 = insertelement <16 x i1> undef, i1 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> undef, i1 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_7 = insertelement <32 x i1> undef, i1 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_8 = insertelement <32 x i1> undef, i1 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_15 = insertelement <32 x i1> undef, i1 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_24 = insertelement <32 x i1> undef, i1 undef, i32 24
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_31 = insertelement <32 x i1> undef, i1 undef, i32 31
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> undef, i1 undef, i32 %arg
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_7 = insertelement <64 x i1> undef, i1 undef, i32 7
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_8 = insertelement <64 x i1> undef, i1 undef, i32 8
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_15 = insertelement <64 x i1> undef, i1 undef, i32 15
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_24 = insertelement <64 x i1> undef, i1 undef, i32 24
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_31 = insertelement <64 x i1> undef, i1 undef, i32 31
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_63 = insertelement <64 x i1> undef, i1 undef, i32 63
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -1004,31 +1004,31 @@
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_1 = insertelement <2 x i1> undef, i1 undef, i32 1
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> undef, i1 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_2 = insertelement <4 x i1> undef, i1 undef, i32 2
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> undef, i1 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> undef, i1 undef, i32 4
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> undef, i1 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_8 = insertelement <16 x i1> undef, i1 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i1_15 = insertelement <16 x i1> undef, i1 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> undef, i1 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_7 = insertelement <32 x i1> undef, i1 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_8 = insertelement <32 x i1> undef, i1 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_15 = insertelement <32 x i1> undef, i1 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_24 = insertelement <32 x i1> undef, i1 undef, i32 24
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v32i1_31 = insertelement <32 x i1> undef, i1 undef, i32 31
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> undef, i1 undef, i32 %arg
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_7 = insertelement <64 x i1> undef, i1 undef, i32 7
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_8 = insertelement <64 x i1> undef, i1 undef, i32 8
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_15 = insertelement <64 x i1> undef, i1 undef, i32 15
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_24 = insertelement <64 x i1> undef, i1 undef, i32 24
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_31 = insertelement <64 x i1> undef, i1 undef, i32 31
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
-; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
+; SSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v64i1_63 = insertelement <64 x i1> undef, i1 undef, i32 63
 ; SSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
@@ -1037,156 +1037,156 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_1 = insertelement <2 x i1> undef, i1 undef, i32 1
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> undef, i1 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_2 = insertelement <4 x i1> undef, i1 undef, i32 2
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> undef, i1 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> undef, i1 undef, i32 4
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> undef, i1 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i1_8 = insertelement <16 x i1> undef, i1 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i1_15 = insertelement <16 x i1> undef, i1 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> undef, i1 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_7 = insertelement <32 x i1> undef, i1 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_8 = insertelement <32 x i1> undef, i1 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_15 = insertelement <32 x i1> undef, i1 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_24 = insertelement <32 x i1> undef, i1 undef, i32 24
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32i1_31 = insertelement <32 x i1> undef, i1 undef, i32 31
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> undef, i1 undef, i32 %arg
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_7 = insertelement <64 x i1> undef, i1 undef, i32 7
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_8 = insertelement <64 x i1> undef, i1 undef, i32 8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_15 = insertelement <64 x i1> undef, i1 undef, i32 15
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_24 = insertelement <64 x i1> undef, i1 undef, i32 24
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_31 = insertelement <64 x i1> undef, i1 undef, i32 31
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_63 = insertelement <64 x i1> undef, i1 undef, i32 63
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'insert_i1'
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> undef, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> undef, i1 undef, i32 1
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> undef, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> undef, i1 undef, i32 2
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> undef, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> undef, i1 undef, i32 4
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> undef, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> undef, i1 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> undef, i1 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> undef, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> undef, i1 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> undef, i1 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> undef, i1 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_24 = insertelement <32 x i1> undef, i1 undef, i32 24
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_31 = insertelement <32 x i1> undef, i1 undef, i32 31
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> undef, i1 undef, i32 %arg
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> undef, i1 undef, i32 7
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> undef, i1 undef, i32 8
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> undef, i1 undef, i32 15
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_24 = insertelement <64 x i1> undef, i1 undef, i32 24
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_31 = insertelement <64 x i1> undef, i1 undef, i32 31
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
-; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
+; SSE4-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_63 = insertelement <64 x i1> undef, i1 undef, i32 63
 ; SSE4-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX-LABEL: 'insert_i1'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> undef, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> undef, i1 undef, i32 1
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> undef, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> undef, i1 undef, i32 2
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> undef, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> undef, i1 undef, i32 4
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> undef, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> undef, i1 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> undef, i1 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> undef, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> undef, i1 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> undef, i1 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> undef, i1 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_24 = insertelement <32 x i1> undef, i1 undef, i32 24
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_31 = insertelement <32 x i1> undef, i1 undef, i32 31
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> undef, i1 undef, i32 %arg
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> undef, i1 undef, i32 7
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> undef, i1 undef, i32 8
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> undef, i1 undef, i32 15
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_24 = insertelement <64 x i1> undef, i1 undef, i32 24
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_31 = insertelement <64 x i1> undef, i1 undef, i32 31
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
-; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_63 = insertelement <64 x i1> undef, i1 undef, i32 63
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'insert_i1'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> undef, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> undef, i1 undef, i32 1
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> undef, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> undef, i1 undef, i32 2
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> undef, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> undef, i1 undef, i32 4
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> undef, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> undef, i1 undef, i32 8
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> undef, i1 undef, i32 15
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32i1_a = insertelement <32 x i1> undef, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> undef, i1 undef, i32 7
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> undef, i1 undef, i32 8
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> undef, i1 undef, i32 15
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_24 = insertelement <32 x i1> undef, i1 undef, i32 24
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_31 = insertelement <32 x i1> undef, i1 undef, i32 31
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v64i1_a = insertelement <64 x i1> undef, i1 undef, i32 %arg
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> undef, i1 undef, i32 7
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> undef, i1 undef, i32 8
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> undef, i1 undef, i32 15
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_24 = insertelement <64 x i1> undef, i1 undef, i32 24
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_31 = insertelement <64 x i1> undef, i1 undef, i32 31
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_63 = insertelement <64 x i1> undef, i1 undef, i32 63
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'insert_i1'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> undef, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> undef, i1 undef, i32 1
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> undef, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> undef, i1 undef, i32 2
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> undef, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> undef, i1 undef, i32 4
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> undef, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> undef, i1 undef, i32 8
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> undef, i1 undef, i32 15
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> undef, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> undef, i1 undef, i32 7
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> undef, i1 undef, i32 8
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> undef, i1 undef, i32 15
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_24 = insertelement <32 x i1> undef, i1 undef, i32 24
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_31 = insertelement <32 x i1> undef, i1 undef, i32 31
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> undef, i1 undef, i32 %arg
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> undef, i1 undef, i32 7
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> undef, i1 undef, i32 8
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> undef, i1 undef, i32 15
@@ -1199,67 +1199,67 @@
 ;
 ; SLM-LABEL: 'insert_i1'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> undef, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> undef, i1 undef, i32 1
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> undef, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> undef, i1 undef, i32 2
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> undef, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> undef, i1 undef, i32 4
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> undef, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> undef, i1 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> undef, i1 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> undef, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> undef, i1 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> undef, i1 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> undef, i1 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_24 = insertelement <32 x i1> undef, i1 undef, i32 24
