[CostModel] Replace getUserCost with getInstructionCost.

[RKSimon]

Summary

• Replace getUserCost with getInstructionCost, covering all cost kinds.
• Remove getInstructionLatency, it's not implemented by any backends, and we should fold the functionality into getUserCost (now getInstructionCost) to make it easier for targets to handle the cost kinds with their existing cost callbacks.

Original Patch by @samparker (Sam Parker)

[samparker]

I'm beginning to change getInstructionThroughput so that it can just call into getUserCost.

• First is cast instructions: D79848.

[samparker]

Next is:

• control-flow ops: D79849.

[samparker]

Merge two of the intrinsic cost APIs: D79941

[samparker]

Unify intrinsic costs: D80012

[samparker]

Updated with final patch now that all the other pieces are in.

[samparker]

ping

[hfinkel]

Is getUserCost a better name than getInstructionCost? I wonder if we're moving in the wrong direction.

FWIW, when someone gets around to wanting to fix the vectorizer's interleaving heuristic, we'll need instruction latencies.

[samparker]

Is getUserCost a better name than getInstructionCost? I wonder if we're moving in the wrong direction.

I'd prefer to use getInstructionCost too.

[samparker]

Renamed getUserCost to getInstructionCost.

[RKSimon]

@samparker Are you intending to look at this again?

[samparker]

@RKSimon I'm not really doing much work on LLVM currently, so no.... But I'd be happy to help out with reviews if someone else wanted to pick it up.

[RKSimon]

Commandeering from @samparker - this has been on my backlog for ages, but you never know, I might get around to it soon...

[samparker]

cheers!

[RKSimon]

rebase - (WIP) still plenty of cost cleanup todo - it looks like some targets aren't correctly filtering costs by cost kind

[RKSimon]

looks like there's a codegen regression as well that needs addressing

[samparker]

Please shout when you're ready for review.

[RKSimon]

Cleaned up default fp arithmetic / select costs - still investigating the remaining regressions......

[RKSimon]

llvm/test/CodeGen/X86/select-optimize.ll

395 ↗

(On Diff #452438)

@apostolakis Please would take a look at this - we're about to start work on improving the cost numbers for latency/size and just this initial cleanup is causing this test to fail - which suggests you're relying on some very inaccurate costs.

[RKSimon]

rebase

[RKSimon]

rebase

[apostolakis]

llvm/test/CodeGen/X86/select-optimize.ll

395 ↗

(On Diff #452438)

I will remove this test. This particular test was a bit flaky to begin with given that it depended on the instruction latency of the instructions. At least it served the purpose of notifying me of the changes in the underlying cost modeling.

Just to clarify the use in the select-optimize pass. The goal was to approximate TargetSchedModel::computeInstrLatency and compute the length of dependence chains. getInstructionCost seemed the best approximation based on the documentation. getUserCost had an unclear purpose to me, did not account for latency queries (I guess this will change now) and takes the operands as input which is worrisome since I just wanted the cost of the instructions on their own regardless of their operands (in practice the operands are mostly ignored so that might not be an issue). The biggest change in cost modeling at least from switching from getInstructionCost to getUserCost seem to be the function calls which might be for the better.

[RKSimon]

llvm/test/CodeGen/X86/select-optimize.ll

395 ↗

(On Diff #452438)

Thanks - if you're wanting to perform this with IR then getUserCost (-> getInstructionCost) will be your best bet - but it does rely on us getting all the costs to be more accurate, which will take time.

I'm intending to fix the bitrot (we changed the way cost-model values are reported) on the script from D103695 which allows us to compare TTI costs vs various scheduler models (via llvm-mca), so eventually most instructions/intrinsics should have values similar to TargetSchedModel::computeInstrLatency. The script just helped check for throughput mismatches, but I did have the other cost kinds in mind as well.

I'd recommend that you do take operands into account where possible, as at least on X86 there are some attempts to use them to match likely codegen (e.g. sign/zero-extended integers that can use smaller arithmetic ops).

[apostolakis]

llvm/test/CodeGen/X86/select-optimize.ll

395 ↗

(On Diff #452438)

Sounds great! Thanks for improving this cost modeling. It is much needed.

For the operands, yes it seems that the purpose is to better predict how they will be lowered which indeed serves to improve cost modeling accuracy.

[apostolakis]

llvm/test/CodeGen/X86/select-optimize.ll

395 ↗

(On Diff #452438)

Test removed in D132029.

[RKSimon]

rebase - we can now make Instruction::Select latency cost override to be x86-only.

I think this is ready for general review now.

[dmgreen]

This is mostly expected to have minimal impact on the decisions the compiler makes, right? Most of the changes are in latency, which are not used very much in heuristics.

I ran some benchmarks and (after fixing what I got wrong the first time), it looks OK.

llvm/lib/Transforms/Scalar/LICM.cpp

1324–1325

getUserCost -> getInstructionCost

[RKSimon]

In D79483#3731290, @dmgreen wrote:

This is mostly expected to have minimal impact on the decisions the compiler makes, right? Most of the changes are in latency, which are not used very much in heuristics.

That's correct - the select-optimize pass is the only one that really tries to use it so far, and that's still disabled by default until the costs have been improved.

I ran some benchmarks and (after fixing what I got wrong the first time), it looks OK.

👍

[samparker]

llvm/include/llvm/Analysis/TargetTransformInfo.h

303–304

nit: three-argument

llvm/include/llvm/Analysis/TargetTransformInfoImpl.h

505

Would it be better, for now, to keep this in the X86 backend, if you need it there?

[RKSimon]

rebase

[RKSimon]

llvm/include/llvm/Analysis/TargetTransformInfoImpl.h

505

getInstructionLatency did have this as the default for all fp results

As well as x86 it causes one aarch64 sve fadd cost test to change from latency cost = 3 to 1 - not sure if thats enough of a reason to keep it generic or not?

[RKSimon]

Make the default fp latency = 3 to be x86 only

[samparker]

LGTM, and thanks for doing this.

llvm/include/llvm/Analysis/TargetTransformInfoImpl.h

505

getInstructionLatency did have this as the default for all fp results

Okay, fine then.

[RKSimon]

llvm/test/Analysis/CostModel/AArch64/sve-math.ll

15

@samparker Just to be clear - you're happy for me to make the latency=3 default x86-only?

[dmgreen]

llvm/test/Analysis/CostModel/AArch64/sve-math.ll

15

I think 3 sounds pretty sensible as a first approximation - it might not be precise but I've commonly seem fp operations in that ballpark. If was used for all targets before then I would keep as all targets now.

[RKSimon]

Default fp arithmmetic latency = 3

llvm/include/llvm/Analysis/TargetTransformInfo.h

public:
   /// target. The normalization of each cost model may be target specific.
   enum TargetCostKind {
     TCK_RecipThroughput, ///< Reciprocal throughput.
     TCK_Latency,         ///< The latency of instruction.
     TCK_CodeSize,        ///< Instruction code size.
     TCK_SizeAndLatency   ///< The weighted sum of size and latency.
   };
 
-  /// Query the cost of a specified instruction.
-  ///
-  /// Clients should use this interface to query the cost of an existing
-  /// instruction. The instruction must have a valid parent (basic block).
-  ///
-  /// Note, this method does not cache the cost calculation and it
-  /// can be expensive in some cases.
-  InstructionCost getInstructionCost(const Instruction *I,
-                                     enum TargetCostKind kind) const {
-    InstructionCost Cost;
-    switch (kind) {
-    case TCK_Latency:
-      Cost = getInstructionLatency(I);
-      break;
-    case TCK_RecipThroughput:
-    case TCK_CodeSize:
-    case TCK_SizeAndLatency:
-      Cost = getUserCost(I, kind);
-      break;
-    }
-    return Cost;
-  }
-
   /// Underlying constants for 'cost' values in this interface.
   ///
   /// Many APIs in this interface return a cost. This enum defines the
   /// fundamental values that should be used to interpret (and produce) those
   /// costs. The costs are returned as an int rather than a member of this
   /// enumeration because it is expected that the cost of one IR instruction
   /// may have a multiplicative factor to it or otherwise won't fit directly
   /// into the enum. Moreover, it is common to sum or average costs which works
public:
   /// \p Operands is a list of operands which can be a result of transformations
   /// of the current operands. The number of the operands on the list must equal
   /// to the number of the current operands the IR user has. Their order on the
   /// list must be the same as the order of the current operands the IR user
   /// has.
   ///
   /// The returned cost is defined in terms of \c TargetCostConstants, see its
   /// comments for a detailed explanation of the cost values.
-  InstructionCost getUserCost(const User *U, ArrayRef<const Value *> Operands,
+  InstructionCost getInstructionCost(const User *U,
+                                     ArrayRef<const Value *> Operands,
                                      TargetCostKind CostKind) const;
 
-  /// This is a helper function which calls the two-argument getUserCost
-  /// with \p Operands which are the current operands U has.
-  InstructionCost getUserCost(const User *U, TargetCostKind CostKind) const {
+  /// This is a helper function which calls the three-argument
+  /// getInstructionCost with \p Operands which are the current operands U has.
+  InstructionCost getInstructionCost(const User *U,
+                                     TargetCostKind CostKind) const {
     SmallVector<const Value *, 4> Operands(U->operand_values());
-    return getUserCost(U, Operands, CostKind);
+    return getInstructionCost(U, Operands, CostKind);
   }
 
