[llvm] [VPlan] Make canonical IV part of the region (PR #156262)

Florian Hahn via llvm-commits llvm-commits at lists.llvm.org
Sun Aug 31 13:08:21 PDT 2025


https://github.com/fhahn updated https://github.com/llvm/llvm-project/pull/156262

>From 1246f74568c9726f271a510dd857b5a58a95d30d Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Mon, 25 Aug 2025 22:22:04 +0100
Subject: [PATCH 1/3] [VPlan] Track VPValues instead of VPRecipes in
 calculateRegisterUsage.

Update calculateRegisterUsageForPlan to track live-ness of VPValues
instead of recipes. This gives slightly more accurate results for
recipes that define multiple values (i.e. VPInterleaveRecipe).

When tracking the live-ness of recipes, all VPValues defined by an
VPInterleaveRecipe are considered alive until the last use of any of
them. When tracking the live-ness of individual VPValues, we can
accurately track the individual values until their last use.

Note the changes in large-loop-rdx.ll and pr47437.ll. This patch
restores the original behavior before introducing VPlan-based liveness
tracking.
---
 .../Transforms/Vectorize/VPlanAnalysis.cpp    |  74 ++---
 .../LoopVectorize/AArch64/reg-usage.ll        |   2 +-
 .../LoopVectorize/PowerPC/large-loop-rdx.ll   | 262 ++++++++++++++----
 .../Transforms/LoopVectorize/X86/pr47437.ll   | 108 ++++++--
 4 files changed, 338 insertions(+), 108 deletions(-)

diff --git a/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp b/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
index 747c6623aa22a..8e5997fa8befe 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
@@ -10,6 +10,7 @@
 #include "VPlan.h"
 #include "VPlanCFG.h"
 #include "VPlanDominatorTree.h"
+#include "VPlanHelpers.h"
 #include "llvm/ADT/PostOrderIterator.h"
 #include "llvm/ADT/TypeSwitch.h"
 #include "llvm/Analysis/ScalarEvolution.h"
@@ -396,7 +397,7 @@ bool VPDominatorTree::properlyDominates(const VPRecipeBase *A,
 
 /// Get the VF scaling factor applied to the recipe's output, if the recipe has
 /// one.
-static unsigned getVFScaleFactor(VPRecipeBase *R) {
+static unsigned getVFScaleFactor(VPValue *R) {
   if (auto *RR = dyn_cast<VPReductionPHIRecipe>(R))
     return RR->getVFScaleFactor();
   if (auto *RR = dyn_cast<VPPartialReductionRecipe>(R))
@@ -422,15 +423,15 @@ SmallVector<VPRegisterUsage, 8> llvm::calculateRegisterUsageForPlan(
     const SmallPtrSetImpl<const Value *> &ValuesToIgnore) {
   // Each 'key' in the map opens a new interval. The values
   // of the map are the index of the 'last seen' usage of the
-  // recipe that is the key.
-  using IntervalMap = SmallDenseMap<VPRecipeBase *, unsigned, 16>;
+  // VPValue that is the key.
+  using IntervalMap = SmallDenseMap<VPValue *, unsigned, 16>;
 
   // Maps indices to recipes.
   SmallVector<VPRecipeBase *, 64> Idx2Recipe;
   // Marks the end of each interval.
   IntervalMap EndPoint;
-  // Saves the list of recipe indices that are used in the loop.
-  SmallPtrSet<VPRecipeBase *, 8> Ends;
+  // Saves the list of VPValues that are used in the loop.
+  SmallPtrSet<VPValue *, 8> Ends;
   // Saves the list of values that are used in the loop but are defined outside
   // the loop (not including non-recipe values such as arguments and
   // constants).
@@ -441,7 +442,7 @@ SmallVector<VPRegisterUsage, 8> llvm::calculateRegisterUsageForPlan(
   // each recipe. We use RPO to ensure that defs are met before their users. We
   // assume that each recipe that has in-loop users starts an interval. We
   // record every time that an in-loop value is used, so we have a list of the
-  // first and last occurrences of each recipe.
+  // first occurences of each recipe and last occurrence of each VPValue.
   VPRegionBlock *LoopRegion = Plan.getVectorLoopRegion();
   ReversePostOrderTraversal<VPBlockDeepTraversalWrapper<VPBlockBase *>> RPOT(
       LoopRegion);
@@ -470,32 +471,32 @@ SmallVector<VPRegisterUsage, 8> llvm::calculateRegisterUsageForPlan(
         }
 
         // Overwrite previous end points.
-        EndPoint[DefR] = Idx2Recipe.size();
-        Ends.insert(DefR);
+        EndPoint[U] = Idx2Recipe.size();
+        Ends.insert(U);
       }
     }
     if (VPBB == LoopRegion->getExiting()) {
       // VPWidenIntOrFpInductionRecipes are used implicitly at the end of the
       // exiting block, where their increment will get materialized eventually.
       for (auto &R : LoopRegion->getEntryBasicBlock()->phis()) {
-        if (isa<VPWidenIntOrFpInductionRecipe>(&R)) {
-          EndPoint[&R] = Idx2Recipe.size();
-          Ends.insert(&R);
+        if (auto *WideIV = dyn_cast<VPWidenIntOrFpInductionRecipe>(&R)) {
+          EndPoint[WideIV] = Idx2Recipe.size();
+          Ends.insert(WideIV);
         }
       }
     }
   }
 
   // Saves the list of intervals that end with the index in 'key'.
-  using RecipeList = SmallVector<VPRecipeBase *, 2>;
-  SmallDenseMap<unsigned, RecipeList, 16> TransposeEnds;
+  using VPValueList = SmallVector<VPValue *, 2>;
+  SmallDenseMap<unsigned, VPValueList, 16> TransposeEnds;
 