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_31 = insertelement <32 x i1> undef, i1 undef, i32 31
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> undef, i1 undef, i32 %arg
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> undef, i1 undef, i32 7
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> undef, i1 undef, i32 8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> undef, i1 undef, i32 15
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_24 = insertelement <64 x i1> undef, i1 undef, i32 24
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_31 = insertelement <64 x i1> undef, i1 undef, i32 31
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_63 = insertelement <64 x i1> undef, i1 undef, i32 63
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'insert_i1'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v2i1_a = insertelement <2 x i1> undef, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i1_0 = insertelement <2 x i1> undef, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i1_1 = insertelement <2 x i1> undef, i1 undef, i32 1
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i1_a = insertelement <4 x i1> undef, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i1_0 = insertelement <4 x i1> undef, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i1_2 = insertelement <4 x i1> undef, i1 undef, i32 2
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v8i1_a = insertelement <8 x i1> undef, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i1_0 = insertelement <8 x i1> undef, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i1_4 = insertelement <8 x i1> undef, i1 undef, i32 4
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16i1_a = insertelement <16 x i1> undef, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i1_0 = insertelement <16 x i1> undef, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_8 = insertelement <16 x i1> undef, i1 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i1_15 = insertelement <16 x i1> undef, i1 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v32i1_a = insertelement <32 x i1> undef, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i1_0 = insertelement <32 x i1> undef, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_7 = insertelement <32 x i1> undef, i1 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_8 = insertelement <32 x i1> undef, i1 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_15 = insertelement <32 x i1> undef, i1 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_24 = insertelement <32 x i1> undef, i1 undef, i32 24
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v32i1_31 = insertelement <32 x i1> undef, i1 undef, i32 31
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v64i1_a = insertelement <64 x i1> undef, i1 undef, i32 %arg
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_0 = insertelement <64 x i1> undef, i1 undef, i32 0
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_7 = insertelement <64 x i1> undef, i1 undef, i32 7
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_8 = insertelement <64 x i1> undef, i1 undef, i32 8
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_15 = insertelement <64 x i1> undef, i1 undef, i32 15
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_24 = insertelement <64 x i1> undef, i1 undef, i32 24
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_31 = insertelement <64 x i1> undef, i1 undef, i32 31
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
-; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_32 = insertelement <64 x i1> undef, i1 undef, i32 32
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i1_48 = insertelement <64 x i1> undef, i1 undef, i32 48
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v64i1_63 = insertelement <64 x i1> undef, i1 undef, i32 63
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-ashr-codesize.ll b/llvm/test/Analysis/CostModel/X86/vshift-ashr-codesize.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-ashr-codesize.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-ashr-codesize.ll
@@ -400,7 +400,7 @@
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %shift = ashr <2 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
@@ -437,7 +437,7 @@
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = ashr <4 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
@@ -486,7 +486,7 @@
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = ashr <8 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
@@ -534,35 +534,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX-LABEL: 'splatvar_shift_v4i32'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, %splat
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; XOP-LABEL: 'splatvar_shift_v4i32'
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, %splat
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX512-LABEL: 'splatvar_shift_v4i32'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, %splat
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
+; CHECK-LABEL: 'splatvar_shift_v4i32'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, %splat
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
 ;
   %insert = insertelement <4 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
@@ -572,7 +548,7 @@
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <8 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
@@ -621,7 +597,7 @@
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <16 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
@@ -811,7 +787,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %shift = ashr <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %shift
@@ -860,7 +836,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %shift = ashr <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <32 x i8> %shift
@@ -915,7 +891,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %shift = ashr <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost-inseltpoison.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost-inseltpoison.ll
@@ -443,17 +443,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %shift = ashr <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %shift = ashr <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v2i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %shift = ashr <2 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v2i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
@@ -492,17 +486,11 @@
 }
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = ashr <4 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = ashr <4 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v4i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = ashr <4 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
@@ -541,17 +529,11 @@
 }
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = ashr <8 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = ashr <8 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v8i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = ashr <8 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
@@ -590,17 +572,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v4i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
 ; AVX-LABEL: 'splatvar_shift_v4i32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
@@ -627,17 +603,11 @@
 }
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <8 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <8 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v8i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <8 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
@@ -676,17 +646,11 @@
 }
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <16 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v16i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <16 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v16i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <16 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v16i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
@@ -879,7 +843,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> poison, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> poison, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> poison, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = ashr <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
@@ -928,7 +892,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> poison, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> poison, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> poison, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %shift = ashr <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
@@ -995,7 +959,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> poison, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> poison, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> poison, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %shift = ashr <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll b/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
@@ -443,17 +443,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %shift = ashr <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %shift = ashr <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v2i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %shift = ashr <2 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v2i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
@@ -492,17 +486,11 @@
 }
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = ashr <4 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = ashr <4 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v4i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = ashr <4 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
@@ -541,17 +529,11 @@
 }
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = ashr <8 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = ashr <8 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v8i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = ashr <8 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
@@ -590,17 +572,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v4i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
 ; AVX-LABEL: 'splatvar_shift_v4i32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
@@ -627,17 +603,11 @@
 }
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <8 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <8 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v8i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <8 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
@@ -676,17 +646,11 @@
 }
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <16 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v16i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <16 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v16i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <16 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v16i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
@@ -879,7 +843,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = ashr <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
@@ -928,7 +892,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %shift = ashr <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
@@ -995,7 +959,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %shift = ashr <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-ashr-latency.ll b/llvm/test/Analysis/CostModel/X86/vshift-ashr-latency.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-ashr-latency.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-ashr-latency.ll
@@ -408,7 +408,7 @@
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %shift = ashr <2 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
@@ -445,7 +445,7 @@
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %shift = ashr <4 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
@@ -494,7 +494,7 @@
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %shift = ashr <8 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
@@ -543,7 +543,7 @@
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
@@ -592,7 +592,7 @@
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <8 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
@@ -641,7 +641,7 @@
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <16 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
@@ -843,7 +843,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %shift = ashr <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %shift
@@ -898,7 +898,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %shift = ashr <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <32 x i8> %shift
@@ -953,7 +953,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %shift = ashr <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-ashr-sizelatency.ll b/llvm/test/Analysis/CostModel/X86/vshift-ashr-sizelatency.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-ashr-sizelatency.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-ashr-sizelatency.ll
@@ -408,7 +408,7 @@
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %shift = ashr <2 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
@@ -445,7 +445,7 @@
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %shift = ashr <4 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
@@ -494,7 +494,7 @@
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %shift = ashr <8 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