   /// If a branch or a select condition is skewed in one direction by more than
   /// this factor, it is very likely to be predicted correctly.
   BranchProbability getPredictableBranchThreshold() const;
 
   /// Return true if branch divergence exists.
   ///
struct LSRCost {
     unsigned ImmCost;
     unsigned SetupCost;
     unsigned ScaleCost;
   };
 
   /// Parameters that control the generic loop unrolling transformation.
   struct UnrollingPreferences {
     /// The cost threshold for the unrolled loop. Should be relative to the
-    /// getUserCost values returned by this API, and the expectation is that
-    /// the unrolled loop's instructions when run through that interface should
-    /// not exceed this cost. However, this is only an estimate. Also, specific
-    /// loops may be unrolled even with a cost above this threshold if deemed
-    /// profitable. Set this to UINT_MAX to disable the loop body cost
+    /// getInstructionCost values returned by this API, and the expectation is
+    /// that the unrolled loop's instructions when run through that interface
+    /// should not exceed this cost. However, this is only an estimate. Also,
+    /// specific loops may be unrolled even with a cost above this threshold if
+    /// deemed profitable. Set this to UINT_MAX to disable the loop body cost
     /// restriction.
     unsigned Threshold;
     /// If complete unrolling will reduce the cost of the loop, we will boost
     /// the Threshold by a certain percent to allow more aggressive complete
     /// unrolling. This value provides the maximum boost percentage that we
     /// can apply to Threshold (The value should be no less than 100).
     /// BoostedThreshold = Threshold * min(RolledCost / UnrolledCost,
     ///                                    MaxPercentThresholdBoost / 100)

   /// \returns How the target needs this vector-predicated operation to be
   /// transformed.
   VPLegalization getVPLegalizationStrategy(const VPIntrinsic &PI) const;
   /// @}
 
   /// @}
 
 private:
-  /// Estimate the latency of specified instruction.
-  /// Returns 1 as the default value.
-  InstructionCost getInstructionLatency(const Instruction *I) const;
-
   /// The abstract base class used to type erase specific TTI
   /// implementations.
   class Concept;
 
   /// The template model for the base class which wraps a concrete
   /// implementation in a type erased interface.
   template <typename T> class Model;
 
public:
   virtual unsigned getInliningThresholdMultiplier() = 0;
   virtual unsigned adjustInliningThreshold(const CallBase *CB) = 0;
   virtual int getInlinerVectorBonusPercent() = 0;
   virtual InstructionCost getMemcpyCost(const Instruction *I) = 0;
   virtual unsigned
   getEstimatedNumberOfCaseClusters(const SwitchInst &SI, unsigned &JTSize,
                                    ProfileSummaryInfo *PSI,
                                    BlockFrequencyInfo *BFI) = 0;
-  virtual InstructionCost getUserCost(const User *U,
+  virtual InstructionCost getInstructionCost(const User *U,
                                              ArrayRef<const Value *> Operands,
                                              TargetCostKind CostKind) = 0;
   virtual BranchProbability getPredictableBranchThreshold() = 0;
   virtual bool hasBranchDivergence() = 0;
   virtual bool useGPUDivergenceAnalysis() = 0;
   virtual bool isSourceOfDivergence(const Value *V) = 0;
   virtual bool isAlwaysUniform(const Value *V) = 0;
   virtual unsigned getFlatAddressSpace() = 0;
   virtual bool collectFlatAddressOperands(SmallVectorImpl<int> &OpIndexes,
                                           Intrinsic::ID IID) const = 0;
virtual bool preferPredicatedReductionSelect(unsigned Opcode, Type *Ty,
                                                ReductionFlags) const = 0;
   virtual bool shouldExpandReduction(const IntrinsicInst *II) const = 0;
   virtual unsigned getGISelRematGlobalCost() const = 0;
   virtual unsigned getMinTripCountTailFoldingThreshold() const = 0;
   virtual bool enableScalableVectorization() const = 0;
   virtual bool supportsScalableVectors() const = 0;
   virtual bool hasActiveVectorLength(unsigned Opcode, Type *DataType,
                                      Align Alignment) const = 0;
-  virtual InstructionCost getInstructionLatency(const Instruction *I) = 0;
   virtual VPLegalization
   getVPLegalizationStrategy(const VPIntrinsic &PI) const = 0;
 };
 
 template <typename T>
 class TargetTransformInfo::Model final : public TargetTransformInfo::Concept {
   T Impl;
 
unsigned adjustInliningThreshold(const CallBase *CB) override {
     return Impl.adjustInliningThreshold(CB);
   }
   int getInlinerVectorBonusPercent() override {
     return Impl.getInlinerVectorBonusPercent();
   }
   InstructionCost getMemcpyCost(const Instruction *I) override {
     return Impl.getMemcpyCost(I);
   }
-  InstructionCost getUserCost(const User *U, ArrayRef<const Value *> Operands,
+  InstructionCost getInstructionCost(const User *U,
+                                     ArrayRef<const Value *> Operands,
                                      TargetCostKind CostKind) override {
-    return Impl.getUserCost(U, Operands, CostKind);
+    return Impl.getInstructionCost(U, Operands, CostKind);
   }
   BranchProbability getPredictableBranchThreshold() override {
     return Impl.getPredictableBranchThreshold();
   }
   bool hasBranchDivergence() override { return Impl.hasBranchDivergence(); }
   bool useGPUDivergenceAnalysis() override {
     return Impl.useGPUDivergenceAnalysis();
   }
bool enableScalableVectorization() const override {
     return Impl.enableScalableVectorization();
   }
 
   bool hasActiveVectorLength(unsigned Opcode, Type *DataType,
                              Align Alignment) const override {
     return Impl.hasActiveVectorLength(Opcode, DataType, Alignment);
   }
 
-  InstructionCost getInstructionLatency(const Instruction *I) override {
-    return Impl.getInstructionLatency(I);
-  }
-
   VPLegalization
   getVPLegalizationStrategy(const VPIntrinsic &PI) const override {
     return Impl.getVPLegalizationStrategy(PI);
   }
 };
 
 template <typename T>
 TargetTransformInfo::TargetTransformInfo(T Impl)

llvm/include/llvm/Analysis/TargetTransformInfoImpl.h

InstructionCost getArithmeticInstrCost(
     case Instruction::FRem:
     case Instruction::SDiv:
     case Instruction::SRem:
     case Instruction::UDiv:
     case Instruction::URem:
       // FIXME: Unlikely to be true for CodeSize.
       return TTI::TCC_Expensive;
     }
+
+    // Assume a 3cy latency for fp arithmetic ops.
+    if (CostKind == TTI::TCK_Latency)
+      if (Ty->getScalarType()->isFloatingPointTy())
+        return 3;
+
     return 1;
   }
 
   InstructionCost getShuffleCost(TTI::ShuffleKind Kind, VectorType *Ty,
                                  ArrayRef<int> Mask, int Index,
                                  VectorType *SubTp,
                                  ArrayRef<const Value *> Args = None) const {
     return 1;
InstructionCost getGEPCost(Type *PointeeType, const Value *Ptr,
     if (static_cast<T *>(this)->isLegalAddressingMode(
             TargetType, const_cast<GlobalValue *>(BaseGV),
             BaseOffset.sextOrTrunc(64).getSExtValue(), HasBaseReg, Scale,
             Ptr->getType()->getPointerAddressSpace()))
       return TTI::TCC_Free;
     return TTI::TCC_Basic;
   }
 
-  InstructionCost getUserCost(const User *U, ArrayRef<const Value *> Operands,
+  InstructionCost getInstructionCost(const User *U,
+                                     ArrayRef<const Value *> Operands,
                                      TTI::TargetCostKind CostKind) {
     using namespace llvm::PatternMatch;
 
     auto *TargetTTI = static_cast<T *>(this);
     // Handle non-intrinsic calls, invokes, and callbr.
     // FIXME: Unlikely to be true for anything but CodeSize.
     auto *CB = dyn_cast<CallBase>(U);
     if (CB && !isa<IntrinsicInst>(U)) {
       if (const Function *F = CB->getCalledFunction()) {
InstructionCost getInstructionCost(const User *U,
     case Instruction::Store: {
       auto *SI = cast<StoreInst>(U);
       Type *ValTy = U->getOperand(0)->getType();
       return TargetTTI->getMemoryOpCost(Opcode, ValTy, SI->getAlign(),
                                         SI->getPointerAddressSpace(),
                                         CostKind, I);
     }
     case Instruction::Load: {
+      // FIXME: Arbitary cost which could come from the backend.
+      if (CostKind == TTI::TCK_Latency)
+        return 4;
       auto *LI = cast<LoadInst>(U);
       Type *LoadType = U->getType();
       // If there is a non-register sized type, the cost estimation may expand
       // it to be several instructions to load into multiple registers on the
       // target.  But, if the only use of the load is a trunc instruction to a
       // register sized type, the instruction selector can combine these
       // instructions to be a single load.  So, in this case, we use the
       // destination type of the trunc instruction rather than the load to
case Instruction::ExtractElement: {
       unsigned Idx = -1;
       if (auto *CI = dyn_cast<ConstantInt>(EEI->getOperand(1)))
         if (CI->getValue().getActiveBits() <= 32)
           Idx = CI->getZExtValue();
       Type *DstTy = U->getOperand(0)->getType();
       return TargetTTI->getVectorInstrCost(*EEI, DstTy, Idx);
     }
     }
-    // By default, just classify everything as 'basic'.
-    return TTI::TCC_Basic;
-  }
 