   // Next, we transpose the EndPoints into a multi map that holds the list of
   // intervals that *end* at a specific location.
   for (auto &Interval : EndPoint)
     TransposeEnds[Interval.second].push_back(Interval.first);
 
-  SmallPtrSet<VPRecipeBase *, 8> OpenIntervals;
+  SmallPtrSet<VPValue *, 8> OpenIntervals;
   SmallVector<VPRegisterUsage, 8> RUs(VFs.size());
   SmallVector<SmallMapVector<unsigned, unsigned, 4>, 8> MaxUsages(VFs.size());
 
@@ -519,14 +520,16 @@ SmallVector<VPRegisterUsage, 8> llvm::calculateRegisterUsageForPlan(
   for (unsigned int Idx = 0, Sz = Idx2Recipe.size(); Idx < Sz; ++Idx) {
     VPRecipeBase *R = Idx2Recipe[Idx];
 
-    // Remove all of the recipes that end at this location.
-    RecipeList &List = TransposeEnds[Idx];
-    for (VPRecipeBase *ToRemove : List)
+    // Remove all of the VPValues that end at this location.
+    VPValueList &List = TransposeEnds[Idx];
+    for (VPValue *ToRemove : List)
       OpenIntervals.erase(ToRemove);
 
     // Ignore recipes that are never used within the loop and do not have side
     // effects.
-    if (!Ends.count(R) && !R->mayHaveSideEffects())
+    if (all_of(R->definedValues(),
+               [&Ends](VPValue *Def) { return !Ends.count(Def); }) &&
+        !R->mayHaveSideEffects())
       continue;
 
     // Skip recipes for ignored values.
@@ -546,41 +549,38 @@ SmallVector<VPRegisterUsage, 8> llvm::calculateRegisterUsageForPlan(
       // there is no previous entry for ClassID.
       SmallMapVector<unsigned, unsigned, 4> RegUsage;
 
-      for (auto *R : OpenIntervals) {
-        // Skip recipes that weren't present in the original loop.
+      for (auto *VPV : OpenIntervals) {
+        // Skip values that weren't present in the original loop.
         // TODO: Remove after removing the legacy
         // LoopVectorizationCostModel::calculateRegisterUsage
         if (isa<VPVectorPointerRecipe, VPVectorEndPointerRecipe,
-                VPBranchOnMaskRecipe>(R))
+                VPBranchOnMaskRecipe>(VPV))
           continue;
 
         if (VFs[J].isScalar() ||
             isa<VPCanonicalIVPHIRecipe, VPReplicateRecipe, VPDerivedIVRecipe,
-                VPEVLBasedIVPHIRecipe, VPScalarIVStepsRecipe>(R) ||
-            (isa<VPInstruction>(R) &&
-             vputils::onlyScalarValuesUsed(cast<VPSingleDefRecipe>(R))) ||
-            (isa<VPReductionPHIRecipe>(R) &&
-             (cast<VPReductionPHIRecipe>(R))->isInLoop())) {
-          unsigned ClassID = TTI.getRegisterClassForType(
-              false, TypeInfo.inferScalarType(R->getVPSingleValue()));
+                VPEVLBasedIVPHIRecipe, VPScalarIVStepsRecipe>(VPV) ||
+            (isa<VPInstruction>(VPV) && vputils::onlyScalarValuesUsed(VPV)) ||
+            (isa<VPReductionPHIRecipe>(VPV) &&
+             (cast<VPReductionPHIRecipe>(VPV))->isInLoop())) {
+          unsigned ClassID =
+              TTI.getRegisterClassForType(false, TypeInfo.inferScalarType(VPV));
           // FIXME: The target might use more than one register for the type
           // even in the scalar case.
           RegUsage[ClassID] += 1;
         } else {
           // The output from scaled phis and scaled reductions actually has
           // fewer lanes than the VF.
-          unsigned ScaleFactor = getVFScaleFactor(R);
+          unsigned ScaleFactor = getVFScaleFactor(VPV);
           ElementCount VF = VFs[J].divideCoefficientBy(ScaleFactor);
           LLVM_DEBUG(if (VF != VFs[J]) {
             dbgs() << "LV(REG): Scaled down VF from " << VFs[J] << " to " << VF
                    << " for " << *R << "\n";
           });
 
-          for (VPValue *DefV : R->definedValues()) {
-            Type *ScalarTy = TypeInfo.inferScalarType(DefV);
-            unsigned ClassID = TTI.getRegisterClassForType(true, ScalarTy);
-            RegUsage[ClassID] += GetRegUsage(ScalarTy, VF);
-          }
+          Type *ScalarTy = TypeInfo.inferScalarType(VPV);
+          unsigned ClassID = TTI.getRegisterClassForType(true, ScalarTy);
+          RegUsage[ClassID] += GetRegUsage(ScalarTy, VF);
         }
       }
 
@@ -593,8 +593,10 @@ SmallVector<VPRegisterUsage, 8> llvm::calculateRegisterUsageForPlan(
     LLVM_DEBUG(dbgs() << "LV(REG): At #" << Idx << " Interval # "
                       << OpenIntervals.size() << '\n');
 
-    // Add the current recipe to the list of open intervals.
-    OpenIntervals.insert(R);
+    // Add the VPValues defined by the current recipe to the list of open
+    // intervals.
+    for (VPValue *DefV : R->definedValues())
+      OpenIntervals.insert(DefV);
   }
 