@@ -542,35 +542,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX-LABEL: 'splatvar_shift_v4i32'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; XOP-LABEL: 'splatvar_shift_v4i32'
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX512-LABEL: 'splatvar_shift_v4i32'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
+; CHECK-LABEL: 'splatvar_shift_v4i32'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %splat
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
 ;
   %insert = insertelement <4 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
@@ -580,7 +556,7 @@
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <8 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
@@ -629,7 +605,7 @@
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <16 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
@@ -819,7 +795,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %shift = ashr <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %shift
@@ -868,7 +844,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %shift = ashr <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <32 x i8> %shift
@@ -923,7 +899,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %shift = ashr <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-lshr-codesize.ll b/llvm/test/Analysis/CostModel/X86/vshift-lshr-codesize.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-lshr-codesize.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-lshr-codesize.ll
@@ -407,35 +407,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; AVX-LABEL: 'splatvar_shift_v2i64'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, %splat
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; XOP-LABEL: 'splatvar_shift_v2i64'
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, %splat
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; AVX512-LABEL: 'splatvar_shift_v2i64'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, %splat
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
+; CHECK-LABEL: 'splatvar_shift_v2i64'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, %splat
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
 ;
   %insert = insertelement <2 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
@@ -445,7 +421,7 @@
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
@@ -494,7 +470,7 @@
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
@@ -542,35 +518,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX-LABEL: 'splatvar_shift_v4i32'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, %splat
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; XOP-LABEL: 'splatvar_shift_v4i32'
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, %splat
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX512-LABEL: 'splatvar_shift_v4i32'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, %splat
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
+; CHECK-LABEL: 'splatvar_shift_v4i32'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, %splat
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
 ;
   %insert = insertelement <4 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
@@ -580,7 +532,7 @@
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <8 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
@@ -629,7 +581,7 @@
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <16 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
@@ -819,7 +771,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %shift = lshr <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %shift
@@ -862,7 +814,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = lshr <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <32 x i8> %shift
@@ -911,7 +863,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = lshr <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost-inseltpoison.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost-inseltpoison.ll
@@ -447,17 +447,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v2i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
 ; AVX-LABEL: 'splatvar_shift_v2i64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
@@ -484,17 +478,11 @@
 }
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v4i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
@@ -533,17 +521,11 @@
 }
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v8i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
@@ -582,17 +564,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v4i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
 ; AVX-LABEL: 'splatvar_shift_v4i32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
@@ -619,17 +595,11 @@
 }
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v8i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
@@ -668,17 +638,11 @@
 }
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v16i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v16i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v16i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
@@ -871,7 +835,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> poison, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> poison, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> poison, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %shift = lshr <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
@@ -920,7 +884,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> poison, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> poison, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> poison, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %shift = lshr <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
@@ -969,7 +933,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> poison, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> poison, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> poison, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %shift = lshr <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll b/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
@@ -447,17 +447,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v2i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
 ; AVX-LABEL: 'splatvar_shift_v2i64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
@@ -484,17 +478,11 @@
 }
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v4i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
@@ -533,17 +521,11 @@
 }
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v8i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
@@ -582,17 +564,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v4i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
 ; AVX-LABEL: 'splatvar_shift_v4i32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
@@ -619,17 +595,11 @@
 }
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v8i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
@@ -668,17 +638,11 @@
 }
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v16i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v16i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v16i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
@@ -871,7 +835,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %shift = lshr <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
@@ -920,7 +884,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %shift = lshr <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
@@ -969,7 +933,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %shift = lshr <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-lshr-latency.ll b/llvm/test/Analysis/CostModel/X86/vshift-lshr-latency.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-lshr-latency.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-lshr-latency.ll
@@ -431,17 +431,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v2i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v2i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
@@ -480,17 +474,11 @@
 }
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v4i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
@@ -529,17 +517,11 @@
 }
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v8i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
@@ -578,17 +560,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v4i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
@@ -627,17 +603,11 @@
 }
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v8i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
@@ -676,17 +646,11 @@
 }
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v16i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v16i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v16i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
@@ -879,7 +843,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %shift = lshr <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %shift
@@ -934,7 +898,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %shift = lshr <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <32 x i8> %shift
@@ -983,7 +947,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %shift = lshr <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-lshr-sizelatency.ll b/llvm/test/Analysis/CostModel/X86/vshift-lshr-sizelatency.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-lshr-sizelatency.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-lshr-sizelatency.ll
@@ -411,35 +411,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; AVX-LABEL: 'splatvar_shift_v2i64'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; XOP-LABEL: 'splatvar_shift_v2i64'
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; AVX512-LABEL: 'splatvar_shift_v2i64'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
+; CHECK-LABEL: 'splatvar_shift_v2i64'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <2 x i64> %a, %splat
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
 ;
   %insert = insertelement <2 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
@@ -449,7 +425,7 @@
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
@@ -498,7 +474,7 @@
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
@@ -546,35 +522,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX-LABEL: 'splatvar_shift_v4i32'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; XOP-LABEL: 'splatvar_shift_v4i32'
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX512-LABEL: 'splatvar_shift_v4i32'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
+; CHECK-LABEL: 'splatvar_shift_v4i32'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <4 x i32> %a, %splat
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
 ;
   %insert = insertelement <4 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
@@ -584,7 +536,7 @@
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
@@ -633,7 +585,7 @@
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
@@ -823,7 +775,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %shift = lshr <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %shift
@@ -866,7 +818,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %shift = lshr <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <32 x i8> %shift
@@ -915,7 +867,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %shift = lshr <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-shl-codesize.ll b/llvm/test/Analysis/CostModel/X86/vshift-shl-codesize.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-shl-codesize.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-shl-codesize.ll
@@ -407,35 +407,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; AVX-LABEL: 'splatvar_shift_v2i64'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %splat
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; XOP-LABEL: 'splatvar_shift_v2i64'
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %splat
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; AVX512-LABEL: 'splatvar_shift_v2i64'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %splat
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
+; CHECK-LABEL: 'splatvar_shift_v2i64'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %splat
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
 ;
   %insert = insertelement <2 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
@@ -445,7 +421,7 @@
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
@@ -494,7 +470,7 @@
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i64> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
@@ -542,35 +518,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX-LABEL: 'splatvar_shift_v4i32'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %splat
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; XOP-LABEL: 'splatvar_shift_v4i32'
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %splat
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX512-LABEL: 'splatvar_shift_v4i32'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %splat
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
+; CHECK-LABEL: 'splatvar_shift_v4i32'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %splat
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
 ;
   %insert = insertelement <4 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
@@ -580,7 +532,7 @@
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <8 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
@@ -629,7 +581,7 @@
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <16 x i32> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
@@ -819,7 +771,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = shl <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %shift
@@ -868,7 +820,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = shl <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <32 x i8> %shift