-  InstructionCost getInstructionLatency(const Instruction *I) {
-    SmallVector<const Value *, 4> Operands(I->operand_values());
-    if (getUserCost(I, Operands, TTI::TCK_Latency) == TTI::TCC_Free)
-      return 0;
-
-    if (isa<LoadInst>(I))
-      return 4;
-
-    Type *DstTy = I->getType();
-
-    // Usually an intrinsic is a simple instruction.
-    // A real function call is much slower.
-    if (auto *CI = dyn_cast<CallInst>(I)) {
-      const Function *F = CI->getCalledFunction();
-      if (!F || static_cast<T *>(this)->isLoweredToCall(F))
-        return 40;
-      // Some intrinsics return a value and a flag, we use the value type
-      // to decide its latency.
-      if (StructType *StructTy = dyn_cast<StructType>(DstTy))
-        DstTy = StructTy->getElementType(0);
-      // Fall through to simple instructions.
-    }
-
-    if (VectorType *VectorTy = dyn_cast<VectorType>(DstTy))
-      DstTy = VectorTy->getElementType();
-    if (DstTy->isFloatingPointTy())
-      return 3;
-
-    return 1;
+    // By default, just classify everything as 'basic' or -1 to represent that
+    // don't know the throughput cost.
+    return CostKind == TTI::TCK_RecipThroughput ? -1 : TTI::TCC_Basic;
   }
 };
 } // namespace llvm
 
 #endif

llvm/include/llvm/CodeGen/BasicTTIImpl.h

Optional<Value *> simplifyDemandedVectorEltsIntrinsic(
       APInt &UndefElts2, APInt &UndefElts3,
       std::function<void(Instruction *, unsigned, APInt, APInt &)>
           SimplifyAndSetOp) {
     return BaseT::simplifyDemandedVectorEltsIntrinsic(
         IC, II, DemandedElts, UndefElts, UndefElts2, UndefElts3,
         SimplifyAndSetOp);
   }
 
-  InstructionCost getInstructionLatency(const Instruction *I) {
-    if (isa<LoadInst>(I))
-      return getST()->getSchedModel().DefaultLoadLatency;
-
-    return BaseT::getInstructionLatency(I);
-  }
-
   virtual Optional<unsigned>
   getCacheSize(TargetTransformInfo::CacheLevel Level) const {
     return Optional<unsigned>(
       getST()->getCacheSize(static_cast<unsigned>(Level)));
   }
 
   virtual Optional<unsigned>
   getCacheAssociativity(TargetTransformInfo::CacheLevel Level) const {

llvm/lib/Analysis/CodeMetrics.cpp

if (const CallInst *CI = dyn_cast<CallInst>(&I)) {
       if (CI->isConvergent())
         convergent = true;
     }
 
     if (const InvokeInst *InvI = dyn_cast<InvokeInst>(&I))
       if (InvI->cannotDuplicate())
         notDuplicatable = true;
 
-    NumInsts += TTI.getUserCost(&I, TargetTransformInfo::TCK_CodeSize);
+    NumInsts += TTI.getInstructionCost(&I, TargetTransformInfo::TCK_CodeSize);
   }
 
   if (isa<ReturnInst>(BB->getTerminator()))
     ++NumRets;
 
   // We never want to inline functions that contain an indirectbr.  This is
   // incorrect because all the blockaddress's (in static global initializers
   // for example) would be referring to the original function, and this indirect

llvm/lib/Analysis/InlineCost.cpp

 bool CallAnalyzer::isGEPFree(GetElementPtrInst &GEP) {
   SmallVector<Value *, 4> Operands;
   Operands.push_back(GEP.getOperand(0));
   for (const Use &Op : GEP.indices())
     if (Constant *SimpleOp = SimplifiedValues.lookup(Op))
       Operands.push_back(SimpleOp);
     else
       Operands.push_back(Op);
-  return TTI.getUserCost(&GEP, Operands,
+  return TTI.getInstructionCost(&GEP, Operands,
                                 TargetTransformInfo::TCK_SizeAndLatency) ==
          TargetTransformInfo::TCC_Free;
 }
 
 bool CallAnalyzer::visitAlloca(AllocaInst &I) {
   disableSROA(I.getOperand(0));
 
   // Check whether inlining will turn a dynamic alloca into a static
   // alloca and handle that case.
bool CallAnalyzer::visitPtrToInt(PtrToIntInst &I) {
   // inlining, it will be nuked, and SROA should proceed. All of the uses which
   // would block SROA would also block SROA if applied directly to a pointer,
   // and so we can just add the integer in here. The only places where SROA is
   // preserved either cannot fire on an integer, or won't in-and-of themselves
   // disable SROA (ext) w/o some later use that we would see and disable.
   if (auto *SROAArg = getSROAArgForValueOrNull(I.getOperand(0)))
     SROAArgValues[&I] = SROAArg;
 
-  return TTI.getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency) ==
+  return TTI.getInstructionCost(&I, TargetTransformInfo::TCK_SizeAndLatency) ==
          TargetTransformInfo::TCC_Free;
 }
 
 bool CallAnalyzer::visitIntToPtr(IntToPtrInst &I) {
   // Propagate constants through ptrtoint.
   if (simplifyInstruction(I))
     return true;
 
   // Track base/offset pairs when round-tripped through a pointer without
   // modifications provided the integer is not too large.
   Value *Op = I.getOperand(0);
   unsigned IntegerSize = Op->getType()->getScalarSizeInBits();
   if (IntegerSize <= DL.getPointerTypeSizeInBits(I.getType())) {
     std::pair<Value *, APInt> BaseAndOffset = ConstantOffsetPtrs.lookup(Op);
     if (BaseAndOffset.first)
       ConstantOffsetPtrs[&I] = BaseAndOffset;
   }
 
   // "Propagate" SROA here in the same manner as we do for ptrtoint above.
   if (auto *SROAArg = getSROAArgForValueOrNull(Op))
     SROAArgValues[&I] = SROAArg;
 
-  return TTI.getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency) ==
+  return TTI.getInstructionCost(&I, TargetTransformInfo::TCK_SizeAndLatency) ==
          TargetTransformInfo::TCC_Free;
 }
 
 bool CallAnalyzer::visitCastInst(CastInst &I) {
   // Propagate constants through casts.
   if (simplifyInstruction(I))
     return true;
 
bool CallAnalyzer::visitCastInst(CastInst &I) {
   case Instruction::FPToSI:
     if (TTI.getFPOpCost(I.getType()) == TargetTransformInfo::TCC_Expensive)
       onCallPenalty();
     break;
   default:
     break;
   }
 
-  return TTI.getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency) ==
+  return TTI.getInstructionCost(&I, TargetTransformInfo::TCK_SizeAndLatency) ==
          TargetTransformInfo::TCC_Free;
 }
 
 bool CallAnalyzer::paramHasAttr(Argument *A, Attribute::AttrKind Attr) {
   return CandidateCall.paramHasAttr(A->getArgNo(), Attr);
 }
 
 bool CallAnalyzer::isKnownNonNullInCallee(Value *V) {
bool CallAnalyzer::visitUnreachableInst(UnreachableInst &I) {
   // to unreachable as they have the lowest possible impact on both runtime and
   // code size.
   return true; // No actual code is needed for unreachable.
 }
 
 bool CallAnalyzer::visitInstruction(Instruction &I) {
   // Some instructions are free. All of the free intrinsics can also be
   // handled by SROA, etc.
-  if (TTI.getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency) ==
+  if (TTI.getInstructionCost(&I, TargetTransformInfo::TCK_SizeAndLatency) ==
       TargetTransformInfo::TCC_Free)
     return true;
 
   // We found something we don't understand or can't handle. Mark any SROA-able
   // values in the operand list as no longer viable.
   for (const Use &Op : I.operands())
     disableSROA(Op);
 

llvm/lib/Analysis/TargetTransformInfo.cpp

 
 unsigned TargetTransformInfo::getEstimatedNumberOfCaseClusters(
     const SwitchInst &SI, unsigned &JTSize, ProfileSummaryInfo *PSI,
     BlockFrequencyInfo *BFI) const {
   return TTIImpl->getEstimatedNumberOfCaseClusters(SI, JTSize, PSI, BFI);
 }
 
 InstructionCost
-TargetTransformInfo::getUserCost(const User *U,
+TargetTransformInfo::getInstructionCost(const User *U,
                                         ArrayRef<const Value *> Operands,
                                         enum TargetCostKind CostKind) const {
-  InstructionCost Cost = TTIImpl->getUserCost(U, Operands, CostKind);
+  InstructionCost Cost = TTIImpl->getInstructionCost(U, Operands, CostKind);
   assert((CostKind == TTI::TCK_RecipThroughput || Cost >= 0) &&
          "TTI should not produce negative costs!");
   return Cost;
 }
 
 BranchProbability TargetTransformInfo::getPredictableBranchThreshold() const {
   return TTIImpl->getPredictableBranchThreshold();
 }
bool TargetTransformInfo::enableScalableVectorization() const {
   return TTIImpl->enableScalableVectorization();
 }
 
 bool TargetTransformInfo::hasActiveVectorLength(unsigned Opcode, Type *DataType,
                                                 Align Alignment) const {
   return TTIImpl->hasActiveVectorLength(Opcode, DataType, Alignment);
 }
 