   // We also search for instructions that are defined outside the loop, but are
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/reg-usage.ll b/llvm/test/Transforms/LoopVectorize/AArch64/reg-usage.ll
index e51a925040a49..23cc7e367776a 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/reg-usage.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/reg-usage.ll
@@ -74,7 +74,7 @@ define dso_local void @dotp_high_register_pressure(ptr %a, ptr %b, ptr %sum, i32
 ; CHECK:       LV(REG): VF = 16
 ; CHECK-NEXT:  LV(REG): Found max usage: 2 item
 ; CHECK-NEXT:  LV(REG): RegisterClass: Generic::ScalarRC, 3 registers
-; CHECK-NEXT:  LV(REG): RegisterClass: Generic::VectorRC, 48 registers
+; CHECK-NEXT:  LV(REG): RegisterClass: Generic::VectorRC, 47 registers
 ; CHECK-NEXT:  LV(REG): Found invariant usage: 1 item
 entry:
   %cmp100 = icmp sgt i32 %n, 0
diff --git a/llvm/test/Transforms/LoopVectorize/PowerPC/large-loop-rdx.ll b/llvm/test/Transforms/LoopVectorize/PowerPC/large-loop-rdx.ll
index 0b23206134bc0..43cce8005bbf6 100644
--- a/llvm/test/Transforms/LoopVectorize/PowerPC/large-loop-rdx.ll
+++ b/llvm/test/Transforms/LoopVectorize/PowerPC/large-loop-rdx.ll
@@ -10,28 +10,43 @@ define void @QLA_F3_r_veq_norm2_V(ptr noalias %r, ptr noalias %a, i32 %n) {
 ; CHECK-SAME: ptr noalias [[R:%.*]], ptr noalias [[A:%.*]], i32 [[N:%.*]]) {
 ; CHECK-NEXT:  [[ENTRY:.*]]:
 ; CHECK-NEXT:    [[CMP24:%.*]] = icmp sgt i32 [[N]], 0
-; CHECK-NEXT:    br i1 [[CMP24]], label %[[FOR_COND1_PREHEADER_PREHEADER:.*]], label %[[FOR_END13:.*]]
-; CHECK:       [[FOR_COND1_PREHEADER_PREHEADER]]:
+; CHECK-NEXT:    br i1 [[CMP24]], label %[[ITER_CHECK:.*]], label %[[FOR_END13:.*]]
+; CHECK:       [[ITER_CHECK]]:
 ; CHECK-NEXT:    [[TMP0:%.*]] = zext i32 [[N]] to i64
-; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 8
-; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
+; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 2
+; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH:.*]], label %[[VECTOR_MAIN_LOOP_ITER_CHECK:.*]]
+; CHECK:       [[VECTOR_MAIN_LOOP_ITER_CHECK]]:
+; CHECK-NEXT:    [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[TMP0]], 16
+; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK1]], label %[[VEC_EPILOG_PH:.*]], label %[[VECTOR_PH:.*]]
 ; CHECK:       [[VECTOR_PH]]:
-; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 8
+; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 16
 ; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
 ; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
 ; CHECK:       [[VECTOR_BODY]]:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP69:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI1:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP65:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI2:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP66:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI3:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP67:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI2:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP65:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI3:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP66:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI4:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP131:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP25:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP132:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP26:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP133:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP27:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP134:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP28:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP135:%.*]], %[[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[INDEX]], 2
 ; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP3:%.*]] = add i64 [[INDEX]], 6
+; CHECK-NEXT:    [[TMP4:%.*]] = add i64 [[INDEX]], 8
+; CHECK-NEXT:    [[TMP5:%.*]] = add i64 [[INDEX]], 10
+; CHECK-NEXT:    [[TMP6:%.*]] = add i64 [[INDEX]], 12
+; CHECK-NEXT:    [[TMP7:%.*]] = add i64 [[INDEX]], 14
 ; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDEX]], i64 0, i32 0
 ; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[TMP1]], i64 0, i32 0
 ; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[TMP2]], i64 0, i32 0
 ; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[TMP3]], i64 0, i32 0
+; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[TMP4]], i64 0, i32 0
+; CHECK-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[TMP5]], i64 0, i32 0
+; CHECK-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[TMP6]], i64 0, i32 0
+; CHECK-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[TMP7]], i64 0, i32 0
 ; CHECK-NEXT:    [[WIDE_VEC35:%.*]] = load <12 x float>, ptr [[TMP13]], align 8
 ; CHECK-NEXT:    [[STRIDED_VEC36:%.*]] = shufflevector <12 x float> [[WIDE_VEC35]], <12 x float> poison, <2 x i32> <i32 0, i32 6>
 ; CHECK-NEXT:    [[STRIDED_VEC37:%.*]] = shufflevector <12 x float> [[WIDE_VEC35]], <12 x float> poison, <2 x i32> <i32 1, i32 7>
@@ -60,116 +75,252 @@ define void @QLA_F3_r_veq_norm2_V(ptr noalias %r, ptr noalias %a, i32 %n) {
 ; CHECK-NEXT:    [[STRIDED_VEC60:%.*]] = shufflevector <12 x float> [[WIDE_VEC56]], <12 x float> poison, <2 x i32> <i32 3, i32 9>
 ; CHECK-NEXT:    [[STRIDED_VEC61:%.*]] = shufflevector <12 x float> [[WIDE_VEC56]], <12 x float> poison, <2 x i32> <i32 4, i32 10>
 ; CHECK-NEXT:    [[STRIDED_VEC62:%.*]] = shufflevector <12 x float> [[WIDE_VEC56]], <12 x float> poison, <2 x i32> <i32 5, i32 11>
+; CHECK-NEXT:    [[WIDE_VEC36:%.*]] = load <12 x float>, ptr [[TMP12]], align 8
+; CHECK-NEXT:    [[STRIDED_VEC42:%.*]] = shufflevector <12 x float> [[WIDE_VEC36]], <12 x float> poison, <2 x i32> <i32 0, i32 6>
+; CHECK-NEXT:    [[STRIDED_VEC49:%.*]] = shufflevector <12 x float> [[WIDE_VEC36]], <12 x float> poison, <2 x i32> <i32 1, i32 7>
+; CHECK-NEXT:    [[STRIDED_VEC56:%.*]] = shufflevector <12 x float> [[WIDE_VEC36]], <12 x float> poison, <2 x i32> <i32 2, i32 8>
+; CHECK-NEXT:    [[STRIDED_VEC63:%.*]] = shufflevector <12 x float> [[WIDE_VEC36]], <12 x float> poison, <2 x i32> <i32 3, i32 9>
+; CHECK-NEXT:    [[STRIDED_VEC64:%.*]] = shufflevector <12 x float> [[WIDE_VEC36]], <12 x float> poison, <2 x i32> <i32 4, i32 10>
+; CHECK-NEXT:    [[STRIDED_VEC65:%.*]] = shufflevector <12 x float> [[WIDE_VEC36]], <12 x float> poison, <2 x i32> <i32 5, i32 11>
+; CHECK-NEXT:    [[WIDE_VEC43:%.*]] = load <12 x float>, ptr [[TMP17]], align 8
+; CHECK-NEXT:    [[STRIDED_VEC66:%.*]] = shufflevector <12 x float> [[WIDE_VEC43]], <12 x float> poison, <2 x i32> <i32 0, i32 6>
+; CHECK-NEXT:    [[STRIDED_VEC67:%.*]] = shufflevector <12 x float> [[WIDE_VEC43]], <12 x float> poison, <2 x i32> <i32 1, i32 7>
+; CHECK-NEXT:    [[STRIDED_VEC68:%.*]] = shufflevector <12 x float> [[WIDE_VEC43]], <12 x float> poison, <2 x i32> <i32 2, i32 8>
+; CHECK-NEXT:    [[STRIDED_VEC69:%.*]] = shufflevector <12 x float> [[WIDE_VEC43]], <12 x float> poison, <2 x i32> <i32 3, i32 9>
+; CHECK-NEXT:    [[STRIDED_VEC70:%.*]] = shufflevector <12 x float> [[WIDE_VEC43]], <12 x float> poison, <2 x i32> <i32 4, i32 10>
+; CHECK-NEXT:    [[STRIDED_VEC71:%.*]] = shufflevector <12 x float> [[WIDE_VEC43]], <12 x float> poison, <2 x i32> <i32 5, i32 11>
+; CHECK-NEXT:    [[WIDE_VEC50:%.*]] = load <12 x float>, ptr [[TMP18]], align 8