@@ -917,7 +869,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = shl <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-shl-cost-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/vshift-shl-cost-inseltpoison.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-shl-cost-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-shl-cost-inseltpoison.ll
@@ -471,17 +471,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v2i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
 ; AVX-LABEL: 'splatvar_shift_v2i64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
@@ -500,12 +494,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v2i64'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> poison, i64 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
   %insert = insertelement <2 x i64> poison, i64 %b, i32 0
   %splat = shufflevector <2 x i64> %insert, <2 x i64> poison, <2 x i32> zeroinitializer
@@ -514,17 +502,11 @@
 }
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v4i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
@@ -555,12 +537,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v4i64'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> poison, i64 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
   %insert = insertelement <4 x i64> poison, i64 %b, i32 0
   %splat = shufflevector <4 x i64> %insert, <4 x i64> poison, <4 x i32> zeroinitializer
@@ -569,17 +545,11 @@
 }
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v8i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
@@ -610,12 +580,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v8i64'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> poison, i64 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
   %insert = insertelement <8 x i64> poison, i64 %b, i32 0
   %splat = shufflevector <8 x i64> %insert, <8 x i64> poison, <8 x i32> zeroinitializer
@@ -624,17 +588,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v4i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
 ; AVX-LABEL: 'splatvar_shift_v4i32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
@@ -653,12 +611,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v4i32'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> poison, i32 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
   %insert = insertelement <4 x i32> poison, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> poison, <4 x i32> zeroinitializer
@@ -667,17 +619,11 @@
 }
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v8i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
@@ -708,12 +654,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <8 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v8i32'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> poison, i32 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
   %insert = insertelement <8 x i32> poison, i32 %b, i32 0
   %splat = shufflevector <8 x i32> %insert, <8 x i32> poison, <8 x i32> zeroinitializer
@@ -722,17 +662,11 @@
 }
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v16i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v16i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v16i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
@@ -763,12 +697,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <16 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v16i32'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> poison, i32 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
   %insert = insertelement <16 x i32> poison, i32 %b, i32 0
   %splat = shufflevector <16 x i32> %insert, <16 x i32> poison, <16 x i32> zeroinitializer
@@ -949,7 +877,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> poison, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> poison, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> poison, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %shift = shl <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
@@ -1004,7 +932,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> poison, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> poison, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> poison, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %shift = shl <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
@@ -1059,7 +987,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> poison, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> poison, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> poison, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %shift = shl <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll b/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
@@ -471,17 +471,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v2i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
 ; AVX-LABEL: 'splatvar_shift_v2i64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
@@ -500,12 +494,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v2i64'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
   %insert = insertelement <2 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
@@ -514,17 +502,11 @@
 }
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v4i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
@@ -555,12 +537,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v4i64'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
   %insert = insertelement <4 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
@@ -569,17 +545,11 @@
 }
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v8i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
@@ -610,12 +580,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v8i64'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
   %insert = insertelement <8 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
@@ -624,17 +588,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v4i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
 ; AVX-LABEL: 'splatvar_shift_v4i32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
@@ -653,12 +611,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v4i32'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
   %insert = insertelement <4 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
@@ -667,17 +619,11 @@
 }
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v8i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
@@ -708,12 +654,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <8 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v8i32'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
   %insert = insertelement <8 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
@@ -722,17 +662,11 @@
 }
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v16i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v16i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v16i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
@@ -763,12 +697,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <16 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v16i32'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
   %insert = insertelement <16 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
@@ -949,7 +877,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %shift = shl <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
@@ -1004,7 +932,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %shift = shl <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
@@ -1059,7 +987,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %shift = shl <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-shl-latency.ll b/llvm/test/Analysis/CostModel/X86/vshift-shl-latency.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-shl-latency.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-shl-latency.ll
@@ -427,17 +427,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v2i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v2i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
@@ -476,17 +470,11 @@
 }
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v4i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
@@ -525,17 +513,11 @@
 }
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v8i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
@@ -574,17 +556,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v4i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
@@ -623,17 +599,11 @@
 }
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v8i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
@@ -672,17 +642,11 @@
 }
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v16i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v16i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v16i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
@@ -875,7 +839,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = shl <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %shift
@@ -924,7 +888,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = shl <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <32 x i8> %shift
@@ -973,7 +937,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %shift = shl <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-shl-sizelatency.ll b/llvm/test/Analysis/CostModel/X86/vshift-shl-sizelatency.ll
--- a/llvm/test/Analysis/CostModel/X86/vshift-shl-sizelatency.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-shl-sizelatency.ll
@@ -475,41 +475,11 @@
 ;
 
 define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v2i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; AVX-LABEL: 'splatvar_shift_v2i64'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; XOP-LABEL: 'splatvar_shift_v2i64'
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; AVX512-LABEL: 'splatvar_shift_v2i64'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v2i64'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
+; CHECK-LABEL: 'splatvar_shift_v2i64'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <2 x i64> %a, %splat
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <2 x i64> %shift
 ;
   %insert = insertelement <2 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
@@ -518,17 +488,11 @@
 }
 
 define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v4i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v4i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
@@ -559,12 +523,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v4i64'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i64> %shift
 ;
   %insert = insertelement <4 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
@@ -573,17 +531,11 @@
 }
 
 define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i64'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
+; SSE-LABEL: 'splatvar_shift_v8i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
@@ -614,12 +566,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <8 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v8i64'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i64> %shift
 ;
   %insert = insertelement <8 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
@@ -628,41 +574,11 @@
 }
 
 define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v4i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX-LABEL: 'splatvar_shift_v4i32'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; XOP-LABEL: 'splatvar_shift_v4i32'
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; AVX512-LABEL: 'splatvar_shift_v4i32'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v4i32'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
+; CHECK-LABEL: 'splatvar_shift_v4i32'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, %splat
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %shift
 ;
   %insert = insertelement <4 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
@@ -671,17 +587,11 @@
 }
 
 define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v8i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v8i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v8i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
@@ -712,12 +622,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %shift = shl <8 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v8i32'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i32> %shift
 ;
   %insert = insertelement <8 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
@@ -726,17 +630,11 @@
 }
 
 define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
-; SSE2-LABEL: 'splatvar_shift_v16i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
-;
-; SSE42-LABEL: 'splatvar_shift_v16i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
+; SSE-LABEL: 'splatvar_shift_v16i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
 ;
 ; AVX1-LABEL: 'splatvar_shift_v16i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
@@ -767,12 +665,6 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %shift = shl <16 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
-;
-; SLM-LABEL: 'splatvar_shift_v16i32'
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
-; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i32> %shift
 ;
   %insert = insertelement <16 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
@@ -941,7 +833,7 @@
 
 define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %shift = shl <16 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %shift
@@ -996,7 +888,7 @@
 
 define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %shift = shl <32 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <32 x i8> %shift
@@ -1051,7 +943,7 @@
 
 define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; SSE2-LABEL: 'splatvar_shift_v64i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %shift = shl <64 x i8> %a, %splat