-InstructionCost
-TargetTransformInfo::getInstructionLatency(const Instruction *I) const {
-  return TTIImpl->getInstructionLatency(I);
-}
-
 TargetTransformInfo::Concept::~Concept() = default;
 
 TargetIRAnalysis::TargetIRAnalysis() : TTICallback(&getDefaultTTI) {}
 
 TargetIRAnalysis::TargetIRAnalysis(
     std::function<Result(const Function &)> TTICallback)
     : TTICallback(std::move(TTICallback)) {}
 

llvm/lib/CodeGen/CodeGenPrepare.cpp

 
 /// Check if V (an operand of a select instruction) is an expensive instruction
 /// that is only used once.
 static bool sinkSelectOperand(const TargetTransformInfo *TTI, Value *V) {
   auto *I = dyn_cast<Instruction>(V);
   // If it's safe to speculatively execute, then it should not have side
   // effects; therefore, it's safe to sink and possibly *not* execute.
   return I && I->hasOneUse() && isSafeToSpeculativelyExecute(I) &&
-         TTI->getUserCost(I, TargetTransformInfo::TCK_SizeAndLatency) >=
+         TTI->getInstructionCost(I, TargetTransformInfo::TCK_SizeAndLatency) >=
              TargetTransformInfo::TCC_Expensive;
 }
 
 /// Returns true if a SelectInst should be turned into an explicit branch.
 static bool isFormingBranchFromSelectProfitable(const TargetTransformInfo *TTI,
                                                 const TargetLowering *TLI,
                                                 SelectInst *SI) {
   // If even a predictable select is cheap, then a branch can't be cheaper.
   if (!TLI->isPredictableSelectExpensive())

llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp

for (auto &I : *BB) {
         if (const Function *F = cast<CallBase>(I).getCalledFunction()) {
           if (!isLoweredToCall(F))
             continue;
         }
         return;
       }
 
       SmallVector<const Value*, 4> Operands(I.operand_values());
-      Cost +=
-        getUserCost(&I, Operands, TargetTransformInfo::TCK_SizeAndLatency);
+      Cost += getInstructionCost(&I, Operands,
+                                 TargetTransformInfo::TCK_SizeAndLatency);
     }
   }
 
   // On v6m cores, there are very few registers available. We can easily end up
   // spilling and reloading more registers in an unrolled loop. Look at the
   // number of LCSSA phis as a rough measure of how many registers will need to
   // be live out of the loop, reducing the default unroll count if more than 1
   // value is needed.  In the long run, all of this should be being learnt by a

llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.h

InstructionCost getCFInstrCost(unsigned Opcode, TTI::TargetCostKind CostKind,
     return 1;
   }
 
   bool isLegalMaskedStore(Type *DataType, Align Alignment);
   bool isLegalMaskedLoad(Type *DataType, Align Alignment);
 
   /// @}
 
-  InstructionCost getUserCost(const User *U, ArrayRef<const Value *> Operands,
+  InstructionCost getInstructionCost(const User *U,
+                                     ArrayRef<const Value *> Operands,
                                      TTI::TargetCostKind CostKind);
 
   // Hexagon specific decision to generate a lookup table.
   bool shouldBuildLookupTables() const;
 };
 
 } // end namespace llvm
 #endif // LLVM_LIB_TARGET_HEXAGON_HEXAGONTARGETTRANSFORMINFO_H

llvm/lib/Target/Hexagon/HexagonTargetTransformInfo.cpp

 unsigned HexagonTTIImpl::getPrefetchDistance() const {
   return ST.getL1PrefetchDistance();
 }
 
 unsigned HexagonTTIImpl::getCacheLineSize() const {
   return ST.getL1CacheLineSize();
 }
 
-InstructionCost HexagonTTIImpl::getUserCost(const User *U,
-                                            ArrayRef<const Value *> Operands,
-                                            TTI::TargetCostKind CostKind) {
+InstructionCost
+HexagonTTIImpl::getInstructionCost(const User *U,
   auto isCastFoldedIntoLoad = [this](const CastInst *CI) -> bool {
     if (!CI->isIntegerCast())
       return false;
     // Only extensions from an integer type shorter than 32-bit to i32
     // can be folded into the load.
     const DataLayout &DL = getDataLayout();
     unsigned SBW = DL.getTypeSizeInBits(CI->getSrcTy());
     unsigned DBW = DL.getTypeSizeInBits(CI->getDestTy());
     if (DBW != 32 || SBW >= DBW)
       return false;
 
     const LoadInst *LI = dyn_cast<const LoadInst>(CI->getOperand(0));
     // Technically, this code could allow multiple uses of the load, and
     // check if all the uses are the same extension operation, but this
     // should be sufficient for most cases.
     return LI && LI->hasOneUse();
   };
 
   if (const CastInst *CI = dyn_cast<const CastInst>(U))
     if (isCastFoldedIntoLoad(CI))
       return TargetTransformInfo::TCC_Free;
-  return BaseT::getUserCost(U, Operands, CostKind);
+  return BaseT::getInstructionCost(U, Operands, CostKind);
 }
 
 bool HexagonTTIImpl::shouldBuildLookupTables() const {
   return EmitLookupTables;
 }

llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h

public:
   InstructionCost getIntImmCostInst(unsigned Opcode, unsigned Idx,
                                     const APInt &Imm, Type *Ty,
                                     TTI::TargetCostKind CostKind,
                                     Instruction *Inst = nullptr);
   InstructionCost getIntImmCostIntrin(Intrinsic::ID IID, unsigned Idx,
                                       const APInt &Imm, Type *Ty,
                                       TTI::TargetCostKind CostKind);
 
-  InstructionCost getUserCost(const User *U, ArrayRef<const Value *> Operands,
+  InstructionCost getInstructionCost(const User *U,
+                                     ArrayRef<const Value *> Operands,
                                      TTI::TargetCostKind CostKind);
 
   TTI::PopcntSupportKind getPopcntSupport(unsigned TyWidth);
   bool isHardwareLoopProfitable(Loop *L, ScalarEvolution &SE,
                                 AssumptionCache &AC,
                                 TargetLibraryInfo *LibInfo,
                                 HardwareLoopInfo &HWLoopInfo);
   bool canSaveCmp(Loop *L, BranchInst **BI, ScalarEvolution *SE, LoopInfo *LI,
                   DominatorTree *DT, AssumptionCache *AC,

llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp

 
 // Check if the current Type is an MMA vector type. Valid MMA types are
 // v256i1 and v512i1 respectively.
 static bool isMMAType(Type *Ty) {
   return Ty->isVectorTy() && (Ty->getScalarSizeInBits() == 1) &&
          (Ty->getPrimitiveSizeInBits() > 128);
 }
 
-InstructionCost PPCTTIImpl::getUserCost(const User *U,
+InstructionCost PPCTTIImpl::getInstructionCost(const User *U,
                                                ArrayRef<const Value *> Operands,
                                                TTI::TargetCostKind CostKind) {
   // We already implement getCastInstrCost and getMemoryOpCost where we perform
   // the vector adjustment there.
   if (isa<CastInst>(U) || isa<LoadInst>(U) || isa<StoreInst>(U))
-    return BaseT::getUserCost(U, Operands, CostKind);
+    return BaseT::getInstructionCost(U, Operands, CostKind);
 
   if (U->getType()->isVectorTy()) {
     // Instructions that need to be split should cost more.
     std::pair<InstructionCost, MVT> LT = getTypeLegalizationCost(U->getType());
-    return LT.first * BaseT::getUserCost(U, Operands, CostKind);
+    return LT.first * BaseT::getInstructionCost(U, Operands, CostKind);
   }
 
-  return BaseT::getUserCost(U, Operands, CostKind);
+  return BaseT::getInstructionCost(U, Operands, CostKind);
 }
 
 // Determining the address of a TLS variable results in a function call in
 // certain TLS models.
 static bool memAddrUsesCTR(const Value *MemAddr, const PPCTargetMachine &TM,
                            SmallPtrSetImpl<const Value *> &Visited) {
   // No need to traverse again if we already checked this operand.
   if (!Visited.insert(MemAddr).second)

llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp

for (auto &I : *BB) {
         if (const Function *F = cast<CallBase>(I).getCalledFunction()) {
           if (!isLoweredToCall(F))
             continue;
         }
         return;
       }
 
       SmallVector<const Value *> Operands(I.operand_values());
-      Cost +=
-          getUserCost(&I, Operands, TargetTransformInfo::TCK_SizeAndLatency);
+      Cost += getInstructionCost(&I, Operands,
+                                 TargetTransformInfo::TCK_SizeAndLatency);
     }
   }
 
   LLVM_DEBUG(dbgs() << "Cost of loop: " << Cost << "\n");
 
   UP.Partial = true;
   UP.Runtime = true;
   UP.UnrollRemainder = true;

llvm/lib/Target/X86/X86TargetTransformInfo.cpp

return AdjustCost(
       BaseT::getCastInstrCost(Opcode, Dst, Src, CCH, CostKind, I));
 }
 
 InstructionCost X86TTIImpl::getCmpSelInstrCost(unsigned Opcode, Type *ValTy,
                                                Type *CondTy,
                                                CmpInst::Predicate VecPred,
                                                TTI::TargetCostKind CostKind,
                                                const Instruction *I) {
+  // Assume a 3cy latency for fp select ops.
+  if (CostKind == TTI::TCK_Latency && Opcode == Instruction::Select)
+    if (ValTy->getScalarType()->isFloatingPointTy())
+      return 3;
+
   // TODO: Handle other cost kinds.
   if (CostKind != TTI::TCK_RecipThroughput)
     return BaseT::getCmpSelInstrCost(Opcode, ValTy, CondTy, VecPred, CostKind,
                                      I);
 