+; CHECK-NEXT:    [[STRIDED_VEC72:%.*]] = shufflevector <12 x float> [[WIDE_VEC50]], <12 x float> poison, <2 x i32> <i32 0, i32 6>
+; CHECK-NEXT:    [[STRIDED_VEC73:%.*]] = shufflevector <12 x float> [[WIDE_VEC50]], <12 x float> poison, <2 x i32> <i32 1, i32 7>
+; CHECK-NEXT:    [[STRIDED_VEC80:%.*]] = shufflevector <12 x float> [[WIDE_VEC50]], <12 x float> poison, <2 x i32> <i32 2, i32 8>
+; CHECK-NEXT:    [[STRIDED_VEC81:%.*]] = shufflevector <12 x float> [[WIDE_VEC50]], <12 x float> poison, <2 x i32> <i32 3, i32 9>
+; CHECK-NEXT:    [[STRIDED_VEC82:%.*]] = shufflevector <12 x float> [[WIDE_VEC50]], <12 x float> poison, <2 x i32> <i32 4, i32 10>
+; CHECK-NEXT:    [[STRIDED_VEC83:%.*]] = shufflevector <12 x float> [[WIDE_VEC50]], <12 x float> poison, <2 x i32> <i32 5, i32 11>
+; CHECK-NEXT:    [[WIDE_VEC57:%.*]] = load <12 x float>, ptr [[TMP19]], align 8
+; CHECK-NEXT:    [[STRIDED_VEC84:%.*]] = shufflevector <12 x float> [[WIDE_VEC57]], <12 x float> poison, <2 x i32> <i32 0, i32 6>
+; CHECK-NEXT:    [[STRIDED_VEC85:%.*]] = shufflevector <12 x float> [[WIDE_VEC57]], <12 x float> poison, <2 x i32> <i32 1, i32 7>
+; CHECK-NEXT:    [[STRIDED_VEC86:%.*]] = shufflevector <12 x float> [[WIDE_VEC57]], <12 x float> poison, <2 x i32> <i32 2, i32 8>
+; CHECK-NEXT:    [[STRIDED_VEC87:%.*]] = shufflevector <12 x float> [[WIDE_VEC57]], <12 x float> poison, <2 x i32> <i32 3, i32 9>
+; CHECK-NEXT:    [[STRIDED_VEC88:%.*]] = shufflevector <12 x float> [[WIDE_VEC57]], <12 x float> poison, <2 x i32> <i32 4, i32 10>
+; CHECK-NEXT:    [[STRIDED_VEC89:%.*]] = shufflevector <12 x float> [[WIDE_VEC57]], <12 x float> poison, <2 x i32> <i32 5, i32 11>
 ; CHECK-NEXT:    [[TMP64:%.*]] = fmul fast <2 x float> [[STRIDED_VEC36]], [[STRIDED_VEC36]]
 ; CHECK-NEXT:    [[TMP97:%.*]] = fmul fast <2 x float> [[STRIDED_VEC43]], [[STRIDED_VEC43]]
 ; CHECK-NEXT:    [[TMP98:%.*]] = fmul fast <2 x float> [[STRIDED_VEC50]], [[STRIDED_VEC50]]
 ; CHECK-NEXT:    [[TMP99:%.*]] = fmul fast <2 x float> [[STRIDED_VEC57]], [[STRIDED_VEC57]]
+; CHECK-NEXT:    [[TMP100:%.*]] = fmul fast <2 x float> [[STRIDED_VEC42]], [[STRIDED_VEC42]]
+; CHECK-NEXT:    [[TMP101:%.*]] = fmul fast <2 x float> [[STRIDED_VEC66]], [[STRIDED_VEC66]]
+; CHECK-NEXT:    [[TMP102:%.*]] = fmul fast <2 x float> [[STRIDED_VEC72]], [[STRIDED_VEC72]]
+; CHECK-NEXT:    [[TMP103:%.*]] = fmul fast <2 x float> [[STRIDED_VEC84]], [[STRIDED_VEC84]]
 ; CHECK-NEXT:    [[TMP72:%.*]] = fmul fast <2 x float> [[STRIDED_VEC37]], [[STRIDED_VEC37]]
 ; CHECK-NEXT:    [[TMP105:%.*]] = fmul fast <2 x float> [[STRIDED_VEC44]], [[STRIDED_VEC44]]
 ; CHECK-NEXT:    [[TMP106:%.*]] = fmul fast <2 x float> [[STRIDED_VEC51]], [[STRIDED_VEC51]]
 ; CHECK-NEXT:    [[TMP107:%.*]] = fmul fast <2 x float> [[STRIDED_VEC58]], [[STRIDED_VEC58]]
+; CHECK-NEXT:    [[TMP108:%.*]] = fmul fast <2 x float> [[STRIDED_VEC49]], [[STRIDED_VEC49]]
+; CHECK-NEXT:    [[TMP109:%.*]] = fmul fast <2 x float> [[STRIDED_VEC67]], [[STRIDED_VEC67]]
+; CHECK-NEXT:    [[TMP110:%.*]] = fmul fast <2 x float> [[STRIDED_VEC73]], [[STRIDED_VEC73]]
+; CHECK-NEXT:    [[TMP111:%.*]] = fmul fast <2 x float> [[STRIDED_VEC85]], [[STRIDED_VEC85]]
 ; CHECK-NEXT:    [[TMP80:%.*]] = fadd fast <2 x float> [[TMP72]], [[TMP64]]
 ; CHECK-NEXT:    [[TMP113:%.*]] = fadd fast <2 x float> [[TMP105]], [[TMP97]]
 ; CHECK-NEXT:    [[TMP114:%.*]] = fadd fast <2 x float> [[TMP106]], [[TMP98]]
 ; CHECK-NEXT:    [[TMP115:%.*]] = fadd fast <2 x float> [[TMP107]], [[TMP99]]
-; CHECK-NEXT:    [[TMP21:%.*]] = fpext <2 x float> [[TMP80]] to <2 x double>
-; CHECK-NEXT:    [[TMP22:%.*]] = fpext <2 x float> [[TMP113]] to <2 x double>
-; CHECK-NEXT:    [[TMP23:%.*]] = fpext <2 x float> [[TMP114]] to <2 x double>
-; CHECK-NEXT:    [[TMP24:%.*]] = fpext <2 x float> [[TMP115]] to <2 x double>