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <64 x i8> %shift
diff --git a/llvm/test/Transforms/SLPVectorizer/X86/cse.ll b/llvm/test/Transforms/SLPVectorizer/X86/cse.ll
--- a/llvm/test/Transforms/SLPVectorizer/X86/cse.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/cse.ll
@@ -69,11 +69,11 @@
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x double>, <4 x double>* [[TMP0]], align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = fmul <4 x double> [[TMP1]], <double 7.900000e+00, double 7.700000e+00, double 7.600000e+00, double 7.400000e+00>
 ; CHECK-NEXT:    [[TMP3:%.*]] = insertelement <4 x double> poison, double [[CONV]], i32 0
-; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <4 x double> [[TMP3]], <4 x double> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP4:%.*]] = fmul <4 x double> [[SHUFFLE]], [[TMP2]]
-; CHECK-NEXT:    [[TMP5:%.*]] = fadd <4 x double> [[TMP4]], <double 6.000000e+00, double 2.000000e+00, double 3.000000e+00, double 4.000000e+00>
-; CHECK-NEXT:    [[TMP6:%.*]] = bitcast double* [[A]] to <4 x double>*
-; CHECK-NEXT:    store <4 x double> [[TMP5]], <4 x double>* [[TMP6]], align 8
+; CHECK-NEXT:    [[TMP4:%.*]] = shufflevector <4 x double> [[TMP3]], <4 x double> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP5:%.*]] = fmul <4 x double> [[TMP4]], [[TMP2]]
+; CHECK-NEXT:    [[TMP6:%.*]] = fadd <4 x double> [[TMP5]], <double 6.000000e+00, double 2.000000e+00, double 3.000000e+00, double 4.000000e+00>
+; CHECK-NEXT:    [[TMP7:%.*]] = bitcast double* [[A]] to <4 x double>*
+; CHECK-NEXT:    store <4 x double> [[TMP6]], <4 x double>* [[TMP7]], align 8
 ; CHECK-NEXT:    ret i32 undef
 ;
 entry:
@@ -196,11 +196,11 @@
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x double>, <4 x double>* [[TMP0]], align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = fmul <4 x double> [[TMP1]], <double 7.900000e+00, double 7.900000e+00, double 7.900000e+00, double 7.900000e+00>
 ; CHECK-NEXT:    [[TMP3:%.*]] = insertelement <4 x double> poison, double [[CONV]], i32 0
-; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <4 x double> [[TMP3]], <4 x double> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP4:%.*]] = fmul <4 x double> [[SHUFFLE]], [[TMP2]]
-; CHECK-NEXT:    [[TMP5:%.*]] = fadd <4 x double> [[TMP4]], <double 6.000000e+00, double 6.000000e+00, double 6.000000e+00, double 6.000000e+00>
-; CHECK-NEXT:    [[TMP6:%.*]] = bitcast double* [[A]] to <4 x double>*
-; CHECK-NEXT:    store <4 x double> [[TMP5]], <4 x double>* [[TMP6]], align 8
+; CHECK-NEXT:    [[TMP4:%.*]] = shufflevector <4 x double> [[TMP3]], <4 x double> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP5:%.*]] = fmul <4 x double> [[TMP4]], [[TMP2]]
+; CHECK-NEXT:    [[TMP6:%.*]] = fadd <4 x double> [[TMP5]], <double 6.000000e+00, double 6.000000e+00, double 6.000000e+00, double 6.000000e+00>
+; CHECK-NEXT:    [[TMP7:%.*]] = bitcast double* [[A]] to <4 x double>*
+; CHECK-NEXT:    store <4 x double> [[TMP6]], <4 x double>* [[TMP7]], align 8
 ; CHECK-NEXT:    ret i32 undef
 ;
 entry:
@@ -246,22 +246,22 @@
 ; CHECK-NEXT:    [[TMP0:%.*]] = bitcast double* [[A:%.*]] to <2 x double>*
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <2 x double>, <2 x double>* [[TMP0]], align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <2 x double> poison, double [[CONV]], i32 0
-; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <2 x double> [[TMP2]], <2 x double> poison, <2 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP3:%.*]] = fmul <2 x double> [[SHUFFLE]], [[TMP1]]
-; CHECK-NEXT:    [[TMP4:%.*]] = bitcast double* [[A]] to <2 x double>*
-; CHECK-NEXT:    store <2 x double> [[TMP3]], <2 x double>* [[TMP4]], align 8
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <2 x double> [[TMP2]], <2 x double> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP4:%.*]] = fmul <2 x double> [[TMP3]], [[TMP1]]
+; CHECK-NEXT:    [[TMP5:%.*]] = bitcast double* [[A]] to <2 x double>*
+; CHECK-NEXT:    store <2 x double> [[TMP4]], <2 x double>* [[TMP5]], align 8
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i32 [[N]], 4
 ; CHECK-NEXT:    br i1 [[CMP]], label [[RETURN:%.*]], label [[IF_END:%.*]]
 ; CHECK:       if.end:
 ; CHECK-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds double, double* [[A]], i64 2
 ; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[N]], 4
 ; CHECK-NEXT:    [[CONV12:%.*]] = sitofp i32 [[ADD]] to double
-; CHECK-NEXT:    [[TMP5:%.*]] = bitcast double* [[ARRAYIDX7]] to <2 x double>*
-; CHECK-NEXT:    [[TMP6:%.*]] = load <2 x double>, <2 x double>* [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP7:%.*]] = insertelement <2 x double> [[TMP2]], double [[CONV12]], i32 1
-; CHECK-NEXT:    [[TMP8:%.*]] = fmul <2 x double> [[TMP7]], [[TMP6]]
-; CHECK-NEXT:    [[TMP9:%.*]] = bitcast double* [[ARRAYIDX7]] to <2 x double>*
-; CHECK-NEXT:    store <2 x double> [[TMP8]], <2 x double>* [[TMP9]], align 8
+; CHECK-NEXT:    [[TMP6:%.*]] = bitcast double* [[ARRAYIDX7]] to <2 x double>*
+; CHECK-NEXT:    [[TMP7:%.*]] = load <2 x double>, <2 x double>* [[TMP6]], align 8
+; CHECK-NEXT:    [[TMP8:%.*]] = insertelement <2 x double> [[TMP2]], double [[CONV12]], i32 1
+; CHECK-NEXT:    [[TMP9:%.*]] = fmul <2 x double> [[TMP8]], [[TMP7]]
+; CHECK-NEXT:    [[TMP10:%.*]] = bitcast double* [[ARRAYIDX7]] to <2 x double>*
+; CHECK-NEXT:    store <2 x double> [[TMP9]], <2 x double>* [[TMP10]], align 8
 ; CHECK-NEXT:    br label [[RETURN]]
 ; CHECK:       return:
 ; CHECK-NEXT:    ret i32 0
@@ -352,18 +352,20 @@
 ; CHECK-LABEL: @cse_for_hoisted_instructions_in_preheader(
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = insertelement <2 x i32> poison, i32 [[A:%.*]], i32 0
-; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <2 x i32> [[TMP0]], <2 x i32> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <2 x i32> [[TMP0]], <2 x i32> poison, <2 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
-; CHECK-NEXT:    [[TMP1:%.*]] = or <2 x i32> <i32 22, i32 22>, [[SHUFFLE]]
+; CHECK-NEXT:    [[TMP2:%.*]] = or <2 x i32> <i32 22, i32 22>, [[TMP1]]
 ; CHECK-NEXT:    [[GEP_0:%.*]] = getelementptr inbounds i32, i32* [[DST:%.*]], i64 0
-; CHECK-NEXT:    [[TMP2:%.*]] = or <2 x i32> [[TMP1]], <i32 3, i32 3>
-; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[GEP_0]] to <2 x i32>*
-; CHECK-NEXT:    store <2 x i32> [[TMP2]], <2 x i32>* [[TMP3]], align 4
-; CHECK-NEXT:    [[TMP4:%.*]] = or <2 x i32> [[SHUFFLE]], <i32 3, i32 3>
+; CHECK-NEXT:    [[TMP3:%.*]] = or <2 x i32> [[TMP2]], <i32 3, i32 3>
+; CHECK-NEXT:    [[TMP4:%.*]] = bitcast i32* [[GEP_0]] to <2 x i32>*
+; CHECK-NEXT:    store <2 x i32> [[TMP3]], <2 x i32>* [[TMP4]], align 4
+; CHECK-NEXT:    [[OR_2:%.*]] = or i32 [[A]], 3
 ; CHECK-NEXT:    [[GEP_2:%.*]] = getelementptr inbounds i32, i32* [[DST]], i64 10
-; CHECK-NEXT:    [[TMP5:%.*]] = bitcast i32* [[GEP_2]] to <2 x i32>*
-; CHECK-NEXT:    store <2 x i32> [[TMP4]], <2 x i32>* [[TMP5]], align 4
+; CHECK-NEXT:    store i32 [[OR_2]], i32* [[GEP_2]], align 4
+; CHECK-NEXT:    [[OR_3:%.*]] = or i32 [[A]], 3
+; CHECK-NEXT:    [[GEP_3:%.*]] = getelementptr inbounds i32, i32* [[DST]], i64 11
+; CHECK-NEXT:    store i32 [[OR_3]], i32* [[GEP_3]], align 4
 ; CHECK-NEXT:    br i1 [[C:%.*]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
diff --git a/llvm/test/Transforms/SLPVectorizer/X86/malformed_phis.ll b/llvm/test/Transforms/SLPVectorizer/X86/malformed_phis.ll
--- a/llvm/test/Transforms/SLPVectorizer/X86/malformed_phis.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/malformed_phis.ll
@@ -11,11 +11,23 @@
 ; CHECK-NEXT:    br label [[BB1:%.*]]
 ; CHECK:       bb1:
 ; CHECK-NEXT:    [[TMP:%.*]] = phi i32 [ undef, [[BB1]] ], [ undef, [[BB:%.*]] ]
-; CHECK-NEXT:    [[TMP2:%.*]] = phi i32 [ [[OP_RDX:%.*]], [[BB1]] ], [ undef, [[BB]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = insertelement <16 x i32> poison, i32 [[TMP]], i32 0
-; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <16 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @llvm.vector.reduce.mul.v16i32(<16 x i32> [[SHUFFLE]])
-; CHECK-NEXT:    [[OP_RDX]] = mul i32 [[TMP1]], undef
+; CHECK-NEXT:    [[TMP2:%.*]] = phi i32 [ [[TMP18:%.*]], [[BB1]] ], [ undef, [[BB]] ]
+; CHECK-NEXT:    [[TMP3:%.*]] = mul i32 undef, [[TMP]]
+; CHECK-NEXT:    [[TMP4:%.*]] = mul i32 [[TMP3]], [[TMP]]
+; CHECK-NEXT:    [[TMP5:%.*]] = mul i32 [[TMP4]], [[TMP]]
+; CHECK-NEXT:    [[TMP6:%.*]] = mul i32 [[TMP5]], [[TMP]]
+; CHECK-NEXT:    [[TMP7:%.*]] = mul i32 [[TMP6]], [[TMP]]
+; CHECK-NEXT:    [[TMP8:%.*]] = mul i32 [[TMP7]], [[TMP]]
+; CHECK-NEXT:    [[TMP9:%.*]] = mul i32 [[TMP8]], [[TMP]]
+; CHECK-NEXT:    [[TMP10:%.*]] = mul i32 [[TMP9]], [[TMP]]
+; CHECK-NEXT:    [[TMP11:%.*]] = mul i32 [[TMP10]], [[TMP]]
+; CHECK-NEXT:    [[TMP12:%.*]] = mul i32 [[TMP11]], [[TMP]]
+; CHECK-NEXT:    [[TMP13:%.*]] = mul i32 [[TMP12]], [[TMP]]
+; CHECK-NEXT:    [[TMP14:%.*]] = mul i32 [[TMP13]], [[TMP]]
+; CHECK-NEXT:    [[TMP15:%.*]] = mul i32 [[TMP14]], [[TMP]]
+; CHECK-NEXT:    [[TMP16:%.*]] = mul i32 [[TMP15]], [[TMP]]
+; CHECK-NEXT:    [[TMP17:%.*]] = mul i32 [[TMP16]], [[TMP]]
+; CHECK-NEXT:    [[TMP18]] = mul i32 [[TMP17]], [[TMP]]
 ; CHECK-NEXT:    br label [[BB1]]
 ;
 bb:
@@ -51,10 +63,10 @@
 ; CHECK-NEXT:    [[TMP:%.*]] = phi i32 [ undef, [[BB:%.*]] ], [ undef, [[BB2]] ]
 ; CHECK-NEXT:    [[TMP3:%.*]] = phi i32 [ 0, [[BB]] ], [ undef, [[BB2]] ]
 ; CHECK-NEXT:    [[TMP0:%.*]] = insertelement <8 x i32> poison, i32 [[TMP]], i32 0
-; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <8 x i32> [[TMP0]], <8 x i32> poison, <8 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> undef)
-; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[SHUFFLE]])
-; CHECK-NEXT:    [[OP_RDX:%.*]] = add i32 [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <8 x i32> [[TMP0]], <8 x i32> poison, <8 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> undef)
+; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP1]])
+; CHECK-NEXT:    [[OP_RDX:%.*]] = add i32 [[TMP2]], [[TMP3]]
 ; CHECK-NEXT:    [[OP_RDX1:%.*]] = add i32 [[OP_RDX]], undef
 ; CHECK-NEXT:    call void @use(i32 [[OP_RDX1]])
 ; CHECK-NEXT:    br label [[BB2]]
@@ -96,66 +108,56 @@
 ; CHECK-NEXT:    br label [[BB3]]
 ; CHECK:       bb3:
 ; CHECK-NEXT:    [[TMP0:%.*]] = phi <2 x i32> [ undef, [[BB1]] ], [ poison, [[BB2:%.*]] ]
-; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <2 x i32> [[TMP0]], <2 x i32> poison, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1>
-; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <8 x i32> [[SHUFFLE]], i32 0
-; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <16 x i32> poison, i32 [[TMP1]], i32 0
-; CHECK-NEXT:    [[TMP3:%.*]] = insertelement <16 x i32> [[TMP2]], i32 [[TMP1]], i32 1
-; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <16 x i32> [[TMP3]], i32 [[TMP1]], i32 2
-; CHECK-NEXT:    [[TMP5:%.*]] = insertelement <16 x i32> [[TMP4]], i32 [[TMP1]], i32 3
-; CHECK-NEXT:    [[TMP6:%.*]] = insertelement <16 x i32> [[TMP5]], i32 [[TMP1]], i32 4
-; CHECK-NEXT:    [[TMP7:%.*]] = insertelement <16 x i32> [[TMP6]], i32 [[TMP1]], i32 5
-; CHECK-NEXT:    [[TMP8:%.*]] = insertelement <16 x i32> [[TMP7]], i32 [[TMP1]], i32 6
-; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <16 x i32> [[TMP8]], i32 [[TMP1]], i32 7
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <16 x i32> [[TMP9]], i32 [[TMP1]], i32 8
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <16 x i32> [[TMP10]], i32 [[TMP1]], i32 9
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <16 x i32> [[TMP11]], i32 [[TMP1]], i32 10
-; CHECK-NEXT:    [[TMP13:%.*]] = insertelement <16 x i32> [[TMP12]], i32 [[TMP1]], i32 11
-; CHECK-NEXT:    [[TMP14:%.*]] = insertelement <16 x i32> [[TMP13]], i32 [[TMP1]], i32 12
-; CHECK-NEXT:    [[TMP15:%.*]] = insertelement <16 x i32> [[TMP14]], i32 [[TMP1]], i32 13
-; CHECK-NEXT:    [[TMP16:%.*]] = insertelement <16 x i32> [[TMP15]], i32 [[TMP1]], i32 14
-; CHECK-NEXT:    [[TMP17:%.*]] = insertelement <16 x i32> [[TMP16]], i32 [[TMP1]], i32 15
-; CHECK-NEXT:    [[TMP18:%.*]] = insertelement <32 x i32> poison, i32 [[TMP1]], i32 0
-; CHECK-NEXT:    [[TMP19:%.*]] = insertelement <32 x i32> [[TMP18]], i32 [[TMP1]], i32 1
-; CHECK-NEXT:    [[TMP20:%.*]] = insertelement <32 x i32> [[TMP19]], i32 [[TMP1]], i32 2
-; CHECK-NEXT:    [[TMP21:%.*]] = insertelement <32 x i32> [[TMP20]], i32 [[TMP1]], i32 3
-; CHECK-NEXT:    [[TMP22:%.*]] = insertelement <32 x i32> [[TMP21]], i32 [[TMP1]], i32 4
-; CHECK-NEXT:    [[TMP23:%.*]] = insertelement <32 x i32> [[TMP22]], i32 [[TMP1]], i32 5
-; CHECK-NEXT:    [[TMP24:%.*]] = insertelement <32 x i32> [[TMP23]], i32 [[TMP1]], i32 6
-; CHECK-NEXT:    [[TMP25:%.*]] = insertelement <32 x i32> [[TMP24]], i32 [[TMP1]], i32 7
-; CHECK-NEXT:    [[TMP26:%.*]] = insertelement <32 x i32> [[TMP25]], i32 [[TMP1]], i32 8
-; CHECK-NEXT:    [[TMP27:%.*]] = insertelement <32 x i32> [[TMP26]], i32 [[TMP1]], i32 9
-; CHECK-NEXT:    [[TMP28:%.*]] = insertelement <32 x i32> [[TMP27]], i32 [[TMP1]], i32 10
-; CHECK-NEXT:    [[TMP29:%.*]] = insertelement <32 x i32> [[TMP28]], i32 [[TMP1]], i32 11
-; CHECK-NEXT:    [[TMP30:%.*]] = insertelement <32 x i32> [[TMP29]], i32 [[TMP1]], i32 12
-; CHECK-NEXT:    [[TMP31:%.*]] = insertelement <32 x i32> [[TMP30]], i32 [[TMP1]], i32 13
-; CHECK-NEXT:    [[TMP32:%.*]] = insertelement <32 x i32> [[TMP31]], i32 [[TMP1]], i32 14
-; CHECK-NEXT:    [[TMP33:%.*]] = insertelement <32 x i32> [[TMP32]], i32 [[TMP1]], i32 15
-; CHECK-NEXT:    [[TMP34:%.*]] = insertelement <32 x i32> [[TMP33]], i32 [[TMP1]], i32 16