   // Legalize the type.
   std::pair<InstructionCost, MVT> LT = getTypeLegalizationCost(ValTy);
 

llvm/lib/Transforms/IPO/FunctionSpecialization.cpp

InstructionCost getUserBonus(User *U, llvm::TargetTransformInfo &TTI,
                                LoopInfo &LI) {
     auto *I = dyn_cast_or_null<Instruction>(U);
     // If not an instruction we do not know how to evaluate.
     // Keep minimum possible cost for now so that it doesnt affect
     // specialization.
     if (!I)
       return std::numeric_limits<unsigned>::min();
 
-    auto Cost = TTI.getUserCost(U, TargetTransformInfo::TCK_SizeAndLatency);
+    InstructionCost Cost =
+        TTI.getInstructionCost(U, TargetTransformInfo::TCK_SizeAndLatency);
 
     // Traverse recursively if there are more uses.
     // TODO: Any other instructions to be added here?
     if (I->mayReadFromMemory() || I->isCast())
       for (auto *User : I->users())
         Cost += getUserBonus(User, TTI, LI);
 
     // Increase the cost if it is inside the loop.

llvm/lib/Transforms/Scalar/JumpThreading.cpp

if (I->getType()->isTokenTy() && I->isUsedOutsideOfBlock(BB))
       return ~0U;
 
     // Blocks with NoDuplicate are modelled as having infinite cost, so they
     // are never duplicated.
     if (const CallInst *CI = dyn_cast<CallInst>(I))
       if (CI->cannotDuplicate() || CI->isConvergent())
         return ~0U;
 
-    if (TTI->getUserCost(&*I, TargetTransformInfo::TCK_SizeAndLatency)
-            == TargetTransformInfo::TCC_Free)
+    if (TTI->getInstructionCost(&*I, TargetTransformInfo::TCK_SizeAndLatency) ==
+        TargetTransformInfo::TCC_Free)
       continue;
 
     // All other instructions count for at least one unit.
     ++Size;
 
     // Calls are more expensive.  If they are non-intrinsic calls, we model them
     // as having cost of 4.  If they are a non-vector intrinsic, we model them
     // as having cost of 2 total, and if they are a vector intrinsic, we model

llvm/lib/Transforms/Scalar/LICM.cpp

static bool isTriviallyReplaceablePHI(const PHINode &PN, const Instruction &I) {
 
   return true;
 }
 
 /// Return true if the instruction is free in the loop.
 static bool isFreeInLoop(const Instruction &I, const Loop *CurLoop,
                          const TargetTransformInfo *TTI) {
   InstructionCost CostI =
-      TTI->getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency);
+      TTI->getInstructionCost(&I, TargetTransformInfo::TCK_SizeAndLatency);
 
   if (auto *GEP = dyn_cast<GetElementPtrInst>(&I)) {
     if (CostI != TargetTransformInfo::TCC_Free)
       return false;
-    // For a GEP, we cannot simply use getUserCost because currently it
-    // optimistically assumes that a GEP will fold into addressing mode
+    // For a GEP, we cannot simply use getInstructionCost because currently
+    // it optimistically assumes that a GEP will fold into addressing mode
     // regardless of its users.
     const BasicBlock *BB = GEP->getParent();
     for (const User *U : GEP->users()) {
       const Instruction *UI = cast<Instruction>(U);
       if (CurLoop->contains(UI) &&
           (BB != UI->getParent() ||
            (!isa<StoreInst>(UI) && !isa<LoadInst>(UI))))
         return false;

llvm/lib/Transforms/Scalar/LoopFlatten.cpp

for (auto &I : *B) {
           Br->getSuccessor(0) == FI.InnerLoop->getHeader())
         continue;
       // Multiplies of the outer iteration variable and inner iteration
       // count will be optimised out.
       if (match(&I, m_c_Mul(m_Specific(FI.OuterInductionPHI),
                             m_Specific(FI.InnerTripCount))))
         continue;
       InstructionCost Cost =
-          TTI->getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency);
+          TTI->getInstructionCost(&I, TargetTransformInfo::TCK_SizeAndLatency);
       LLVM_DEBUG(dbgs() << "Cost " << Cost << ": "; I.dump());
       RepeatedInstrCost += Cost;
     }
   }
 
   LLVM_DEBUG(dbgs() << "Cost of instructions that will be repeated: "
                     << RepeatedInstrCost << "\n");
   // Bail out if flattening the loops would cause instructions in the outer

llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp

for (;; --Iteration) {
                     PhiI->getIncomingValueForBlock(L->getLoopLatch())))
               if (L->contains(OpI))
                 PHIUsedList.push_back(OpI);
             continue;
           }
 
         // First accumulate the cost of this instruction.
         if (!Cost.IsFree) {
-          UnrolledCost += TTI.getUserCost(I, CostKind);
+          UnrolledCost += TTI.getInstructionCost(I, CostKind);
           LLVM_DEBUG(dbgs() << "Adding cost of instruction (iteration "
                             << Iteration << "): ");
           LLVM_DEBUG(I->dump());
         }
 
         // We must count the cost of every operand which is not free,
         // recursively. If we reach a loop PHI node, simply add it to the set
         // to be considered on the next iteration (backwards!).
for (unsigned Idx = 0; Idx != BBWorklist.size(); ++Idx) {
       for (Instruction &I : *BB) {
         // These won't get into the final code - don't even try calculating the
         // cost for them.
         if (isa<DbgInfoIntrinsic>(I) || EphValues.count(&I))
           continue;
 
         // Track this instruction's expected baseline cost when executing the
         // rolled loop form.
-        RolledDynamicCost += TTI.getUserCost(&I, CostKind);
+        RolledDynamicCost += TTI.getInstructionCost(&I, CostKind);
 
         // Visit the instruction to analyze its loop cost after unrolling,
         // and if the visitor returns true, mark the instruction as free after
         // unrolling and continue.
         bool IsFree = Analyzer.visit(I);
         bool Inserted = InstCostMap.insert({&I, (int)Iteration,
                                            (unsigned)IsFree,
                                            /*IsCounted*/ false}).second;

llvm/lib/Transforms/Scalar/SimpleLoopUnswitch.cpp

for (auto &I : *BB) {
         continue;
 
       if (I.getType()->isTokenTy() && I.isUsedOutsideOfBlock(BB))
         return false;
       if (auto *CB = dyn_cast<CallBase>(&I))
         if (CB->isConvergent() || CB->cannotDuplicate())
           return false;
 
-      Cost += TTI.getUserCost(&I, CostKind);
+      Cost += TTI.getInstructionCost(&I, CostKind);
     }
     assert(Cost >= 0 && "Must not have negative costs!");
     LoopCost += Cost;
     assert(LoopCost >= 0 && "Must not have negative loop costs!");
     BBCostMap[BB] = Cost;
   }
   LLVM_DEBUG(dbgs() << "  Total loop cost: " << LoopCost << "\n");
 

llvm/lib/Transforms/Scalar/SpeculativeExecution.cpp

switch (Operator::getOpcode(I)) {
     case Instruction::FCmp:
     case Instruction::Trunc:
     case Instruction::Freeze:
     case Instruction::ExtractElement:
     case Instruction::InsertElement:
     case Instruction::ShuffleVector:
     case Instruction::ExtractValue:
     case Instruction::InsertValue:
-      return TTI.getUserCost(I, TargetTransformInfo::TCK_SizeAndLatency);
+      return TTI.getInstructionCost(I, TargetTransformInfo::TCK_SizeAndLatency);
 
     default:
       return InstructionCost::getInvalid(); // Disallow anything not explicitly
                                             // listed.
   }
 }
 
 bool SpeculativeExecutionPass::considerHoistingFromTo(

llvm/lib/Transforms/Utils/SimplifyCFG.cpp

 /// which is assumed to be safe to speculate. TCC_Free means cheap,
 /// TCC_Basic means less cheap, and TCC_Expensive means prohibitively
 /// expensive.
 static InstructionCost computeSpeculationCost(const User *I,
                                               const TargetTransformInfo &TTI) {
   assert((!isa<Instruction>(I) ||
           isSafeToSpeculativelyExecute(cast<Instruction>(I))) &&
          "Instruction is not safe to speculatively execute!");
-  return TTI.getUserCost(I, TargetTransformInfo::TCK_SizeAndLatency);
+  return TTI.getInstructionCost(I, TargetTransformInfo::TCK_SizeAndLatency);
 }
 
 /// If we have a merge point of an "if condition" as accepted above,
 /// return true if the specified value dominates the block.  We
 /// don't handle the true generality of domination here, just a special case
 /// which works well enough for us.
 ///
 /// If AggressiveInsts is non-null, and if V does not dominate BB, we check to
if (isa<DbgInfoIntrinsic>(I) || isa<BranchInst>(I))
       continue;
     // I must be safe to execute unconditionally.
     if (!isSafeToSpeculativelyExecute(&I))
       return false;
     SawVectorOp |= isVectorOp(I);
 