-; CHECK-NEXT:    [[TMP25:%.*]] = fadd fast <2 x double> [[TMP21]], [[VEC_PHI]]
-; CHECK-NEXT:    [[TMP26:%.*]] = fadd fast <2 x double> [[TMP22]], [[VEC_PHI1]]
-; CHECK-NEXT:    [[TMP27:%.*]] = fadd fast <2 x double> [[TMP23]], [[VEC_PHI2]]
-; CHECK-NEXT:    [[TMP28:%.*]] = fadd fast <2 x double> [[TMP24]], [[VEC_PHI3]]
-; CHECK-NEXT:    [[TMP100:%.*]] = fmul fast <2 x float> [[STRIDED_VEC38]], [[STRIDED_VEC38]]
-; CHECK-NEXT:    [[TMP101:%.*]] = fmul fast <2 x float> [[STRIDED_VEC45]], [[STRIDED_VEC45]]
-; CHECK-NEXT:    [[TMP102:%.*]] = fmul fast <2 x float> [[STRIDED_VEC52]], [[STRIDED_VEC52]]
-; CHECK-NEXT:    [[TMP103:%.*]] = fmul fast <2 x float> [[STRIDED_VEC59]], [[STRIDED_VEC59]]
-; CHECK-NEXT:    [[TMP108:%.*]] = fmul fast <2 x float> [[STRIDED_VEC39]], [[STRIDED_VEC39]]
-; CHECK-NEXT:    [[TMP109:%.*]] = fmul fast <2 x float> [[STRIDED_VEC46]], [[STRIDED_VEC46]]
-; CHECK-NEXT:    [[TMP110:%.*]] = fmul fast <2 x float> [[STRIDED_VEC53]], [[STRIDED_VEC53]]
-; CHECK-NEXT:    [[TMP111:%.*]] = fmul fast <2 x float> [[STRIDED_VEC60]], [[STRIDED_VEC60]]
 ; CHECK-NEXT:    [[TMP116:%.*]] = fadd fast <2 x float> [[TMP108]], [[TMP100]]
 ; CHECK-NEXT:    [[TMP117:%.*]] = fadd fast <2 x float> [[TMP109]], [[TMP101]]
 ; CHECK-NEXT:    [[TMP118:%.*]] = fadd fast <2 x float> [[TMP110]], [[TMP102]]
 ; CHECK-NEXT:    [[TMP119:%.*]] = fadd fast <2 x float> [[TMP111]], [[TMP103]]
+; CHECK-NEXT:    [[TMP40:%.*]] = fpext <2 x float> [[TMP80]] to <2 x double>
+; CHECK-NEXT:    [[TMP52:%.*]] = fpext <2 x float> [[TMP113]] to <2 x double>
+; CHECK-NEXT:    [[TMP53:%.*]] = fpext <2 x float> [[TMP114]] to <2 x double>
+; CHECK-NEXT:    [[TMP54:%.*]] = fpext <2 x float> [[TMP115]] to <2 x double>
 ; CHECK-NEXT:    [[TMP41:%.*]] = fpext <2 x float> [[TMP116]] to <2 x double>
 ; CHECK-NEXT:    [[TMP42:%.*]] = fpext <2 x float> [[TMP117]] to <2 x double>
 ; CHECK-NEXT:    [[TMP43:%.*]] = fpext <2 x float> [[TMP118]] to <2 x double>
 ; CHECK-NEXT:    [[TMP44:%.*]] = fpext <2 x float> [[TMP119]] to <2 x double>
+; CHECK-NEXT:    [[TMP55:%.*]] = fadd fast <2 x double> [[TMP40]], [[VEC_PHI]]
+; CHECK-NEXT:    [[TMP49:%.*]] = fadd fast <2 x double> [[TMP52]], [[VEC_PHI2]]
+; CHECK-NEXT:    [[TMP50:%.*]] = fadd fast <2 x double> [[TMP53]], [[VEC_PHI3]]
+; CHECK-NEXT:    [[TMP51:%.*]] = fadd fast <2 x double> [[TMP54]], [[VEC_PHI4]]
 ; CHECK-NEXT:    [[TMP45:%.*]] = fadd fast <2 x double> [[TMP41]], [[TMP25]]
 ; CHECK-NEXT:    [[TMP46:%.*]] = fadd fast <2 x double> [[TMP42]], [[TMP26]]
 ; CHECK-NEXT:    [[TMP47:%.*]] = fadd fast <2 x double> [[TMP43]], [[TMP27]]
 ; CHECK-NEXT:    [[TMP48:%.*]] = fadd fast <2 x double> [[TMP44]], [[TMP28]]
-; CHECK-NEXT:    [[TMP104:%.*]] = fmul fast <2 x float> [[STRIDED_VEC40]], [[STRIDED_VEC40]]
-; CHECK-NEXT:    [[TMP142:%.*]] = fmul fast <2 x float> [[STRIDED_VEC47]], [[STRIDED_VEC47]]
-; CHECK-NEXT:    [[TMP147:%.*]] = fmul fast <2 x float> [[STRIDED_VEC54]], [[STRIDED_VEC54]]
-; CHECK-NEXT:    [[TMP152:%.*]] = fmul fast <2 x float> [[STRIDED_VEC61]], [[STRIDED_VEC61]]
-; CHECK-NEXT:    [[TMP112:%.*]] = fmul fast <2 x float> [[STRIDED_VEC41]], [[STRIDED_VEC41]]
-; CHECK-NEXT:    [[TMP143:%.*]] = fmul fast <2 x float> [[STRIDED_VEC48]], [[STRIDED_VEC48]]
-; CHECK-NEXT:    [[TMP148:%.*]] = fmul fast <2 x float> [[STRIDED_VEC55]], [[STRIDED_VEC55]]
-; CHECK-NEXT:    [[TMP153:%.*]] = fmul fast <2 x float> [[STRIDED_VEC62]], [[STRIDED_VEC62]]
+; CHECK-NEXT:    [[TMP104:%.*]] = fmul fast <2 x float> [[STRIDED_VEC38]], [[STRIDED_VEC38]]
+; CHECK-NEXT:    [[TMP142:%.*]] = fmul fast <2 x float> [[STRIDED_VEC45]], [[STRIDED_VEC45]]
+; CHECK-NEXT:    [[TMP147:%.*]] = fmul fast <2 x float> [[STRIDED_VEC52]], [[STRIDED_VEC52]]