-; CHECK-NEXT:    [[TMP35:%.*]] = insertelement <32 x i32> [[TMP34]], i32 [[TMP1]], i32 17
-; CHECK-NEXT:    [[TMP36:%.*]] = insertelement <32 x i32> [[TMP35]], i32 [[TMP1]], i32 18
-; CHECK-NEXT:    [[TMP37:%.*]] = insertelement <32 x i32> [[TMP36]], i32 [[TMP1]], i32 19
-; CHECK-NEXT:    [[TMP38:%.*]] = insertelement <32 x i32> [[TMP37]], i32 [[TMP1]], i32 20
-; CHECK-NEXT:    [[TMP39:%.*]] = insertelement <32 x i32> [[TMP38]], i32 [[TMP1]], i32 21
-; CHECK-NEXT:    [[TMP40:%.*]] = insertelement <32 x i32> [[TMP39]], i32 [[TMP1]], i32 22
-; CHECK-NEXT:    [[TMP41:%.*]] = insertelement <32 x i32> [[TMP40]], i32 [[TMP1]], i32 23
-; CHECK-NEXT:    [[TMP42:%.*]] = insertelement <32 x i32> [[TMP41]], i32 [[TMP1]], i32 24
-; CHECK-NEXT:    [[TMP43:%.*]] = insertelement <32 x i32> [[TMP42]], i32 [[TMP1]], i32 25
-; CHECK-NEXT:    [[TMP44:%.*]] = insertelement <32 x i32> [[TMP43]], i32 [[TMP1]], i32 26
-; CHECK-NEXT:    [[TMP45:%.*]] = insertelement <32 x i32> [[TMP44]], i32 [[TMP1]], i32 27
-; CHECK-NEXT:    [[TMP46:%.*]] = insertelement <32 x i32> [[TMP45]], i32 [[TMP1]], i32 28
-; CHECK-NEXT:    [[TMP47:%.*]] = insertelement <32 x i32> [[TMP46]], i32 [[TMP1]], i32 29
-; CHECK-NEXT:    [[TMP48:%.*]] = insertelement <32 x i32> [[TMP47]], i32 [[TMP1]], i32 30
-; CHECK-NEXT:    [[TMP49:%.*]] = insertelement <32 x i32> [[TMP48]], i32 [[TMP1]], i32 31
-; CHECK-NEXT:    [[TMP50:%.*]] = call i32 @llvm.vector.reduce.mul.v32i32(<32 x i32> [[TMP49]])
-; CHECK-NEXT:    [[TMP51:%.*]] = call i32 @llvm.vector.reduce.mul.v16i32(<16 x i32> [[TMP17]])
-; CHECK-NEXT:    [[OP_RDX:%.*]] = mul i32 [[TMP50]], [[TMP51]]
-; CHECK-NEXT:    [[TMP52:%.*]] = call i32 @llvm.vector.reduce.mul.v8i32(<8 x i32> [[SHUFFLE]])
-; CHECK-NEXT:    [[OP_RDX1:%.*]] = mul i32 [[OP_RDX]], [[TMP52]]
-; CHECK-NEXT:    [[OP_RDX2:%.*]] = mul i32 [[OP_RDX1]], [[TMP1]]
-; CHECK-NEXT:    [[OP_RDX3:%.*]] = mul i32 [[TMP1]], [[TMP1]]
-; CHECK-NEXT:    [[OP_RDX4:%.*]] = mul i32 [[OP_RDX2]], [[OP_RDX3]]
-; CHECK-NEXT:    [[OP_RDX5:%.*]] = mul i32 [[OP_RDX4]], [[TMP1]]
-; CHECK-NEXT:    [[VAL64:%.*]] = add i32 undef, [[OP_RDX5]]
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <2 x i32> [[TMP0]], <2 x i32> poison, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1>
+; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <16 x i32> [[TMP1]], i32 0
+; CHECK-NEXT:    [[TMP3:%.*]] = insertelement <32 x i32> poison, i32 [[TMP2]], i32 0
+; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <32 x i32> [[TMP3]], i32 [[TMP2]], i32 1
+; CHECK-NEXT:    [[TMP5:%.*]] = insertelement <32 x i32> [[TMP4]], i32 [[TMP2]], i32 2
+; CHECK-NEXT:    [[TMP6:%.*]] = insertelement <32 x i32> [[TMP5]], i32 [[TMP2]], i32 3
+; CHECK-NEXT:    [[TMP7:%.*]] = insertelement <32 x i32> [[TMP6]], i32 [[TMP2]], i32 4
+; CHECK-NEXT:    [[TMP8:%.*]] = insertelement <32 x i32> [[TMP7]], i32 [[TMP2]], i32 5
+; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <32 x i32> [[TMP8]], i32 [[TMP2]], i32 6
+; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <32 x i32> [[TMP9]], i32 [[TMP2]], i32 7
+; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <32 x i32> [[TMP10]], i32 [[TMP2]], i32 8
+; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <32 x i32> [[TMP11]], i32 [[TMP2]], i32 9
+; CHECK-NEXT:    [[TMP13:%.*]] = insertelement <32 x i32> [[TMP12]], i32 [[TMP2]], i32 10
+; CHECK-NEXT:    [[TMP14:%.*]] = insertelement <32 x i32> [[TMP13]], i32 [[TMP2]], i32 11
+; CHECK-NEXT:    [[TMP15:%.*]] = insertelement <32 x i32> [[TMP14]], i32 [[TMP2]], i32 12
+; CHECK-NEXT:    [[TMP16:%.*]] = insertelement <32 x i32> [[TMP15]], i32 [[TMP2]], i32 13
+; CHECK-NEXT:    [[TMP17:%.*]] = insertelement <32 x i32> [[TMP16]], i32 [[TMP2]], i32 14
+; CHECK-NEXT:    [[TMP18:%.*]] = insertelement <32 x i32> [[TMP17]], i32 [[TMP2]], i32 15
+; CHECK-NEXT:    [[TMP19:%.*]] = insertelement <32 x i32> [[TMP18]], i32 [[TMP2]], i32 16
+; CHECK-NEXT:    [[TMP20:%.*]] = insertelement <32 x i32> [[TMP19]], i32 [[TMP2]], i32 17
+; CHECK-NEXT:    [[TMP21:%.*]] = insertelement <32 x i32> [[TMP20]], i32 [[TMP2]], i32 18
+; CHECK-NEXT:    [[TMP22:%.*]] = insertelement <32 x i32> [[TMP21]], i32 [[TMP2]], i32 19
+; CHECK-NEXT:    [[TMP23:%.*]] = insertelement <32 x i32> [[TMP22]], i32 [[TMP2]], i32 20
+; CHECK-NEXT:    [[TMP24:%.*]] = insertelement <32 x i32> [[TMP23]], i32 [[TMP2]], i32 21
+; CHECK-NEXT:    [[TMP25:%.*]] = insertelement <32 x i32> [[TMP24]], i32 [[TMP2]], i32 22
+; CHECK-NEXT:    [[TMP26:%.*]] = insertelement <32 x i32> [[TMP25]], i32 [[TMP2]], i32 23
+; CHECK-NEXT:    [[TMP27:%.*]] = insertelement <32 x i32> [[TMP26]], i32 [[TMP2]], i32 24
+; CHECK-NEXT:    [[TMP28:%.*]] = insertelement <32 x i32> [[TMP27]], i32 [[TMP2]], i32 25
+; CHECK-NEXT:    [[TMP29:%.*]] = insertelement <32 x i32> [[TMP28]], i32 [[TMP2]], i32 26
+; CHECK-NEXT:    [[TMP30:%.*]] = insertelement <32 x i32> [[TMP29]], i32 [[TMP2]], i32 27
+; CHECK-NEXT:    [[TMP31:%.*]] = insertelement <32 x i32> [[TMP30]], i32 [[TMP2]], i32 28
+; CHECK-NEXT:    [[TMP32:%.*]] = insertelement <32 x i32> [[TMP31]], i32 [[TMP2]], i32 29
+; CHECK-NEXT:    [[TMP33:%.*]] = insertelement <32 x i32> [[TMP32]], i32 [[TMP2]], i32 30
+; CHECK-NEXT:    [[TMP34:%.*]] = insertelement <32 x i32> [[TMP33]], i32 [[TMP2]], i32 31
+; CHECK-NEXT:    [[TMP35:%.*]] = call i32 @llvm.vector.reduce.mul.v32i32(<32 x i32> [[TMP34]])
+; CHECK-NEXT:    [[TMP36:%.*]] = call i32 @llvm.vector.reduce.mul.v16i32(<16 x i32> [[TMP1]])
+; CHECK-NEXT:    [[OP_RDX:%.*]] = mul i32 [[TMP35]], [[TMP36]]
+; CHECK-NEXT:    [[OP_RDX1:%.*]] = mul i32 [[OP_RDX]], [[TMP2]]
+; CHECK-NEXT:    [[OP_RDX2:%.*]] = mul i32 [[TMP2]], [[TMP2]]
+; CHECK-NEXT:    [[OP_RDX3:%.*]] = mul i32 [[TMP2]], [[TMP2]]
+; CHECK-NEXT:    [[OP_RDX4:%.*]] = mul i32 [[TMP2]], [[TMP2]]
+; CHECK-NEXT:    [[OP_RDX5:%.*]] = mul i32 [[TMP2]], [[TMP2]]
+; CHECK-NEXT:    [[OP_RDX6:%.*]] = mul i32 [[TMP2]], [[TMP2]]
+; CHECK-NEXT:    [[OP_RDX7:%.*]] = mul i32 [[OP_RDX1]], [[OP_RDX2]]
+; CHECK-NEXT:    [[OP_RDX8:%.*]] = mul i32 [[OP_RDX3]], [[OP_RDX4]]
+; CHECK-NEXT:    [[OP_RDX9:%.*]] = mul i32 [[OP_RDX5]], [[OP_RDX6]]
+; CHECK-NEXT:    [[OP_RDX10:%.*]] = mul i32 [[OP_RDX7]], [[OP_RDX8]]
+; CHECK-NEXT:    [[OP_RDX11:%.*]] = mul i32 [[OP_RDX9]], [[TMP2]]
+; CHECK-NEXT:    [[OP_RDX12:%.*]] = mul i32 [[OP_RDX10]], [[OP_RDX11]]
+; CHECK-NEXT:    [[VAL64:%.*]] = add i32 undef, [[OP_RDX12]]
 ; CHECK-NEXT:    [[VAL65:%.*]] = sext i32 [[VAL64]] to i64
 ; CHECK-NEXT:    ret i64 [[VAL65]]
 ;
diff --git a/llvm/test/Transforms/SLPVectorizer/X86/used-reduced-op.ll b/llvm/test/Transforms/SLPVectorizer/X86/used-reduced-op.ll
--- a/llvm/test/Transforms/SLPVectorizer/X86/used-reduced-op.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/used-reduced-op.ll
@@ -21,222 +21,233 @@
 ; CHECK-NEXT:    [[TMP10:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 3, i64 1), align 4
 ; CHECK-NEXT:    [[TMP11:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 3, i64 2), align 8
 ; CHECK-NEXT:    [[TMP12:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 3, i64 3), align 4
-; CHECK-NEXT:    [[TMP13:%.*]] = load <16 x i32>, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 4, i64 0), align 16
+; CHECK-NEXT:    [[TMP13:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 4, i64 0), align 16
+; CHECK-NEXT:    [[TMP14:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 4, i64 1), align 4
+; CHECK-NEXT:    [[TMP15:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 4, i64 2), align 8
+; CHECK-NEXT:    [[TMP16:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 4, i64 3), align 4
+; CHECK-NEXT:    [[TMP17:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 5, i64 0), align 16
+; CHECK-NEXT:    [[TMP18:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 5, i64 1), align 4
+; CHECK-NEXT:    [[TMP19:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 5, i64 2), align 8
+; CHECK-NEXT:    [[TMP20:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 5, i64 3), align 4
+; CHECK-NEXT:    [[TMP21:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 6, i64 0), align 16
+; CHECK-NEXT:    [[TMP22:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 6, i64 1), align 4
+; CHECK-NEXT:    [[TMP23:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 6, i64 2), align 8
+; CHECK-NEXT:    [[TMP24:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 6, i64 3), align 4
+; CHECK-NEXT:    [[TMP25:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 7, i64 0), align 16
+; CHECK-NEXT:    [[TMP26:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 7, i64 1), align 4
+; CHECK-NEXT:    [[TMP27:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 7, i64 2), align 8
+; CHECK-NEXT:    [[TMP28:%.*]] = load i32, ptr getelementptr inbounds ([8 x [4 x i32]], ptr @k, i64 0, i64 7, i64 3), align 4
 ; CHECK-NEXT:    br label [[FOR_COND:%.*]]
 ; CHECK:       for.cond:
 ; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_COND]] ], [ 0, [[ENTRY:%.*]] ]
 ; CHECK-NEXT:    [[B_0:%.*]] = phi i32 [ [[SPEC_SELECT8_3_7:%.*]], [[FOR_COND]] ], [ undef, [[ENTRY]] ]
-; CHECK-NEXT:    [[TMP14:%.*]] = trunc i64 [[INDVARS_IV]] to i32
-; CHECK-NEXT:    [[TMP15:%.*]] = add i32 [[TMP14]], -183
-; CHECK-NEXT:    [[TMP16:%.*]] = insertelement <4 x i32> poison, i32 [[TMP15]], i32 0
-; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <4 x i32> [[TMP16]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP17:%.*]] = sub <4 x i32> [[SHUFFLE]], [[TMP0]]
-; CHECK-NEXT:    [[TMP18:%.*]] = icmp slt <4 x i32> [[TMP17]], zeroinitializer
-; CHECK-NEXT:    [[TMP19:%.*]] = sub nsw <4 x i32> zeroinitializer, [[TMP17]]
-; CHECK-NEXT:    [[TMP20:%.*]] = select <4 x i1> [[TMP18]], <4 x i32> [[TMP19]], <4 x i32> [[TMP17]]
-; CHECK-NEXT:    [[TMP21:%.*]] = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> [[TMP20]])
-; CHECK-NEXT:    [[OP_RDX:%.*]] = icmp slt i32 [[TMP21]], [[B_0]]
-; CHECK-NEXT:    [[OP_RDX1:%.*]] = select i1 [[OP_RDX]], i32 [[TMP21]], i32 [[B_0]]
-; CHECK-NEXT:    [[SUB_116:%.*]] = sub i32 [[TMP15]], [[TMP1]]
-; CHECK-NEXT:    [[TMP22:%.*]] = icmp slt i32 [[SUB_116]], 0
+; CHECK-NEXT:    [[TMP29:%.*]] = trunc i64 [[INDVARS_IV]] to i32
+; CHECK-NEXT:    [[TMP30:%.*]] = add i32 [[TMP29]], -183
+; CHECK-NEXT:    [[TMP31:%.*]] = insertelement <4 x i32> poison, i32 [[TMP30]], i32 0
+; CHECK-NEXT:    [[TMP32:%.*]] = shufflevector <4 x i32> [[TMP31]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP33:%.*]] = sub <4 x i32> [[TMP32]], [[TMP0]]
+; CHECK-NEXT:    [[TMP34:%.*]] = icmp slt <4 x i32> [[TMP33]], zeroinitializer
+; CHECK-NEXT:    [[TMP35:%.*]] = sub nsw <4 x i32> zeroinitializer, [[TMP33]]
+; CHECK-NEXT:    [[TMP36:%.*]] = select <4 x i1> [[TMP34]], <4 x i32> [[TMP35]], <4 x i32> [[TMP33]]
+; CHECK-NEXT:    [[TMP37:%.*]] = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> [[TMP36]])
+; CHECK-NEXT:    [[OP_RDX:%.*]] = icmp slt i32 [[TMP37]], [[B_0]]
+; CHECK-NEXT:    [[OP_RDX1:%.*]] = select i1 [[OP_RDX]], i32 [[TMP37]], i32 [[B_0]]
+; CHECK-NEXT:    [[SUB_116:%.*]] = sub i32 [[TMP30]], [[TMP1]]
+; CHECK-NEXT:    [[TMP38:%.*]] = icmp slt i32 [[SUB_116]], 0
 ; CHECK-NEXT:    [[NEG_117:%.*]] = sub nsw i32 0, [[SUB_116]]
-; CHECK-NEXT:    [[TMP23:%.*]] = select i1 [[TMP22]], i32 [[NEG_117]], i32 [[SUB_116]]
-; CHECK-NEXT:    [[CMP12_118:%.*]] = icmp slt i32 [[TMP23]], [[OP_RDX1]]
-; CHECK-NEXT:    [[SPEC_SELECT8_120:%.*]] = select i1 [[CMP12_118]], i32 [[TMP23]], i32 [[OP_RDX1]]
-; CHECK-NEXT:    [[SUB_1_1:%.*]] = sub i32 [[TMP15]], [[TMP2]]
-; CHECK-NEXT:    [[TMP24:%.*]] = icmp slt i32 [[SUB_1_1]], 0
+; CHECK-NEXT:    [[TMP39:%.*]] = select i1 [[TMP38]], i32 [[NEG_117]], i32 [[SUB_116]]
+; CHECK-NEXT:    [[CMP12_118:%.*]] = icmp slt i32 [[TMP39]], [[OP_RDX1]]
+; CHECK-NEXT:    [[SPEC_SELECT8_120:%.*]] = select i1 [[CMP12_118]], i32 [[TMP39]], i32 [[OP_RDX1]]
+; CHECK-NEXT:    [[SUB_1_1:%.*]] = sub i32 [[TMP30]], [[TMP2]]
+; CHECK-NEXT:    [[TMP40:%.*]] = icmp slt i32 [[SUB_1_1]], 0
 ; CHECK-NEXT:    [[NEG_1_1:%.*]] = sub nsw i32 0, [[SUB_1_1]]
-; CHECK-NEXT:    [[TMP25:%.*]] = select i1 [[TMP24]], i32 [[NEG_1_1]], i32 [[SUB_1_1]]
-; CHECK-NEXT:    [[CMP12_1_1:%.*]] = icmp slt i32 [[TMP25]], [[SPEC_SELECT8_120]]
+; CHECK-NEXT:    [[TMP41:%.*]] = select i1 [[TMP40]], i32 [[NEG_1_1]], i32 [[SUB_1_1]]
+; CHECK-NEXT:    [[CMP12_1_1:%.*]] = icmp slt i32 [[TMP41]], [[SPEC_SELECT8_120]]
 ; CHECK-NEXT:    [[NARROW:%.*]] = or i1 [[CMP12_1_1]], [[CMP12_118]]
-; CHECK-NEXT:    [[SPEC_SELECT8_1_1:%.*]] = select i1 [[CMP12_1_1]], i32 [[TMP25]], i32 [[SPEC_SELECT8_120]]
-; CHECK-NEXT:    [[SUB_2_1:%.*]] = sub i32 [[TMP15]], [[TMP3]]
-; CHECK-NEXT:    [[TMP26:%.*]] = icmp slt i32 [[SUB_2_1]], 0
+; CHECK-NEXT:    [[SPEC_SELECT8_1_1:%.*]] = select i1 [[CMP12_1_1]], i32 [[TMP41]], i32 [[SPEC_SELECT8_120]]
+; CHECK-NEXT:    [[SUB_2_1:%.*]] = sub i32 [[TMP30]], [[TMP3]]
+; CHECK-NEXT:    [[TMP42:%.*]] = icmp slt i32 [[SUB_2_1]], 0
 ; CHECK-NEXT:    [[NEG_2_1:%.*]] = sub nsw i32 0, [[SUB_2_1]]
-; CHECK-NEXT:    [[TMP27:%.*]] = select i1 [[TMP26]], i32 [[NEG_2_1]], i32 [[SUB_2_1]]
-; CHECK-NEXT:    [[CMP12_2_1:%.*]] = icmp slt i32 [[TMP27]], [[SPEC_SELECT8_1_1]]
+; CHECK-NEXT:    [[TMP43:%.*]] = select i1 [[TMP42]], i32 [[NEG_2_1]], i32 [[SUB_2_1]]
+; CHECK-NEXT:    [[CMP12_2_1:%.*]] = icmp slt i32 [[TMP43]], [[SPEC_SELECT8_1_1]]
 ; CHECK-NEXT:    [[NARROW34:%.*]] = or i1 [[CMP12_2_1]], [[NARROW]]
-; CHECK-NEXT:    [[SPEC_SELECT8_2_1:%.*]] = select i1 [[CMP12_2_1]], i32 [[TMP27]], i32 [[SPEC_SELECT8_1_1]]
-; CHECK-NEXT:    [[SUB_3_1:%.*]] = sub i32 [[TMP15]], [[TMP4]]
-; CHECK-NEXT:    [[TMP28:%.*]] = icmp slt i32 [[SUB_3_1]], 0