     // Account for the cost of duplicating this instruction into each
     // predecessor. Ignore free instructions.
-    if (!TTI ||
-        TTI->getUserCost(&I, CostKind) != TargetTransformInfo::TCC_Free) {
+    if (!TTI || TTI->getInstructionCost(&I, CostKind) !=
+                    TargetTransformInfo::TCC_Free) {
       NumBonusInsts += PredCount;
 
       // Early exits once we reach the limit.
       if (NumBonusInsts >
           BonusInstThreshold * BranchFoldToCommonDestVectorMultiplier)
         return false;
     }
 
for (auto &I : BB->instructionsWithoutDebug(false)) {
       if (auto *S = dyn_cast<StoreInst>(&I))
         if (llvm::find(FreeStores, S))
           continue;
       // Else, we have a white-list of instructions that we are ak speculating.
       if (!isa<BinaryOperator>(I) && !isa<GetElementPtrInst>(I))
         return false; // Not in white-list - not worthwhile folding.
       // And finally, if this is a non-free instruction that we are okay
       // speculating, ensure that we consider the speculation budget.
-      Cost += TTI.getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency);
+      Cost +=
+          TTI.getInstructionCost(&I, TargetTransformInfo::TCK_SizeAndLatency);
       if (Cost > Budget)
         return false; // Eagerly refuse to fold as soon as we're out of budget.
     }
     assert(Cost <= Budget &&
            "When we run out of budget we will eagerly return from within the "
            "per-instruction loop.");
     return true;
   };

llvm/test/Analysis/CostModel/AArch64/sve-math.ll

 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
 ; RUN: opt -mtriple=aarch64-- -mattr=+sve -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=throughput   < %s | FileCheck %s --check-prefix=THRU
 ; RUN: opt -mtriple=aarch64-- -mattr=+sve -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency      < %s | FileCheck %s --check-prefix=LATE
 ; RUN: opt -mtriple=aarch64-- -mattr=+sve -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size    < %s | FileCheck %s --check-prefix=SIZE
 ; RUN: opt -mtriple=aarch64-- -mattr=+sve -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency < %s | FileCheck %s --check-prefix=SIZE_LATE
 
 declare <vscale x 2 x double> @llvm.sqrt.v2f64(<vscale x 2 x double>)
 
 define <vscale x 2 x double> @fadd_v2f64(<vscale x 2 x double> %a, <vscale x 2 x double> %b) {
 ; THRU-LABEL: 'fadd_v2f64'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r = fadd <vscale x 2 x double> %a, %b
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <vscale x 2 x double> %r
 ;
 ; LATE-LABEL: 'fadd_v2f64'
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %r = fadd <vscale x 2 x double> %a, %b
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <vscale x 2 x double> %r
 ;
 ; SIZE-LABEL: 'fadd_v2f64'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r = fadd <vscale x 2 x double> %a, %b
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <vscale x 2 x double> %r
 ;
 ; SIZE_LATE-LABEL: 'fadd_v2f64'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %r = fadd <vscale x 2 x double> %a, %b
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <vscale x 2 x double> %r
 ;
   %r = fadd <vscale x 2 x double> %a, %b
   ret <vscale x 2 x double> %r
 }
 
 define <vscale x 2 x double> @sqrt_v2f64(<vscale x 2 x double> %a) {
 ; THRU-LABEL: 'sqrt_v2f64'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r = call <vscale x 2 x double> @llvm.sqrt.nxv2f64(<vscale x 2 x double> %a)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <vscale x 2 x double> %r
 ;
 ; LATE-LABEL: 'sqrt_v2f64'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %r = call <vscale x 2 x double> @llvm.sqrt.nxv2f64(<vscale x 2 x double> %a)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r = call <vscale x 2 x double> @llvm.sqrt.nxv2f64(<vscale x 2 x double> %a)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <vscale x 2 x double> %r
 ;
 ; SIZE-LABEL: 'sqrt_v2f64'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r = call <vscale x 2 x double> @llvm.sqrt.nxv2f64(<vscale x 2 x double> %a)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <vscale x 2 x double> %r
 ;
 ; SIZE_LATE-LABEL: 'sqrt_v2f64'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r = call <vscale x 2 x double> @llvm.sqrt.nxv2f64(<vscale x 2 x double> %a)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <vscale x 2 x double> %r
 ;
   %r = call <vscale x 2 x double> @llvm.sqrt.v2f64(<vscale x 2 x double> %a)
   ret <vscale x 2 x double> %r
 }

llvm/test/Analysis/CostModel/ARM/intrinsic-cost-kinds.ll

 
 define void @smax(i32 %a, i32 %b, <16 x i32> %va, <16 x i32> %vb) {
 ; THRU-LABEL: 'smax'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'smax'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'smax'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'smax'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
   %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
   ret void
 }
 
 define void @fmuladd(float %a, float %b, float %c, <16 x float> %va, <16 x float> %vb, <16 x float> %vc) {
 ; THRU-LABEL: 'fmuladd'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'fmuladd'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'fmuladd'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'fmuladd'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
   %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
   ret void
 }
 
 define void @log2(float %a, <16 x float> %va) {
 ; THRU-LABEL: 'log2'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.log2.f32(float %a)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'log2'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %s = call float @llvm.log2.f32(float %a)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.log2.f32(float %a)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'log2'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.log2.f32(float %a)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'log2'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.log2.f32(float %a)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call float @llvm.log2.f32(float %a)
   %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
   ret void
 }
 
 define void @constrained_fadd(float %a, <16 x float> %va) {
 ; THRU-LABEL: 'constrained_fadd'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'constrained_fadd'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+; LATE-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'constrained_fadd'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'constrained_fadd'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
   %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
   ret void
 }
 
 define void @fmaximum(float %a, float %b, <16 x float> %va, <16 x float> %vb) {
 ; THRU-LABEL: 'fmaximum'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.maximum.f32(float %a, float %b)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 208 for instruction: %v = call <16 x float> @llvm.maximum.v16f32(<16 x float> %va, <16 x float> %vb)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'fmaximum'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call float @llvm.maximum.f32(float %a, float %b)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call <16 x float> @llvm.maximum.v16f32(<16 x float> %va, <16 x float> %vb)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.maximum.f32(float %a, float %b)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 208 for instruction: %v = call <16 x float> @llvm.maximum.v16f32(<16 x float> %va, <16 x float> %vb)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'fmaximum'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.maximum.f32(float %a, float %b)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %v = call <16 x float> @llvm.maximum.v16f32(<16 x float> %va, <16 x float> %vb)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'fmaximum'

 define void @cttz(i32 %a, <16 x i32> %va) {
 ; THRU-LABEL: 'cttz'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.cttz.i32(i32 %a, i1 false)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x i32> @llvm.cttz.v16i32(<16 x i32> %va, i1 false)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'cttz'
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.cttz.i32(i32 %a, i1 false)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v = call <16 x i32> @llvm.cttz.v16i32(<16 x i32> %va, i1 false)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x i32> @llvm.cttz.v16i32(<16 x i32> %va, i1 false)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'cttz'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.cttz.i32(i32 %a, i1 false)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x i32> @llvm.cttz.v16i32(<16 x i32> %va, i1 false)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'cttz'

 define void @ctlz(i32 %a, <16 x i32> %va) {
 ; THRU-LABEL: 'ctlz'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'ctlz'
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'ctlz'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'ctlz'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
   %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
   ret void
 }
 
 define void @fshl(i32 %a, i32 %b, i32 %c, <16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc) {
 ; THRU-LABEL: 'fshl'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 256 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'fshl'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 250 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'fshl'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 229 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'fshl'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 250 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
   %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
   ret void
 }
 
 define void @maskedgather(<16 x float*> %va, <16 x i1> %vb, <16 x float> %vc) {
 ; THRU-LABEL: 'maskedgather'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'maskedgather'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'maskedgather'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'maskedgather'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
   ret void
 }
 
 define void @maskedscatter(<16 x float> %va, <16 x float*> %vb, <16 x i1> %vc) {
 ; THRU-LABEL: 'maskedscatter'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'maskedscatter'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'maskedscatter'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'maskedscatter'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
   ret void
 }
 
 define void @reduce_fmax(<16 x float> %va) {
 ; THRU-LABEL: 'reduce_fmax'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'reduce_fmax'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'reduce_fmax'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 122 for instruction: %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'reduce_fmax'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
   ret void
 }
 
 define void @memcpy(i8* %a, i8* %b, i32 %c) {
 ; THRU-LABEL: 'memcpy'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'memcpy'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'memcpy'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'memcpy'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
   ret void
 }