+; CHECK-NEXT:    [[TMP152:%.*]] = fmul fast <2 x float> [[STRIDED_VEC59]], [[STRIDED_VEC59]]
+; CHECK-NEXT:    [[TMP60:%.*]] = fmul fast <2 x float> [[STRIDED_VEC56]], [[STRIDED_VEC56]]
+; CHECK-NEXT:    [[TMP67:%.*]] = fmul fast <2 x float> [[STRIDED_VEC68]], [[STRIDED_VEC68]]
+; CHECK-NEXT:    [[TMP73:%.*]] = fmul fast <2 x float> [[STRIDED_VEC80]], [[STRIDED_VEC80]]
+; CHECK-NEXT:    [[TMP74:%.*]] = fmul fast <2 x float> [[STRIDED_VEC86]], [[STRIDED_VEC86]]
+; CHECK-NEXT:    [[TMP112:%.*]] = fmul fast <2 x float> [[STRIDED_VEC39]], [[STRIDED_VEC39]]
+; CHECK-NEXT:    [[TMP143:%.*]] = fmul fast <2 x float> [[STRIDED_VEC46]], [[STRIDED_VEC46]]
+; CHECK-NEXT:    [[TMP148:%.*]] = fmul fast <2 x float> [[STRIDED_VEC53]], [[STRIDED_VEC53]]
+; CHECK-NEXT:    [[TMP153:%.*]] = fmul fast <2 x float> [[STRIDED_VEC60]], [[STRIDED_VEC60]]
+; CHECK-NEXT:    [[TMP75:%.*]] = fmul fast <2 x float> [[STRIDED_VEC63]], [[STRIDED_VEC63]]
+; CHECK-NEXT:    [[TMP81:%.*]] = fmul fast <2 x float> [[STRIDED_VEC69]], [[STRIDED_VEC69]]
+; CHECK-NEXT:    [[TMP70:%.*]] = fmul fast <2 x float> [[STRIDED_VEC81]], [[STRIDED_VEC81]]
+; CHECK-NEXT:    [[TMP71:%.*]] = fmul fast <2 x float> [[STRIDED_VEC87]], [[STRIDED_VEC87]]
 ; CHECK-NEXT:    [[TMP120:%.*]] = fadd fast <2 x float> [[TMP112]], [[TMP104]]
 ; CHECK-NEXT:    [[TMP144:%.*]] = fadd fast <2 x float> [[TMP143]], [[TMP142]]
 ; CHECK-NEXT:    [[TMP149:%.*]] = fadd fast <2 x float> [[TMP148]], [[TMP147]]
 ; CHECK-NEXT:    [[TMP154:%.*]] = fadd fast <2 x float> [[TMP153]], [[TMP152]]
+; CHECK-NEXT:    [[TMP76:%.*]] = fadd fast <2 x float> [[TMP75]], [[TMP60]]
+; CHECK-NEXT:    [[TMP77:%.*]] = fadd fast <2 x float> [[TMP81]], [[TMP67]]
+; CHECK-NEXT:    [[TMP78:%.*]] = fadd fast <2 x float> [[TMP70]], [[TMP73]]
+; CHECK-NEXT:    [[TMP79:%.*]] = fadd fast <2 x float> [[TMP71]], [[TMP74]]
 ; CHECK-NEXT:    [[TMP61:%.*]] = fpext <2 x float> [[TMP120]] to <2 x double>
 ; CHECK-NEXT:    [[TMP62:%.*]] = fpext <2 x float> [[TMP144]] to <2 x double>
 ; CHECK-NEXT:    [[TMP63:%.*]] = fpext <2 x float> [[TMP149]] to <2 x double>
 ; CHECK-NEXT:    [[TMP155:%.*]] = fpext <2 x float> [[TMP154]] to <2 x double>
-; CHECK-NEXT:    [[TMP69]] = fadd fast <2 x double> [[TMP61]], [[TMP45]]
-; CHECK-NEXT:    [[TMP65]] = fadd fast <2 x double> [[TMP62]], [[TMP46]]
-; CHECK-NEXT:    [[TMP66]] = fadd fast <2 x double> [[TMP63]], [[TMP47]]
-; CHECK-NEXT:    [[TMP67]] = fadd fast <2 x double> [[TMP155]], [[TMP48]]
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
+; CHECK-NEXT:    [[TMP84:%.*]] = fpext <2 x float> [[TMP76]] to <2 x double>
+; CHECK-NEXT:    [[TMP85:%.*]] = fpext <2 x float> [[TMP77]] to <2 x double>
+; CHECK-NEXT:    [[TMP86:%.*]] = fpext <2 x float> [[TMP78]] to <2 x double>
+; CHECK-NEXT:    [[TMP87:%.*]] = fpext <2 x float> [[TMP79]] to <2 x double>
+; CHECK-NEXT:    [[TMP88:%.*]] = fadd fast <2 x double> [[TMP61]], [[TMP55]]
+; CHECK-NEXT:    [[TMP89:%.*]] = fadd fast <2 x double> [[TMP62]], [[TMP49]]
+; CHECK-NEXT:    [[TMP90:%.*]] = fadd fast <2 x double> [[TMP63]], [[TMP50]]
+; CHECK-NEXT:    [[TMP91:%.*]] = fadd fast <2 x double> [[TMP155]], [[TMP51]]
+; CHECK-NEXT:    [[TMP92:%.*]] = fadd fast <2 x double> [[TMP84]], [[TMP45]]
+; CHECK-NEXT:    [[TMP93:%.*]] = fadd fast <2 x double> [[TMP85]], [[TMP46]]
+; CHECK-NEXT:    [[TMP94:%.*]] = fadd fast <2 x double> [[TMP86]], [[TMP47]]
+; CHECK-NEXT:    [[TMP95:%.*]] = fadd fast <2 x double> [[TMP87]], [[TMP48]]
+; CHECK-NEXT:    [[TMP96:%.*]] = fmul fast <2 x float> [[STRIDED_VEC40]], [[STRIDED_VEC40]]
+; CHECK-NEXT:    [[TMP128:%.*]] = fmul fast <2 x float> [[STRIDED_VEC47]], [[STRIDED_VEC47]]
+; CHECK-NEXT:    [[TMP129:%.*]] = fmul fast <2 x float> [[STRIDED_VEC54]], [[STRIDED_VEC54]]