+; CHECK-NEXT:    [[SPEC_SELECT8_2_1:%.*]] = select i1 [[CMP12_2_1]], i32 [[TMP43]], i32 [[SPEC_SELECT8_1_1]]
+; CHECK-NEXT:    [[SUB_3_1:%.*]] = sub i32 [[TMP30]], [[TMP4]]
+; CHECK-NEXT:    [[TMP44:%.*]] = icmp slt i32 [[SUB_3_1]], 0
 ; CHECK-NEXT:    [[NEG_3_1:%.*]] = sub nsw i32 0, [[SUB_3_1]]
-; CHECK-NEXT:    [[TMP29:%.*]] = select i1 [[TMP28]], i32 [[NEG_3_1]], i32 [[SUB_3_1]]
-; CHECK-NEXT:    [[CMP12_3_1:%.*]] = icmp slt i32 [[TMP29]], [[SPEC_SELECT8_2_1]]
+; CHECK-NEXT:    [[TMP45:%.*]] = select i1 [[TMP44]], i32 [[NEG_3_1]], i32 [[SUB_3_1]]
+; CHECK-NEXT:    [[CMP12_3_1:%.*]] = icmp slt i32 [[TMP45]], [[SPEC_SELECT8_2_1]]
 ; CHECK-NEXT:    [[NARROW35:%.*]] = or i1 [[CMP12_3_1]], [[NARROW34]]
 ; CHECK-NEXT:    [[SPEC_SELECT_3_1:%.*]] = zext i1 [[NARROW35]] to i32
-; CHECK-NEXT:    [[SPEC_SELECT8_3_1:%.*]] = select i1 [[CMP12_3_1]], i32 [[TMP29]], i32 [[SPEC_SELECT8_2_1]]
-; CHECK-NEXT:    [[SUB_222:%.*]] = sub i32 [[TMP15]], [[TMP5]]
-; CHECK-NEXT:    [[TMP30:%.*]] = icmp slt i32 [[SUB_222]], 0
+; CHECK-NEXT:    [[SPEC_SELECT8_3_1:%.*]] = select i1 [[CMP12_3_1]], i32 [[TMP45]], i32 [[SPEC_SELECT8_2_1]]
+; CHECK-NEXT:    [[SUB_222:%.*]] = sub i32 [[TMP30]], [[TMP5]]
+; CHECK-NEXT:    [[TMP46:%.*]] = icmp slt i32 [[SUB_222]], 0
 ; CHECK-NEXT:    [[NEG_223:%.*]] = sub nsw i32 0, [[SUB_222]]
-; CHECK-NEXT:    [[TMP31:%.*]] = select i1 [[TMP30]], i32 [[NEG_223]], i32 [[SUB_222]]
-; CHECK-NEXT:    [[CMP12_224:%.*]] = icmp slt i32 [[TMP31]], [[SPEC_SELECT8_3_1]]
-; CHECK-NEXT:    [[SPEC_SELECT8_226:%.*]] = select i1 [[CMP12_224]], i32 [[TMP31]], i32 [[SPEC_SELECT8_3_1]]
-; CHECK-NEXT:    [[SUB_1_2:%.*]] = sub i32 [[TMP15]], [[TMP6]]
-; CHECK-NEXT:    [[TMP32:%.*]] = icmp slt i32 [[SUB_1_2]], 0
+; CHECK-NEXT:    [[TMP47:%.*]] = select i1 [[TMP46]], i32 [[NEG_223]], i32 [[SUB_222]]
+; CHECK-NEXT:    [[CMP12_224:%.*]] = icmp slt i32 [[TMP47]], [[SPEC_SELECT8_3_1]]
+; CHECK-NEXT:    [[SPEC_SELECT8_226:%.*]] = select i1 [[CMP12_224]], i32 [[TMP47]], i32 [[SPEC_SELECT8_3_1]]
+; CHECK-NEXT:    [[SUB_1_2:%.*]] = sub i32 [[TMP30]], [[TMP6]]
+; CHECK-NEXT:    [[TMP48:%.*]] = icmp slt i32 [[SUB_1_2]], 0
 ; CHECK-NEXT:    [[NEG_1_2:%.*]] = sub nsw i32 0, [[SUB_1_2]]
-; CHECK-NEXT:    [[TMP33:%.*]] = select i1 [[TMP32]], i32 [[NEG_1_2]], i32 [[SUB_1_2]]
-; CHECK-NEXT:    [[CMP12_1_2:%.*]] = icmp slt i32 [[TMP33]], [[SPEC_SELECT8_226]]
-; CHECK-NEXT:    [[TMP34:%.*]] = or i1 [[CMP12_1_2]], [[CMP12_224]]
-; CHECK-NEXT:    [[SPEC_SELECT8_1_2:%.*]] = select i1 [[CMP12_1_2]], i32 [[TMP33]], i32 [[SPEC_SELECT8_226]]
-; CHECK-NEXT:    [[SUB_2_2:%.*]] = sub i32 [[TMP15]], [[TMP7]]
-; CHECK-NEXT:    [[TMP35:%.*]] = icmp slt i32 [[SUB_2_2]], 0
+; CHECK-NEXT:    [[TMP49:%.*]] = select i1 [[TMP48]], i32 [[NEG_1_2]], i32 [[SUB_1_2]]
+; CHECK-NEXT:    [[CMP12_1_2:%.*]] = icmp slt i32 [[TMP49]], [[SPEC_SELECT8_226]]
+; CHECK-NEXT:    [[TMP50:%.*]] = or i1 [[CMP12_1_2]], [[CMP12_224]]
+; CHECK-NEXT:    [[SPEC_SELECT8_1_2:%.*]] = select i1 [[CMP12_1_2]], i32 [[TMP49]], i32 [[SPEC_SELECT8_226]]
+; CHECK-NEXT:    [[SUB_2_2:%.*]] = sub i32 [[TMP30]], [[TMP7]]
+; CHECK-NEXT:    [[TMP51:%.*]] = icmp slt i32 [[SUB_2_2]], 0
 ; CHECK-NEXT:    [[NEG_2_2:%.*]] = sub nsw i32 0, [[SUB_2_2]]
-; CHECK-NEXT:    [[TMP36:%.*]] = select i1 [[TMP35]], i32 [[NEG_2_2]], i32 [[SUB_2_2]]
-; CHECK-NEXT:    [[CMP12_2_2:%.*]] = icmp slt i32 [[TMP36]], [[SPEC_SELECT8_1_2]]
-; CHECK-NEXT:    [[TMP37:%.*]] = or i1 [[CMP12_2_2]], [[TMP34]]
-; CHECK-NEXT:    [[SPEC_SELECT8_2_2:%.*]] = select i1 [[CMP12_2_2]], i32 [[TMP36]], i32 [[SPEC_SELECT8_1_2]]
-; CHECK-NEXT:    [[SUB_3_2:%.*]] = sub i32 [[TMP15]], [[TMP8]]
-; CHECK-NEXT:    [[TMP38:%.*]] = icmp slt i32 [[SUB_3_2]], 0
+; CHECK-NEXT:    [[TMP52:%.*]] = select i1 [[TMP51]], i32 [[NEG_2_2]], i32 [[SUB_2_2]]
+; CHECK-NEXT:    [[CMP12_2_2:%.*]] = icmp slt i32 [[TMP52]], [[SPEC_SELECT8_1_2]]
+; CHECK-NEXT:    [[TMP53:%.*]] = or i1 [[CMP12_2_2]], [[TMP50]]
+; CHECK-NEXT:    [[SPEC_SELECT8_2_2:%.*]] = select i1 [[CMP12_2_2]], i32 [[TMP52]], i32 [[SPEC_SELECT8_1_2]]
+; CHECK-NEXT:    [[SUB_3_2:%.*]] = sub i32 [[TMP30]], [[TMP8]]
+; CHECK-NEXT:    [[TMP54:%.*]] = icmp slt i32 [[SUB_3_2]], 0
 ; CHECK-NEXT:    [[NEG_3_2:%.*]] = sub nsw i32 0, [[SUB_3_2]]
-; CHECK-NEXT:    [[TMP39:%.*]] = select i1 [[TMP38]], i32 [[NEG_3_2]], i32 [[SUB_3_2]]
-; CHECK-NEXT:    [[CMP12_3_2:%.*]] = icmp slt i32 [[TMP39]], [[SPEC_SELECT8_2_2]]
-; CHECK-NEXT:    [[TMP40:%.*]] = or i1 [[CMP12_3_2]], [[TMP37]]
-; CHECK-NEXT:    [[SPEC_SELECT_3_2:%.*]] = select i1 [[TMP40]], i32 2, i32 [[SPEC_SELECT_3_1]]
-; CHECK-NEXT:    [[SPEC_SELECT8_3_2:%.*]] = select i1 [[CMP12_3_2]], i32 [[TMP39]], i32 [[SPEC_SELECT8_2_2]]
-; CHECK-NEXT:    [[SUB_328:%.*]] = sub i32 [[TMP15]], [[TMP9]]
-; CHECK-NEXT:    [[TMP41:%.*]] = icmp slt i32 [[SUB_328]], 0
+; CHECK-NEXT:    [[TMP55:%.*]] = select i1 [[TMP54]], i32 [[NEG_3_2]], i32 [[SUB_3_2]]
+; CHECK-NEXT:    [[CMP12_3_2:%.*]] = icmp slt i32 [[TMP55]], [[SPEC_SELECT8_2_2]]
+; CHECK-NEXT:    [[TMP56:%.*]] = or i1 [[CMP12_3_2]], [[TMP53]]
+; CHECK-NEXT:    [[SPEC_SELECT_3_2:%.*]] = select i1 [[TMP56]], i32 2, i32 [[SPEC_SELECT_3_1]]
+; CHECK-NEXT:    [[SPEC_SELECT8_3_2:%.*]] = select i1 [[CMP12_3_2]], i32 [[TMP55]], i32 [[SPEC_SELECT8_2_2]]
+; CHECK-NEXT:    [[SUB_328:%.*]] = sub i32 [[TMP30]], [[TMP9]]
+; CHECK-NEXT:    [[TMP57:%.*]] = icmp slt i32 [[SUB_328]], 0
 ; CHECK-NEXT:    [[NEG_329:%.*]] = sub nsw i32 0, [[SUB_328]]
-; CHECK-NEXT:    [[TMP42:%.*]] = select i1 [[TMP41]], i32 [[NEG_329]], i32 [[SUB_328]]
-; CHECK-NEXT:    [[CMP12_330:%.*]] = icmp slt i32 [[TMP42]], [[SPEC_SELECT8_3_2]]
-; CHECK-NEXT:    [[SPEC_SELECT8_332:%.*]] = select i1 [[CMP12_330]], i32 [[TMP42]], i32 [[SPEC_SELECT8_3_2]]
-; CHECK-NEXT:    [[SUB_1_3:%.*]] = sub i32 [[TMP15]], [[TMP10]]
-; CHECK-NEXT:    [[TMP43:%.*]] = icmp slt i32 [[SUB_1_3]], 0
+; CHECK-NEXT:    [[TMP58:%.*]] = select i1 [[TMP57]], i32 [[NEG_329]], i32 [[SUB_328]]
+; CHECK-NEXT:    [[CMP12_330:%.*]] = icmp slt i32 [[TMP58]], [[SPEC_SELECT8_3_2]]
+; CHECK-NEXT:    [[SPEC_SELECT8_332:%.*]] = select i1 [[CMP12_330]], i32 [[TMP58]], i32 [[SPEC_SELECT8_3_2]]
+; CHECK-NEXT:    [[SUB_1_3:%.*]] = sub i32 [[TMP30]], [[TMP10]]
+; CHECK-NEXT:    [[TMP59:%.*]] = icmp slt i32 [[SUB_1_3]], 0
 ; CHECK-NEXT:    [[NEG_1_3:%.*]] = sub nsw i32 0, [[SUB_1_3]]
-; CHECK-NEXT:    [[TMP44:%.*]] = select i1 [[TMP43]], i32 [[NEG_1_3]], i32 [[SUB_1_3]]
-; CHECK-NEXT:    [[CMP12_1_3:%.*]] = icmp slt i32 [[TMP44]], [[SPEC_SELECT8_332]]
-; CHECK-NEXT:    [[TMP45:%.*]] = or i1 [[CMP12_1_3]], [[CMP12_330]]
-; CHECK-NEXT:    [[SPEC_SELECT8_1_3:%.*]] = select i1 [[CMP12_1_3]], i32 [[TMP44]], i32 [[SPEC_SELECT8_332]]
-; CHECK-NEXT:    [[SUB_2_3:%.*]] = sub i32 [[TMP15]], [[TMP11]]
-; CHECK-NEXT:    [[TMP46:%.*]] = icmp slt i32 [[SUB_2_3]], 0
+; CHECK-NEXT:    [[TMP60:%.*]] = select i1 [[TMP59]], i32 [[NEG_1_3]], i32 [[SUB_1_3]]
+; CHECK-NEXT:    [[CMP12_1_3:%.*]] = icmp slt i32 [[TMP60]], [[SPEC_SELECT8_332]]
+; CHECK-NEXT:    [[TMP61:%.*]] = or i1 [[CMP12_1_3]], [[CMP12_330]]
+; CHECK-NEXT:    [[SPEC_SELECT8_1_3:%.*]] = select i1 [[CMP12_1_3]], i32 [[TMP60]], i32 [[SPEC_SELECT8_332]]
+; CHECK-NEXT:    [[SUB_2_3:%.*]] = sub i32 [[TMP30]], [[TMP11]]
+; CHECK-NEXT:    [[TMP62:%.*]] = icmp slt i32 [[SUB_2_3]], 0
 ; CHECK-NEXT:    [[NEG_2_3:%.*]] = sub nsw i32 0, [[SUB_2_3]]
-; CHECK-NEXT:    [[TMP47:%.*]] = select i1 [[TMP46]], i32 [[NEG_2_3]], i32 [[SUB_2_3]]
-; CHECK-NEXT:    [[CMP12_2_3:%.*]] = icmp slt i32 [[TMP47]], [[SPEC_SELECT8_1_3]]
-; CHECK-NEXT:    [[TMP48:%.*]] = or i1 [[CMP12_2_3]], [[TMP45]]
-; CHECK-NEXT:    [[SPEC_SELECT8_2_3:%.*]] = select i1 [[CMP12_2_3]], i32 [[TMP47]], i32 [[SPEC_SELECT8_1_3]]
-; CHECK-NEXT:    [[SUB_3_3:%.*]] = sub i32 [[TMP15]], [[TMP12]]
-; CHECK-NEXT:    [[TMP49:%.*]] = icmp slt i32 [[SUB_3_3]], 0
+; CHECK-NEXT:    [[TMP63:%.*]] = select i1 [[TMP62]], i32 [[NEG_2_3]], i32 [[SUB_2_3]]
+; CHECK-NEXT:    [[CMP12_2_3:%.*]] = icmp slt i32 [[TMP63]], [[SPEC_SELECT8_1_3]]
+; CHECK-NEXT:    [[TMP64:%.*]] = or i1 [[CMP12_2_3]], [[TMP61]]
+; CHECK-NEXT:    [[SPEC_SELECT8_2_3:%.*]] = select i1 [[CMP12_2_3]], i32 [[TMP63]], i32 [[SPEC_SELECT8_1_3]]
+; CHECK-NEXT:    [[SUB_3_3:%.*]] = sub i32 [[TMP30]], [[TMP12]]
+; CHECK-NEXT:    [[TMP65:%.*]] = icmp slt i32 [[SUB_3_3]], 0
 ; CHECK-NEXT:    [[NEG_3_3:%.*]] = sub nsw i32 0, [[SUB_3_3]]
-; CHECK-NEXT:    [[TMP50:%.*]] = select i1 [[TMP49]], i32 [[NEG_3_3]], i32 [[SUB_3_3]]
-; CHECK-NEXT:    [[CMP12_3_3:%.*]] = icmp slt i32 [[TMP50]], [[SPEC_SELECT8_2_3]]
-; CHECK-NEXT:    [[TMP51:%.*]] = or i1 [[CMP12_3_3]], [[TMP48]]
-; CHECK-NEXT:    [[SPEC_SELECT_3_3:%.*]] = select i1 [[TMP51]], i32 3, i32 [[SPEC_SELECT_3_2]]
-; CHECK-NEXT:    [[SPEC_SELECT8_3_3:%.*]] = select i1 [[CMP12_3_3]], i32 [[TMP50]], i32 [[SPEC_SELECT8_2_3]]
-; CHECK-NEXT:    [[TMP52:%.*]] = insertelement <16 x i32> poison, i32 [[TMP15]], i32 0
-; CHECK-NEXT:    [[SHUFFLE2:%.*]] = shufflevector <16 x i32> [[TMP52]], <16 x i32> poison, <16 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP53:%.*]] = sub <16 x i32> [[SHUFFLE2]], [[TMP13]]
-; CHECK-NEXT:    [[TMP54:%.*]] = extractelement <16 x i32> [[TMP53]], i32 0
-; CHECK-NEXT:    [[NEG_4:%.*]] = sub nsw i32 0, [[TMP54]]
-; CHECK-NEXT:    [[TMP55:%.*]] = icmp slt <16 x i32> [[TMP53]], zeroinitializer
-; CHECK-NEXT:    [[TMP56:%.*]] = extractelement <16 x i1> [[TMP55]], i32 0
-; CHECK-NEXT:    [[TMP57:%.*]] = select i1 [[TMP56]], i32 [[NEG_4]], i32 [[TMP54]]
-; CHECK-NEXT:    [[CMP12_4:%.*]] = icmp slt i32 [[TMP57]], [[SPEC_SELECT8_3_3]]
-; CHECK-NEXT:    [[SPEC_SELECT8_4:%.*]] = select i1 [[CMP12_4]], i32 [[TMP57]], i32 [[SPEC_SELECT8_3_3]]
-; CHECK-NEXT:    [[TMP58:%.*]] = extractelement <16 x i32> [[TMP53]], i32 1
-; CHECK-NEXT:    [[NEG_1_4:%.*]] = sub nsw i32 0, [[TMP58]]
-; CHECK-NEXT:    [[TMP59:%.*]] = extractelement <16 x i1> [[TMP55]], i32 1
-; CHECK-NEXT:    [[TMP60:%.*]] = select i1 [[TMP59]], i32 [[NEG_1_4]], i32 [[TMP58]]
-; CHECK-NEXT:    [[CMP12_1_4:%.*]] = icmp slt i32 [[TMP60]], [[SPEC_SELECT8_4]]
-; CHECK-NEXT:    [[TMP61:%.*]] = or i1 [[CMP12_1_4]], [[CMP12_4]]
-; CHECK-NEXT:    [[SPEC_SELECT8_1_4:%.*]] = select i1 [[CMP12_1_4]], i32 [[TMP60]], i32 [[SPEC_SELECT8_4]]
-; CHECK-NEXT:    [[TMP62:%.*]] = extractelement <16 x i32> [[TMP53]], i32 2
-; CHECK-NEXT:    [[NEG_2_4:%.*]] = sub nsw i32 0, [[TMP62]]
-; CHECK-NEXT:    [[TMP63:%.*]] = extractelement <16 x i1> [[TMP55]], i32 2
-; CHECK-NEXT:    [[TMP64:%.*]] = select i1 [[TMP63]], i32 [[NEG_2_4]], i32 [[TMP62]]
-; CHECK-NEXT:    [[CMP12_2_4:%.*]] = icmp slt i32 [[TMP64]], [[SPEC_SELECT8_1_4]]
-; CHECK-NEXT:    [[TMP65:%.*]] = or i1 [[CMP12_2_4]], [[TMP61]]
-; CHECK-NEXT:    [[SPEC_SELECT8_2_4:%.*]] = select i1 [[CMP12_2_4]], i32 [[TMP64]], i32 [[SPEC_SELECT8_1_4]]
-; CHECK-NEXT:    [[TMP66:%.*]] = extractelement <16 x i32> [[TMP53]], i32 3
-; CHECK-NEXT:    [[NEG_3_4:%.*]] = sub nsw i32 0, [[TMP66]]
-; CHECK-NEXT:    [[TMP67:%.*]] = extractelement <16 x i1> [[TMP55]], i32 3
-; CHECK-NEXT:    [[TMP68:%.*]] = select i1 [[TMP67]], i32 [[NEG_3_4]], i32 [[TMP66]]
-; CHECK-NEXT:    [[CMP12_3_4:%.*]] = icmp slt i32 [[TMP68]], [[SPEC_SELECT8_2_4]]
-; CHECK-NEXT:    [[TMP69:%.*]] = or i1 [[CMP12_3_4]], [[TMP65]]
-; CHECK-NEXT:    [[SPEC_SELECT_3_4:%.*]] = select i1 [[TMP69]], i32 4, i32 [[SPEC_SELECT_3_3]]
-; CHECK-NEXT:    [[SPEC_SELECT8_3_4:%.*]] = select i1 [[CMP12_3_4]], i32 [[TMP68]], i32 [[SPEC_SELECT8_2_4]]
-; CHECK-NEXT:    [[TMP70:%.*]] = extractelement <16 x i32> [[TMP53]], i32 4
-; CHECK-NEXT:    [[NEG_5:%.*]] = sub nsw i32 0, [[TMP70]]
-; CHECK-NEXT:    [[TMP71:%.*]] = extractelement <16 x i1> [[TMP55]], i32 4
-; CHECK-NEXT:    [[TMP72:%.*]] = select i1 [[TMP71]], i32 [[NEG_5]], i32 [[TMP70]]
-; CHECK-NEXT:    [[CMP12_5:%.*]] = icmp slt i32 [[TMP72]], [[SPEC_SELECT8_3_4]]
-; CHECK-NEXT:    [[SPEC_SELECT8_5:%.*]] = select i1 [[CMP12_5]], i32 [[TMP72]], i32 [[SPEC_SELECT8_3_4]]
-; CHECK-NEXT:    [[TMP73:%.*]] = extractelement <16 x i32> [[TMP53]], i32 5
-; CHECK-NEXT:    [[NEG_1_5:%.*]] = sub nsw i32 0, [[TMP73]]
-; CHECK-NEXT:    [[TMP74:%.*]] = extractelement <16 x i1> [[TMP55]], i32 5
-; CHECK-NEXT:    [[TMP75:%.*]] = select i1 [[TMP74]], i32 [[NEG_1_5]], i32 [[TMP73]]
-; CHECK-NEXT:    [[CMP12_1_5:%.*]] = icmp slt i32 [[TMP75]], [[SPEC_SELECT8_5]]
-; CHECK-NEXT:    [[TMP76:%.*]] = or i1 [[CMP12_1_5]], [[CMP12_5]]
-; CHECK-NEXT:    [[SPEC_SELECT8_1_5:%.*]] = select i1 [[CMP12_1_5]], i32 [[TMP75]], i32 [[SPEC_SELECT8_5]]
-; CHECK-NEXT:    [[TMP77:%.*]] = extractelement <16 x i32> [[TMP53]], i32 6
-; CHECK-NEXT:    [[NEG_2_5:%.*]] = sub nsw i32 0, [[TMP77]]
-; CHECK-NEXT:    [[TMP78:%.*]] = extractelement <16 x i1> [[TMP55]], i32 6
-; CHECK-NEXT:    [[TMP79:%.*]] = select i1 [[TMP78]], i32 [[NEG_2_5]], i32 [[TMP77]]
-; CHECK-NEXT:    [[CMP12_2_5:%.*]] = icmp slt i32 [[TMP79]], [[SPEC_SELECT8_1_5]]
-; CHECK-NEXT:    [[TMP80:%.*]] = or i1 [[CMP12_2_5]], [[TMP76]]
-; CHECK-NEXT:    [[SPEC_SELECT8_2_5:%.*]] = select i1 [[CMP12_2_5]], i32 [[TMP79]], i32 [[SPEC_SELECT8_1_5]]
-; CHECK-NEXT:    [[TMP81:%.*]] = extractelement <16 x i32> [[TMP53]], i32 7
-; CHECK-NEXT:    [[NEG_3_5:%.*]] = sub nsw i32 0, [[TMP81]]
-; CHECK-NEXT:    [[TMP82:%.*]] = extractelement <16 x i1> [[TMP55]], i32 7
-; CHECK-NEXT:    [[TMP83:%.*]] = select i1 [[TMP82]], i32 [[NEG_3_5]], i32 [[TMP81]]
-; CHECK-NEXT:    [[CMP12_3_5:%.*]] = icmp slt i32 [[TMP83]], [[SPEC_SELECT8_2_5]]
-; CHECK-NEXT:    [[TMP84:%.*]] = or i1 [[CMP12_3_5]], [[TMP80]]
-; CHECK-NEXT:    [[SPEC_SELECT_3_5:%.*]] = select i1 [[TMP84]], i32 5, i32 [[SPEC_SELECT_3_4]]
-; CHECK-NEXT:    [[SPEC_SELECT8_3_5:%.*]] = select i1 [[CMP12_3_5]], i32 [[TMP83]], i32 [[SPEC_SELECT8_2_5]]
-; CHECK-NEXT:    [[TMP85:%.*]] = extractelement <16 x i32> [[TMP53]], i32 8