llvm/test/Analysis/CostModel/ARM/target-intrinsics.ll

 ; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t1 = call i32 @llvm.arm.ssat(i32 undef, i32 undef)
 ; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t2 = tail call { <8 x half>, <8 x half> } @llvm.arm.mve.vld2q.v8f16.p0f16(half* undef)
 ; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t3 = call { i32, i32 } @llvm.arm.mve.sqrshrl(i32 undef, i32 undef, i32 undef, i32 48)
 ; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t4 = tail call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 0, i32 0, i32 0, i32 0, i32 0, <8 x i16> undef, <8 x i16> undef)
 ; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; CHECK-THUMB2-LAT-LABEL: 'intrinsics'
 ; CHECK-THUMB2-LAT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t1 = call i32 @llvm.arm.ssat(i32 undef, i32 undef)
-; CHECK-THUMB2-LAT-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %t2 = tail call { <8 x half>, <8 x half> } @llvm.arm.mve.vld2q.v8f16.p0f16(half* undef)
+; CHECK-THUMB2-LAT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t2 = tail call { <8 x half>, <8 x half> } @llvm.arm.mve.vld2q.v8f16.p0f16(half* undef)
 ; CHECK-THUMB2-LAT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t3 = call { i32, i32 } @llvm.arm.mve.sqrshrl(i32 undef, i32 undef, i32 undef, i32 48)
 ; CHECK-THUMB2-LAT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t4 = tail call { i32, i32 } @llvm.arm.mve.vmlldava.v8i16(i32 0, i32 0, i32 0, i32 0, i32 0, <8 x i16> undef, <8 x i16> undef)
 ; CHECK-THUMB2-LAT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-THUMB2-SIZE-LABEL: 'intrinsics'
 ; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t1 = call i32 @llvm.arm.ssat(i32 undef, i32 undef)
 ; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t2 = tail call { <8 x half>, <8 x half> } @llvm.arm.mve.vld2q.v8f16.p0f16(half* undef)
 ; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %t3 = call { i32, i32 } @llvm.arm.mve.sqrshrl(i32 undef, i32 undef, i32 undef, i32 48)

llvm/test/Analysis/CostModel/SystemZ/ext-of-icmp-cost.ll

 ; RUN: opt < %s -passes="print<cost-model>" -cost-kind=code-size 2>&1 -disable-output \
 ; RUN:   -mtriple=s390x-unknown-linux -mcpu=z13  | FileCheck %s
 ;
-; Check that getUserCost() does not return TCC_Free for extensions of
+; Check that getInstructionCost() does not return TCC_Free for extensions of
 ; i1 returned from icmp.
 
 define i64 @fun1(i64 %v) {
 ; CHECK-LABEL: 'fun1'
 ; CHECK: Cost Model: Found an estimated cost of 1 for instruction:   %cmp = icmp eq i64 %v, 0
 ; CHECK: Cost Model: Found an estimated cost of 1 for instruction:   %z = zext i1 %cmp to i64
 ; CHECK: Cost Model: Found an estimated cost of 1 for instruction:   ret i64 %z
   %cmp = icmp eq i64 %v, 0

llvm/test/Analysis/CostModel/X86/arith-fp-costkinds.ll

 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=latency      < %s | FileCheck %s --check-prefixes=LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=SIZE,SSE-SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=SIZE_LATE,SSE-SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=latency      < %s | FileCheck %s --check-prefixes=LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=SIZE,AVX-SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=SIZE_LATE,AVX-SIZE_LATE,AVX1-SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=latency      < %s | FileCheck %s --check-prefixes=LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=SIZE,AVX-SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=SIZE_LATE,AVX-SIZE_LATE,AVX2-SIZE_LATE
+; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=latency      < %s | FileCheck %s --check-prefixes=CHECK,LATE,SSE-LATE
+; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=CHECK,SIZE,SSE-SIZE
+; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=CHECK,SIZE_LATE,SSE-SIZE_LATE
+;
+; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=latency      < %s | FileCheck %s --check-prefixes=CHECK,LATE,AVX-LATE,AVX1-LATE
+; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=CHECK,SIZE,AVX-SIZE
+; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=CHECK,SIZE_LATE,AVX-SIZE_LATE,AVX1-SIZE_LATE
+;
+; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=latency      < %s | FileCheck %s --check-prefixes=CHECK,LATE,AVX-LATE,AVX2-LATE
+; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=CHECK,SIZE,AVX-SIZE
+; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=CHECK,SIZE_LATE,AVX-SIZE_LATE,AVX2-SIZE_LATE
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 target triple = "x86_64-apple-macosx10.8.0"
 
 define i32 @fadd(i32 %arg) {
 ; LATE-LABEL: 'fadd'
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fadd float undef, undef
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fadd <4 x float> undef, undef
;
   %V2F64 = fmul <2 x double> undef, undef
   %V4F64 = fmul <4 x double> undef, undef
   %V8F64 = fmul <8 x double> undef, undef
 
   ret i32 undef
 }
 
 define i32 @fdiv(i32 %arg) {
-; LATE-LABEL: 'fdiv'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fdiv float undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fdiv <4 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fdiv <8 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fdiv <16 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fdiv double undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fdiv <2 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fdiv <4 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fdiv <8 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'fdiv'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = fdiv float undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fdiv <4 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fdiv <8 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fdiv <16 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = fdiv double undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fdiv <2 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fdiv <4 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fdiv <8 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'fdiv'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = fdiv float undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fdiv <4 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fdiv <8 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fdiv <16 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = fdiv double undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fdiv <2 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fdiv <4 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fdiv <8 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+; CHECK-LABEL: 'fdiv'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = fdiv float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fdiv <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fdiv <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fdiv <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = fdiv double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fdiv <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fdiv <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fdiv <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
   %F32 = fdiv float undef, undef
   %V4F32 = fdiv <4 x float> undef, undef
   %V8F32 = fdiv <8 x float> undef, undef
   %V16F32 = fdiv <16 x float> undef, undef
 
   %F64 = fdiv double undef, undef
   %V2F64 = fdiv <2 x double> undef, undef
   %V4F64 = fdiv <4 x double> undef, undef
   %V8F64 = fdiv <8 x double> undef, undef
 
   ret i32 undef
 }
 
 define i32 @frem(i32 %arg) {
-; LATE-LABEL: 'frem'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = frem float undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = frem <4 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = frem <8 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = frem <16 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = frem double undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = frem <2 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = frem <4 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = frem <8 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'frem'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = frem float undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = frem <4 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = frem <8 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = frem <16 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = frem double undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = frem <2 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = frem <4 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = frem <8 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'frem'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = frem float undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = frem <4 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = frem <8 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = frem <16 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = frem double undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = frem <2 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = frem <4 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = frem <8 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+; CHECK-LABEL: 'frem'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = frem float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = frem <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = frem <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = frem <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = frem double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = frem <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = frem <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = frem <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
   %F32 = frem float undef, undef
   %V4F32 = frem <4 x float> undef, undef
   %V8F32 = frem <8 x float> undef, undef
   %V16F32 = frem <16 x float> undef, undef
 
   %F64 = frem double undef, undef
   %V2F64 = frem <2 x double> undef, undef
   %V4F64 = frem <4 x double> undef, undef
   %V8F64 = frem <8 x double> undef, undef
 
   ret i32 undef
 }
 
 define i32 @fsqrt(i32 %arg) {
-; LATE-LABEL: 'fsqrt'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+; SSE-LATE-LABEL: 'fsqrt'
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; SSE-SIZE-LABEL: 'fsqrt'
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; SSE-SIZE_LATE-LABEL: 'fsqrt'
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
+; AVX1-LATE-LABEL: 'fsqrt'
+; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
 ; AVX1-SIZE_LATE-LABEL: 'fsqrt'
 ; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
 ; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
 ; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
 ; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
 ; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
 ; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
 ; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
 ; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
 ; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
+; AVX2-LATE-LABEL: 'fsqrt'
+; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
 ; AVX2-SIZE_LATE-LABEL: 'fsqrt'
 ; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
 ; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
 ; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
 ; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
 ; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
 ; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
 ; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
;
   %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
   %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
   %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
 
   ret i32 undef
 }
 
 define i32 @fabs(i32 %arg) {
-; LATE-LABEL: 'fabs'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+; SSE-LATE-LABEL: 'fabs'
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; SSE-SIZE-LABEL: 'fabs'
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; SSE-SIZE_LATE-LABEL: 'fabs'
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
+; AVX-LATE-LABEL: 'fabs'
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
 ; AVX-SIZE-LABEL: 'fabs'
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
;
   %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
   %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
   %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
 
   ret i32 undef
 }
 
 define i32 @fcopysign(i32 %arg) {
-; LATE-LABEL: 'fcopysign'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+; SSE-LATE-LABEL: 'fcopysign'
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; SSE-SIZE-LABEL: 'fcopysign'
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; SSE-SIZE_LATE-LABEL: 'fcopysign'
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
+; AVX-LATE-LABEL: 'fcopysign'
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
 ; AVX-SIZE-LABEL: 'fcopysign'
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
;
   %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
   %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
   %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
 
   ret i32 undef
 }
 
 define i32 @fma(i32 %arg) {
-; LATE-LABEL: 'fma'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+; SSE-LATE-LABEL: 'fma'
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; SSE-SIZE-LABEL: 'fma'
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
 ; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; SSE-SIZE_LATE-LABEL: 'fma'
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
 ; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
+; AVX-LATE-LABEL: 'fma'
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
 ; AVX-SIZE-LABEL: 'fma'
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
 ; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)

llvm/test/Analysis/CostModel/X86/costmodel.ll

 ; LATENCY-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %A2 = alloca i64, i64 undef, align 8
 ; LATENCY-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = add i64 undef, undef
 ; LATENCY-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %1 = load i64, i64* undef, align 4
 ; LATENCY-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %BC = bitcast i8* undef to i32*
 ; LATENCY-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %I2P = inttoptr i64 undef to i8*
 ; LATENCY-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %P2I = ptrtoint i8* undef to i64
 ; LATENCY-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %TC = trunc i64 undef to i32
 ; LATENCY-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %uadd = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
-; LATENCY-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: call void undef()
+; LATENCY-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void undef()
 ; LATENCY-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i64 undef
 ;
 ; CODESIZE-LABEL: 'foo'
 ; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %A1 = alloca i32, align 8
 ; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %A2 = alloca i64, i64 undef, align 8
 ; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = add i64 undef, undef
 ; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = load i64, i64* undef, align 4
 ; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %BC = bitcast i8* undef to i32*