-; CHECK-NEXT:    [[NEG_6:%.*]] = sub nsw i32 0, [[TMP85]]
-; CHECK-NEXT:    [[TMP86:%.*]] = extractelement <16 x i1> [[TMP55]], i32 8
-; CHECK-NEXT:    [[TMP87:%.*]] = select i1 [[TMP86]], i32 [[NEG_6]], i32 [[TMP85]]
-; CHECK-NEXT:    [[CMP12_6:%.*]] = icmp slt i32 [[TMP87]], [[SPEC_SELECT8_3_5]]
-; CHECK-NEXT:    [[SPEC_SELECT8_6:%.*]] = select i1 [[CMP12_6]], i32 [[TMP87]], i32 [[SPEC_SELECT8_3_5]]
-; CHECK-NEXT:    [[TMP88:%.*]] = extractelement <16 x i32> [[TMP53]], i32 9
-; CHECK-NEXT:    [[NEG_1_6:%.*]] = sub nsw i32 0, [[TMP88]]
-; CHECK-NEXT:    [[TMP89:%.*]] = extractelement <16 x i1> [[TMP55]], i32 9
-; CHECK-NEXT:    [[TMP90:%.*]] = select i1 [[TMP89]], i32 [[NEG_1_6]], i32 [[TMP88]]
-; CHECK-NEXT:    [[CMP12_1_6:%.*]] = icmp slt i32 [[TMP90]], [[SPEC_SELECT8_6]]
-; CHECK-NEXT:    [[TMP91:%.*]] = or i1 [[CMP12_1_6]], [[CMP12_6]]
-; CHECK-NEXT:    [[SPEC_SELECT8_1_6:%.*]] = select i1 [[CMP12_1_6]], i32 [[TMP90]], i32 [[SPEC_SELECT8_6]]
-; CHECK-NEXT:    [[TMP92:%.*]] = extractelement <16 x i32> [[TMP53]], i32 10
-; CHECK-NEXT:    [[NEG_2_6:%.*]] = sub nsw i32 0, [[TMP92]]
-; CHECK-NEXT:    [[TMP93:%.*]] = extractelement <16 x i1> [[TMP55]], i32 10
-; CHECK-NEXT:    [[TMP94:%.*]] = select i1 [[TMP93]], i32 [[NEG_2_6]], i32 [[TMP92]]
-; CHECK-NEXT:    [[CMP12_2_6:%.*]] = icmp slt i32 [[TMP94]], [[SPEC_SELECT8_1_6]]
-; CHECK-NEXT:    [[TMP95:%.*]] = or i1 [[CMP12_2_6]], [[TMP91]]
-; CHECK-NEXT:    [[SPEC_SELECT8_2_6:%.*]] = select i1 [[CMP12_2_6]], i32 [[TMP94]], i32 [[SPEC_SELECT8_1_6]]
-; CHECK-NEXT:    [[TMP96:%.*]] = extractelement <16 x i32> [[TMP53]], i32 11
-; CHECK-NEXT:    [[NEG_3_6:%.*]] = sub nsw i32 0, [[TMP96]]
-; CHECK-NEXT:    [[TMP97:%.*]] = extractelement <16 x i1> [[TMP55]], i32 11
-; CHECK-NEXT:    [[TMP98:%.*]] = select i1 [[TMP97]], i32 [[NEG_3_6]], i32 [[TMP96]]
-; CHECK-NEXT:    [[CMP12_3_6:%.*]] = icmp slt i32 [[TMP98]], [[SPEC_SELECT8_2_6]]
-; CHECK-NEXT:    [[TMP99:%.*]] = or i1 [[CMP12_3_6]], [[TMP95]]
-; CHECK-NEXT:    [[SPEC_SELECT_3_6:%.*]] = select i1 [[TMP99]], i32 6, i32 [[SPEC_SELECT_3_5]]
-; CHECK-NEXT:    [[SPEC_SELECT8_3_6:%.*]] = select i1 [[CMP12_3_6]], i32 [[TMP98]], i32 [[SPEC_SELECT8_2_6]]
-; CHECK-NEXT:    [[TMP100:%.*]] = extractelement <16 x i32> [[TMP53]], i32 12
-; CHECK-NEXT:    [[NEG_7:%.*]] = sub nsw i32 0, [[TMP100]]
-; CHECK-NEXT:    [[TMP101:%.*]] = extractelement <16 x i1> [[TMP55]], i32 12
-; CHECK-NEXT:    [[TMP102:%.*]] = select i1 [[TMP101]], i32 [[NEG_7]], i32 [[TMP100]]
+; CHECK-NEXT:    [[TMP66:%.*]] = select i1 [[TMP65]], i32 [[NEG_3_3]], i32 [[SUB_3_3]]
+; CHECK-NEXT:    [[CMP12_3_3:%.*]] = icmp slt i32 [[TMP66]], [[SPEC_SELECT8_2_3]]
+; CHECK-NEXT:    [[TMP67:%.*]] = or i1 [[CMP12_3_3]], [[TMP64]]
+; CHECK-NEXT:    [[SPEC_SELECT_3_3:%.*]] = select i1 [[TMP67]], i32 3, i32 [[SPEC_SELECT_3_2]]
+; CHECK-NEXT:    [[SPEC_SELECT8_3_3:%.*]] = select i1 [[CMP12_3_3]], i32 [[TMP66]], i32 [[SPEC_SELECT8_2_3]]
+; CHECK-NEXT:    [[SUB_4:%.*]] = sub i32 [[TMP30]], [[TMP13]]
+; CHECK-NEXT:    [[TMP68:%.*]] = icmp slt i32 [[SUB_4]], 0
+; CHECK-NEXT:    [[NEG_4:%.*]] = sub nsw i32 0, [[SUB_4]]
+; CHECK-NEXT:    [[TMP69:%.*]] = select i1 [[TMP68]], i32 [[NEG_4]], i32 [[SUB_4]]
+; CHECK-NEXT:    [[CMP12_4:%.*]] = icmp slt i32 [[TMP69]], [[SPEC_SELECT8_3_3]]
+; CHECK-NEXT:    [[SPEC_SELECT8_4:%.*]] = select i1 [[CMP12_4]], i32 [[TMP69]], i32 [[SPEC_SELECT8_3_3]]
+; CHECK-NEXT:    [[SUB_1_4:%.*]] = sub i32 [[TMP30]], [[TMP14]]
+; CHECK-NEXT:    [[TMP70:%.*]] = icmp slt i32 [[SUB_1_4]], 0
+; CHECK-NEXT:    [[NEG_1_4:%.*]] = sub nsw i32 0, [[SUB_1_4]]
+; CHECK-NEXT:    [[TMP71:%.*]] = select i1 [[TMP70]], i32 [[NEG_1_4]], i32 [[SUB_1_4]]
+; CHECK-NEXT:    [[CMP12_1_4:%.*]] = icmp slt i32 [[TMP71]], [[SPEC_SELECT8_4]]
+; CHECK-NEXT:    [[TMP72:%.*]] = or i1 [[CMP12_1_4]], [[CMP12_4]]
+; CHECK-NEXT:    [[SPEC_SELECT8_1_4:%.*]] = select i1 [[CMP12_1_4]], i32 [[TMP71]], i32 [[SPEC_SELECT8_4]]
+; CHECK-NEXT:    [[SUB_2_4:%.*]] = sub i32 [[TMP30]], [[TMP15]]
+; CHECK-NEXT:    [[TMP73:%.*]] = icmp slt i32 [[SUB_2_4]], 0
+; CHECK-NEXT:    [[NEG_2_4:%.*]] = sub nsw i32 0, [[SUB_2_4]]
+; CHECK-NEXT:    [[TMP74:%.*]] = select i1 [[TMP73]], i32 [[NEG_2_4]], i32 [[SUB_2_4]]
+; CHECK-NEXT:    [[CMP12_2_4:%.*]] = icmp slt i32 [[TMP74]], [[SPEC_SELECT8_1_4]]
+; CHECK-NEXT:    [[TMP75:%.*]] = or i1 [[CMP12_2_4]], [[TMP72]]
+; CHECK-NEXT:    [[SPEC_SELECT8_2_4:%.*]] = select i1 [[CMP12_2_4]], i32 [[TMP74]], i32 [[SPEC_SELECT8_1_4]]
+; CHECK-NEXT:    [[SUB_3_4:%.*]] = sub i32 [[TMP30]], [[TMP16]]
+; CHECK-NEXT:    [[TMP76:%.*]] = icmp slt i32 [[SUB_3_4]], 0
+; CHECK-NEXT:    [[NEG_3_4:%.*]] = sub nsw i32 0, [[SUB_3_4]]
+; CHECK-NEXT:    [[TMP77:%.*]] = select i1 [[TMP76]], i32 [[NEG_3_4]], i32 [[SUB_3_4]]
+; CHECK-NEXT:    [[CMP12_3_4:%.*]] = icmp slt i32 [[TMP77]], [[SPEC_SELECT8_2_4]]
+; CHECK-NEXT:    [[TMP78:%.*]] = or i1 [[CMP12_3_4]], [[TMP75]]
+; CHECK-NEXT:    [[SPEC_SELECT_3_4:%.*]] = select i1 [[TMP78]], i32 4, i32 [[SPEC_SELECT_3_3]]
+; CHECK-NEXT:    [[SPEC_SELECT8_3_4:%.*]] = select i1 [[CMP12_3_4]], i32 [[TMP77]], i32 [[SPEC_SELECT8_2_4]]
+; CHECK-NEXT:    [[SUB_5:%.*]] = sub i32 [[TMP30]], [[TMP17]]
+; CHECK-NEXT:    [[TMP79:%.*]] = icmp slt i32 [[SUB_5]], 0
+; CHECK-NEXT:    [[NEG_5:%.*]] = sub nsw i32 0, [[SUB_5]]
+; CHECK-NEXT:    [[TMP80:%.*]] = select i1 [[TMP79]], i32 [[NEG_5]], i32 [[SUB_5]]
+; CHECK-NEXT:    [[CMP12_5:%.*]] = icmp slt i32 [[TMP80]], [[SPEC_SELECT8_3_4]]
+; CHECK-NEXT:    [[SPEC_SELECT8_5:%.*]] = select i1 [[CMP12_5]], i32 [[TMP80]], i32 [[SPEC_SELECT8_3_4]]
+; CHECK-NEXT:    [[SUB_1_5:%.*]] = sub i32 [[TMP30]], [[TMP18]]
+; CHECK-NEXT:    [[TMP81:%.*]] = icmp slt i32 [[SUB_1_5]], 0
+; CHECK-NEXT:    [[NEG_1_5:%.*]] = sub nsw i32 0, [[SUB_1_5]]
+; CHECK-NEXT:    [[TMP82:%.*]] = select i1 [[TMP81]], i32 [[NEG_1_5]], i32 [[SUB_1_5]]
+; CHECK-NEXT:    [[CMP12_1_5:%.*]] = icmp slt i32 [[TMP82]], [[SPEC_SELECT8_5]]
+; CHECK-NEXT:    [[TMP83:%.*]] = or i1 [[CMP12_1_5]], [[CMP12_5]]
+; CHECK-NEXT:    [[SPEC_SELECT8_1_5:%.*]] = select i1 [[CMP12_1_5]], i32 [[TMP82]], i32 [[SPEC_SELECT8_5]]
+; CHECK-NEXT:    [[SUB_2_5:%.*]] = sub i32 [[TMP30]], [[TMP19]]
+; CHECK-NEXT:    [[TMP84:%.*]] = icmp slt i32 [[SUB_2_5]], 0
+; CHECK-NEXT:    [[NEG_2_5:%.*]] = sub nsw i32 0, [[SUB_2_5]]
+; CHECK-NEXT:    [[TMP85:%.*]] = select i1 [[TMP84]], i32 [[NEG_2_5]], i32 [[SUB_2_5]]
+; CHECK-NEXT:    [[CMP12_2_5:%.*]] = icmp slt i32 [[TMP85]], [[SPEC_SELECT8_1_5]]
+; CHECK-NEXT:    [[TMP86:%.*]] = or i1 [[CMP12_2_5]], [[TMP83]]
+; CHECK-NEXT:    [[SPEC_SELECT8_2_5:%.*]] = select i1 [[CMP12_2_5]], i32 [[TMP85]], i32 [[SPEC_SELECT8_1_5]]
+; CHECK-NEXT:    [[SUB_3_5:%.*]] = sub i32 [[TMP30]], [[TMP20]]
+; CHECK-NEXT:    [[TMP87:%.*]] = icmp slt i32 [[SUB_3_5]], 0
+; CHECK-NEXT:    [[NEG_3_5:%.*]] = sub nsw i32 0, [[SUB_3_5]]
+; CHECK-NEXT:    [[TMP88:%.*]] = select i1 [[TMP87]], i32 [[NEG_3_5]], i32 [[SUB_3_5]]
+; CHECK-NEXT:    [[CMP12_3_5:%.*]] = icmp slt i32 [[TMP88]], [[SPEC_SELECT8_2_5]]
+; CHECK-NEXT:    [[TMP89:%.*]] = or i1 [[CMP12_3_5]], [[TMP86]]
+; CHECK-NEXT:    [[SPEC_SELECT_3_5:%.*]] = select i1 [[TMP89]], i32 5, i32 [[SPEC_SELECT_3_4]]
+; CHECK-NEXT:    [[SPEC_SELECT8_3_5:%.*]] = select i1 [[CMP12_3_5]], i32 [[TMP88]], i32 [[SPEC_SELECT8_2_5]]
+; CHECK-NEXT:    [[SUB_6:%.*]] = sub i32 [[TMP30]], [[TMP21]]
+; CHECK-NEXT:    [[TMP90:%.*]] = icmp slt i32 [[SUB_6]], 0
+; CHECK-NEXT:    [[NEG_6:%.*]] = sub nsw i32 0, [[SUB_6]]
+; CHECK-NEXT:    [[TMP91:%.*]] = select i1 [[TMP90]], i32 [[NEG_6]], i32 [[SUB_6]]
+; CHECK-NEXT:    [[CMP12_6:%.*]] = icmp slt i32 [[TMP91]], [[SPEC_SELECT8_3_5]]
+; CHECK-NEXT:    [[SPEC_SELECT8_6:%.*]] = select i1 [[CMP12_6]], i32 [[TMP91]], i32 [[SPEC_SELECT8_3_5]]
+; CHECK-NEXT:    [[SUB_1_6:%.*]] = sub i32 [[TMP30]], [[TMP22]]
+; CHECK-NEXT:    [[TMP92:%.*]] = icmp slt i32 [[SUB_1_6]], 0
+; CHECK-NEXT:    [[NEG_1_6:%.*]] = sub nsw i32 0, [[SUB_1_6]]
+; CHECK-NEXT:    [[TMP93:%.*]] = select i1 [[TMP92]], i32 [[NEG_1_6]], i32 [[SUB_1_6]]
+; CHECK-NEXT:    [[CMP12_1_6:%.*]] = icmp slt i32 [[TMP93]], [[SPEC_SELECT8_6]]
+; CHECK-NEXT:    [[TMP94:%.*]] = or i1 [[CMP12_1_6]], [[CMP12_6]]
+; CHECK-NEXT:    [[SPEC_SELECT8_1_6:%.*]] = select i1 [[CMP12_1_6]], i32 [[TMP93]], i32 [[SPEC_SELECT8_6]]
+; CHECK-NEXT:    [[SUB_2_6:%.*]] = sub i32 [[TMP30]], [[TMP23]]
+; CHECK-NEXT:    [[TMP95:%.*]] = icmp slt i32 [[SUB_2_6]], 0
+; CHECK-NEXT:    [[NEG_2_6:%.*]] = sub nsw i32 0, [[SUB_2_6]]
+; CHECK-NEXT:    [[TMP96:%.*]] = select i1 [[TMP95]], i32 [[NEG_2_6]], i32 [[SUB_2_6]]
+; CHECK-NEXT:    [[CMP12_2_6:%.*]] = icmp slt i32 [[TMP96]], [[SPEC_SELECT8_1_6]]
+; CHECK-NEXT:    [[TMP97:%.*]] = or i1 [[CMP12_2_6]], [[TMP94]]
+; CHECK-NEXT:    [[SPEC_SELECT8_2_6:%.*]] = select i1 [[CMP12_2_6]], i32 [[TMP96]], i32 [[SPEC_SELECT8_1_6]]
+; CHECK-NEXT:    [[SUB_3_6:%.*]] = sub i32 [[TMP30]], [[TMP24]]
+; CHECK-NEXT:    [[TMP98:%.*]] = icmp slt i32 [[SUB_3_6]], 0
+; CHECK-NEXT:    [[NEG_3_6:%.*]] = sub nsw i32 0, [[SUB_3_6]]
+; CHECK-NEXT:    [[TMP99:%.*]] = select i1 [[TMP98]], i32 [[NEG_3_6]], i32 [[SUB_3_6]]
+; CHECK-NEXT:    [[CMP12_3_6:%.*]] = icmp slt i32 [[TMP99]], [[SPEC_SELECT8_2_6]]
+; CHECK-NEXT:    [[TMP100:%.*]] = or i1 [[CMP12_3_6]], [[TMP97]]
+; CHECK-NEXT:    [[SPEC_SELECT_3_6:%.*]] = select i1 [[TMP100]], i32 6, i32 [[SPEC_SELECT_3_5]]
+; CHECK-NEXT:    [[SPEC_SELECT8_3_6:%.*]] = select i1 [[CMP12_3_6]], i32 [[TMP99]], i32 [[SPEC_SELECT8_2_6]]
+; CHECK-NEXT:    [[SUB_7:%.*]] = sub i32 [[TMP30]], [[TMP25]]
+; CHECK-NEXT:    [[TMP101:%.*]] = icmp slt i32 [[SUB_7]], 0
+; CHECK-NEXT:    [[NEG_7:%.*]] = sub nsw i32 0, [[SUB_7]]
+; CHECK-NEXT:    [[TMP102:%.*]] = select i1 [[TMP101]], i32 [[NEG_7]], i32 [[SUB_7]]
 ; CHECK-NEXT:    [[CMP12_7:%.*]] = icmp slt i32 [[TMP102]], [[SPEC_SELECT8_3_6]]
 ; CHECK-NEXT:    [[SPEC_SELECT8_7:%.*]] = select i1 [[CMP12_7]], i32 [[TMP102]], i32 [[SPEC_SELECT8_3_6]]
-; CHECK-NEXT:    [[TMP103:%.*]] = extractelement <16 x i32> [[TMP53]], i32 13
-; CHECK-NEXT:    [[NEG_1_7:%.*]] = sub nsw i32 0, [[TMP103]]
-; CHECK-NEXT:    [[TMP104:%.*]] = extractelement <16 x i1> [[TMP55]], i32 13
-; CHECK-NEXT:    [[TMP105:%.*]] = select i1 [[TMP104]], i32 [[NEG_1_7]], i32 [[TMP103]]
-; CHECK-NEXT:    [[CMP12_1_7:%.*]] = icmp slt i32 [[TMP105]], [[SPEC_SELECT8_7]]
-; CHECK-NEXT:    [[TMP106:%.*]] = or i1 [[CMP12_1_7]], [[CMP12_7]]
-; CHECK-NEXT:    [[SPEC_SELECT8_1_7:%.*]] = select i1 [[CMP12_1_7]], i32 [[TMP105]], i32 [[SPEC_SELECT8_7]]
-; CHECK-NEXT:    [[TMP107:%.*]] = extractelement <16 x i32> [[TMP53]], i32 14
-; CHECK-NEXT:    [[NEG_2_7:%.*]] = sub nsw i32 0, [[TMP107]]
-; CHECK-NEXT:    [[TMP108:%.*]] = extractelement <16 x i1> [[TMP55]], i32 14
-; CHECK-NEXT:    [[TMP109:%.*]] = select i1 [[TMP108]], i32 [[NEG_2_7]], i32 [[TMP107]]
-; CHECK-NEXT:    [[CMP12_2_7:%.*]] = icmp slt i32 [[TMP109]], [[SPEC_SELECT8_1_7]]
-; CHECK-NEXT:    [[TMP110:%.*]] = or i1 [[CMP12_2_7]], [[TMP106]]
-; CHECK-NEXT:    [[SPEC_SELECT8_2_7:%.*]] = select i1 [[CMP12_2_7]], i32 [[TMP109]], i32 [[SPEC_SELECT8_1_7]]
-; CHECK-NEXT:    [[TMP111:%.*]] = extractelement <16 x i32> [[TMP53]], i32 15
-; CHECK-NEXT:    [[NEG_3_7:%.*]] = sub nsw i32 0, [[TMP111]]
-; CHECK-NEXT:    [[TMP112:%.*]] = extractelement <16 x i1> [[TMP55]], i32 15
-; CHECK-NEXT:    [[TMP113:%.*]] = select i1 [[TMP112]], i32 [[NEG_3_7]], i32 [[TMP111]]
-; CHECK-NEXT:    [[CMP12_3_7:%.*]] = icmp slt i32 [[TMP113]], [[SPEC_SELECT8_2_7]]
-; CHECK-NEXT:    [[TMP114:%.*]] = or i1 [[CMP12_3_7]], [[TMP110]]
-; CHECK-NEXT:    [[SPEC_SELECT_3_7:%.*]] = select i1 [[TMP114]], i32 7, i32 [[SPEC_SELECT_3_6]]
-; CHECK-NEXT:    [[SPEC_SELECT8_3_7]] = select i1 [[CMP12_3_7]], i32 [[TMP113]], i32 [[SPEC_SELECT8_2_7]]
+; CHECK-NEXT:    [[SUB_1_7:%.*]] = sub i32 [[TMP30]], [[TMP26]]
+; CHECK-NEXT:    [[TMP103:%.*]] = icmp slt i32 [[SUB_1_7]], 0
+; CHECK-NEXT:    [[NEG_1_7:%.*]] = sub nsw i32 0, [[SUB_1_7]]
+; CHECK-NEXT:    [[TMP104:%.*]] = select i1 [[TMP103]], i32 [[NEG_1_7]], i32 [[SUB_1_7]]
+; CHECK-NEXT:    [[CMP12_1_7:%.*]] = icmp slt i32 [[TMP104]], [[SPEC_SELECT8_7]]
+; CHECK-NEXT:    [[TMP105:%.*]] = or i1 [[CMP12_1_7]], [[CMP12_7]]
+; CHECK-NEXT:    [[SPEC_SELECT8_1_7:%.*]] = select i1 [[CMP12_1_7]], i32 [[TMP104]], i32 [[SPEC_SELECT8_7]]
+; CHECK-NEXT:    [[SUB_2_7:%.*]] = sub i32 [[TMP30]], [[TMP27]]
+; CHECK-NEXT:    [[TMP106:%.*]] = icmp slt i32 [[SUB_2_7]], 0
+; CHECK-NEXT:    [[NEG_2_7:%.*]] = sub nsw i32 0, [[SUB_2_7]]
+; CHECK-NEXT:    [[TMP107:%.*]] = select i1 [[TMP106]], i32 [[NEG_2_7]], i32 [[SUB_2_7]]
+; CHECK-NEXT:    [[CMP12_2_7:%.*]] = icmp slt i32 [[TMP107]], [[SPEC_SELECT8_1_7]]
+; CHECK-NEXT:    [[TMP108:%.*]] = or i1 [[CMP12_2_7]], [[TMP105]]
+; CHECK-NEXT:    [[SPEC_SELECT8_2_7:%.*]] = select i1 [[CMP12_2_7]], i32 [[TMP107]], i32 [[SPEC_SELECT8_1_7]]
+; CHECK-NEXT:    [[SUB_3_7:%.*]] = sub i32 [[TMP30]], [[TMP28]]
+; CHECK-NEXT:    [[TMP109:%.*]] = icmp slt i32 [[SUB_3_7]], 0
+; CHECK-NEXT:    [[NEG_3_7:%.*]] = sub nsw i32 0, [[SUB_3_7]]
+; CHECK-NEXT:    [[TMP110:%.*]] = select i1 [[TMP109]], i32 [[NEG_3_7]], i32 [[SUB_3_7]]
+; CHECK-NEXT:    [[CMP12_3_7:%.*]] = icmp slt i32 [[TMP110]], [[SPEC_SELECT8_2_7]]
+; CHECK-NEXT:    [[TMP111:%.*]] = or i1 [[CMP12_3_7]], [[TMP108]]
+; CHECK-NEXT:    [[SPEC_SELECT_3_7:%.*]] = select i1 [[TMP111]], i32 7, i32 [[SPEC_SELECT_3_6]]
+; CHECK-NEXT:    [[SPEC_SELECT8_3_7]] = select i1 [[CMP12_3_7]], i32 [[TMP110]], i32 [[SPEC_SELECT8_2_7]]
 ; CHECK-NEXT:    [[K:%.*]] = getelementptr inbounds [366 x i32], ptr @l, i64 0, i64 [[INDVARS_IV]]
 ; CHECK-NEXT:    store i32 [[SPEC_SELECT_3_7]], ptr [[K]], align 4
 ; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1