llvm/test/Analysis/CostModel/X86/intrinsic-cost-kinds.ll

 
 define void @umul(i32 %a, i32 %b, <16 x i32> %va, <16 x i32> %vb) {
 ; THRU-LABEL: 'umul'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %v = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'umul'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> %va, <16 x i32> %vb)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'umul'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'umul'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
   %v = call {<16 x i32>, <16 x i1>} @llvm.umul.with.overflow.v16i32(<16 x i32> %va, <16 x i32> %vb)
   ret void
 }
 
 define void @smax(i32 %a, i32 %b, <16 x i32> %va, <16 x i32> %vb) {
 ; THRU-LABEL: 'smax'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'smax'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'smax'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'smax'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
   %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
   ret void
 }
 
 define void @fcopysign(float %a, float %b, <16 x float> %va, <16 x float> %vb) {
 ; THRU-LABEL: 'fcopysign'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call float @llvm.copysign.f32(float %a, float %b)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x float> @llvm.copysign.v16f32(<16 x float> %va, <16 x float> %vb)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'fcopysign'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call float @llvm.copysign.f32(float %a, float %b)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call <16 x float> @llvm.copysign.v16f32(<16 x float> %va, <16 x float> %vb)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call float @llvm.copysign.f32(float %a, float %b)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x float> @llvm.copysign.v16f32(<16 x float> %va, <16 x float> %vb)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'fcopysign'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call float @llvm.copysign.f32(float %a, float %b)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x float> @llvm.copysign.v16f32(<16 x float> %va, <16 x float> %vb)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'fcopysign'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call float @llvm.copysign.f32(float %a, float %b)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v = call <16 x float> @llvm.copysign.v16f32(<16 x float> %va, <16 x float> %vb)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call float @llvm.copysign.f32(float %a, float %b)
   %v = call <16 x float> @llvm.copysign.v16f32(<16 x float> %va, <16 x float> %vb)
   ret void
 }
 
 define void @fmuladd(float %a, float %b, float %c, <16 x float> %va, <16 x float> %vb, <16 x float> %vc) {
 ; THRU-LABEL: 'fmuladd'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'fmuladd'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'fmuladd'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'fmuladd'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
   %v = call <16 x float> @llvm.fmuladd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x float> %vc)
   ret void
 }
 
 define void @log2(float %a, <16 x float> %va) {
 ; THRU-LABEL: 'log2'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.log2.f32(float %a)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'log2'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call float @llvm.log2.f32(float %a)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.log2.f32(float %a)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'log2'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.log2.f32(float %a)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'log2'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.log2.f32(float %a)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call float @llvm.log2.f32(float %a)
   %v = call <16 x float> @llvm.log2.v16f32(<16 x float> %va)
   ret void
 }
 
 define void @constrained_fadd(float %a, <16 x float> %va) {
 ; THRU-LABEL: 'constrained_fadd'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'constrained_fadd'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+; LATE-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'constrained_fadd'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'constrained_fadd'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call float @llvm.experimental.constrained.fadd.f32(float %a, float %a, metadata !"round.dynamic", metadata !"fpexcept.ignore")
   %t = call <16 x float> @llvm.experimental.constrained.fadd.v16f32(<16 x float> %va, <16 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
   ret void
 }
 
 define void @fmaximum(float %a, float %b, <16 x float> %va, <16 x float> %vb) {
 ; THRU-LABEL: 'fmaximum'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.maximum.f32(float %a, float %b)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 196 for instruction: %v = call <16 x float> @llvm.maximum.v16f32(<16 x float> %va, <16 x float> %vb)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'fmaximum'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call float @llvm.maximum.f32(float %a, float %b)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call <16 x float> @llvm.maximum.v16f32(<16 x float> %va, <16 x float> %vb)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.maximum.f32(float %a, float %b)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 196 for instruction: %v = call <16 x float> @llvm.maximum.v16f32(<16 x float> %va, <16 x float> %vb)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'fmaximum'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call float @llvm.maximum.f32(float %a, float %b)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %v = call <16 x float> @llvm.maximum.v16f32(<16 x float> %va, <16 x float> %vb)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'fmaximum'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %s = call float @llvm.maximum.f32(float %a, float %b)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 196 for instruction: %v = call <16 x float> @llvm.maximum.v16f32(<16 x float> %va, <16 x float> %vb)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call float @llvm.maximum.f32(float %a, float %b)
   %v = call <16 x float> @llvm.maximum.v16f32(<16 x float> %va, <16 x float> %vb)
   ret void
 }
 
 define void @cttz(i32 %a, <16 x i32> %va) {
 ; THRU-LABEL: 'cttz'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call i32 @llvm.cttz.i32(i32 %a, i1 false)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %v = call <16 x i32> @llvm.cttz.v16i32(<16 x i32> %va, i1 false)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'cttz'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.cttz.i32(i32 %a, i1 false)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v = call <16 x i32> @llvm.cttz.v16i32(<16 x i32> %va, i1 false)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call i32 @llvm.cttz.i32(i32 %a, i1 false)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %v = call <16 x i32> @llvm.cttz.v16i32(<16 x i32> %va, i1 false)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'cttz'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call i32 @llvm.cttz.i32(i32 %a, i1 false)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %v = call <16 x i32> @llvm.cttz.v16i32(<16 x i32> %va, i1 false)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'cttz'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %s = call i32 @llvm.cttz.i32(i32 %a, i1 false)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %v = call <16 x i32> @llvm.cttz.v16i32(<16 x i32> %va, i1 false)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call i32 @llvm.cttz.i32(i32 %a, i1 false)
   %v = call <16 x i32> @llvm.cttz.v16i32(<16 x i32> %va, i1 false)
   ret void
 }
 
 define void @ctlz(i32 %a, <16 x i32> %va) {
 ; THRU-LABEL: 'ctlz'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'ctlz'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'ctlz'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'ctlz'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call i32 @llvm.ctlz.i32(i32 %a, i1 true)
   %v = call <16 x i32> @llvm.ctlz.v16i32(<16 x i32> %va, i1 true)
   ret void
 }
 
 define void @fshl(i32 %a, i32 %b, i32 %c, <16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc) {
 ; THRU-LABEL: 'fshl'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'fshl'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'fshl'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'fshl'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
   %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
   ret void
 }
 
 define void @maskedgather(<16 x float*> %va, <16 x i1> %vb, <16 x float> %vc) {
 ; THRU-LABEL: 'maskedgather'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'maskedgather'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'maskedgather'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'maskedgather'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %v = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %va, i32 1, <16 x i1> %vb, <16 x float> %vc)
   ret void
 }
 
 define void @maskedscatter(<16 x float> %va, <16 x float*> %vb, <16 x i1> %vc) {
 ; THRU-LABEL: 'maskedscatter'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'maskedscatter'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'maskedscatter'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'maskedscatter'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> %va, <16 x float*> %vb, i32 1, <16 x i1> %vc)
   ret void
 }
 
 define void @reduce_fmax(<16 x float> %va) {
 ; THRU-LABEL: 'reduce_fmax'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'reduce_fmax'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'reduce_fmax'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'reduce_fmax'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %v = call float @llvm.vector.reduce.fmax.v16f32(<16 x float> %va)
   ret void
 }
 
 define void @reduce_fmul(<16 x float> %va) {
 ; THRU-LABEL: 'reduce_fmul'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v = call float @llvm.vector.reduce.fmul.v16f32(float 4.200000e+01, <16 x float> %va)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'reduce_fmul'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call float @llvm.vector.reduce.fmul.v16f32(float 4.200000e+01, <16 x float> %va)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %v = call float @llvm.vector.reduce.fmul.v16f32(float 4.200000e+01, <16 x float> %va)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'reduce_fmul'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v = call float @llvm.vector.reduce.fmul.v16f32(float 4.200000e+01, <16 x float> %va)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'reduce_fmul'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v = call float @llvm.vector.reduce.fmul.v16f32(float 4.200000e+01, <16 x float> %va)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %v = call float @llvm.vector.reduce.fmul.v16f32(float 42.0, <16 x float> %va)
   ret void
 }
 
 define void @reduce_fadd_fast(<16 x float> %va) {
 ; THRU-LABEL: 'reduce_fadd_fast'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v = call fast float @llvm.vector.reduce.fadd.v16f32(float 0.000000e+00, <16 x float> %va)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'reduce_fadd_fast'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v = call fast float @llvm.vector.reduce.fadd.v16f32(float 0.000000e+00, <16 x float> %va)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %v = call fast float @llvm.vector.reduce.fadd.v16f32(float 0.000000e+00, <16 x float> %va)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'reduce_fadd_fast'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v = call fast float @llvm.vector.reduce.fadd.v16f32(float 0.000000e+00, <16 x float> %va)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'reduce_fadd_fast'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v = call fast float @llvm.vector.reduce.fadd.v16f32(float 0.000000e+00, <16 x float> %va)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %v = call fast float @llvm.vector.reduce.fadd.v16f32(float 0.0, <16 x float> %va)
   ret void
 }
 
 define void @memcpy(i8* %a, i8* %b, i32 %c) {
 ; THRU-LABEL: 'memcpy'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'memcpy'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
+; LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
 ; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE-LABEL: 'memcpy'
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SIZE_LATE-LABEL: 'memcpy'
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
 ; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
   ret void
 }