+; CHECK-NEXT:    [[TMP130:%.*]] = fmul fast <2 x float> [[STRIDED_VEC61]], [[STRIDED_VEC61]]
+; CHECK-NEXT:    [[TMP136:%.*]] = fmul fast <2 x float> [[STRIDED_VEC64]], [[STRIDED_VEC64]]
+; CHECK-NEXT:    [[TMP137:%.*]] = fmul fast <2 x float> [[STRIDED_VEC70]], [[STRIDED_VEC70]]
+; CHECK-NEXT:    [[TMP139:%.*]] = fmul fast <2 x float> [[STRIDED_VEC82]], [[STRIDED_VEC82]]
+; CHECK-NEXT:    [[TMP157:%.*]] = fmul fast <2 x float> [[STRIDED_VEC88]], [[STRIDED_VEC88]]
+; CHECK-NEXT:    [[TMP159:%.*]] = fmul fast <2 x float> [[STRIDED_VEC41]], [[STRIDED_VEC41]]
+; CHECK-NEXT:    [[TMP160:%.*]] = fmul fast <2 x float> [[STRIDED_VEC48]], [[STRIDED_VEC48]]
+; CHECK-NEXT:    [[TMP161:%.*]] = fmul fast <2 x float> [[STRIDED_VEC55]], [[STRIDED_VEC55]]
+; CHECK-NEXT:    [[TMP162:%.*]] = fmul fast <2 x float> [[STRIDED_VEC62]], [[STRIDED_VEC62]]
+; CHECK-NEXT:    [[TMP163:%.*]] = fmul fast <2 x float> [[STRIDED_VEC65]], [[STRIDED_VEC65]]
+; CHECK-NEXT:    [[TMP164:%.*]] = fmul fast <2 x float> [[STRIDED_VEC71]], [[STRIDED_VEC71]]
+; CHECK-NEXT:    [[TMP165:%.*]] = fmul fast <2 x float> [[STRIDED_VEC83]], [[STRIDED_VEC83]]
+; CHECK-NEXT:    [[TMP166:%.*]] = fmul fast <2 x float> [[STRIDED_VEC89]], [[STRIDED_VEC89]]
+; CHECK-NEXT:    [[TMP167:%.*]] = fadd fast <2 x float> [[TMP159]], [[TMP96]]
+; CHECK-NEXT:    [[TMP168:%.*]] = fadd fast <2 x float> [[TMP160]], [[TMP128]]
+; CHECK-NEXT:    [[TMP169:%.*]] = fadd fast <2 x float> [[TMP161]], [[TMP129]]
+; CHECK-NEXT:    [[TMP170:%.*]] = fadd fast <2 x float> [[TMP162]], [[TMP130]]
+; CHECK-NEXT:    [[TMP171:%.*]] = fadd fast <2 x float> [[TMP163]], [[TMP136]]
+; CHECK-NEXT:    [[TMP172:%.*]] = fadd fast <2 x float> [[TMP164]], [[TMP137]]
+; CHECK-NEXT:    [[TMP173:%.*]] = fadd fast <2 x float> [[TMP165]], [[TMP139]]
+; CHECK-NEXT:    [[TMP174:%.*]] = fadd fast <2 x float> [[TMP166]], [[TMP157]]
+; CHECK-NEXT:    [[TMP175:%.*]] = fpext <2 x float> [[TMP167]] to <2 x double>
+; CHECK-NEXT:    [[TMP121:%.*]] = fpext <2 x float> [[TMP168]] to <2 x double>
+; CHECK-NEXT:    [[TMP122:%.*]] = fpext <2 x float> [[TMP169]] to <2 x double>
+; CHECK-NEXT:    [[TMP123:%.*]] = fpext <2 x float> [[TMP170]] to <2 x double>
+; CHECK-NEXT:    [[TMP124:%.*]] = fpext <2 x float> [[TMP171]] to <2 x double>
+; CHECK-NEXT:    [[TMP125:%.*]] = fpext <2 x float> [[TMP172]] to <2 x double>
+; CHECK-NEXT:    [[TMP126:%.*]] = fpext <2 x float> [[TMP173]] to <2 x double>
+; CHECK-NEXT:    [[TMP127:%.*]] = fpext <2 x float> [[TMP174]] to <2 x double>
+; CHECK-NEXT:    [[TMP69]] = fadd fast <2 x double> [[TMP175]], [[TMP88]]
+; CHECK-NEXT:    [[TMP65]] = fadd fast <2 x double> [[TMP121]], [[TMP89]]
+; CHECK-NEXT:    [[TMP66]] = fadd fast <2 x double> [[TMP122]], [[TMP90]]
+; CHECK-NEXT:    [[TMP131]] = fadd fast <2 x double> [[TMP123]], [[TMP91]]
+; CHECK-NEXT:    [[TMP132]] = fadd fast <2 x double> [[TMP124]], [[TMP92]]
+; CHECK-NEXT:    [[TMP133]] = fadd fast <2 x double> [[TMP125]], [[TMP93]]
+; CHECK-NEXT:    [[TMP134]] = fadd fast <2 x double> [[TMP126]], [[TMP94]]
+; CHECK-NEXT:    [[TMP135]] = fadd fast <2 x double> [[TMP127]], [[TMP95]]
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
 ; CHECK-NEXT:    [[TMP68:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[TMP68]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       [[MIDDLE_BLOCK]]:
 ; CHECK-NEXT:    [[BIN_RDX:%.*]] = fadd fast <2 x double> [[TMP65]], [[TMP69]]
 ; CHECK-NEXT:    [[BIN_RDX30:%.*]] = fadd fast <2 x double> [[TMP66]], [[BIN_RDX]]
-; CHECK-NEXT:    [[TMP156:%.*]] = fadd fast <2 x double> [[TMP67]], [[BIN_RDX30]]
+; CHECK-NEXT:    [[BIN_RDX64:%.*]] = fadd fast <2 x double> [[TMP131]], [[BIN_RDX30]]
+; CHECK-NEXT:    [[BIN_RDX65:%.*]] = fadd fast <2 x double> [[TMP132]], [[BIN_RDX64]]
+; CHECK-NEXT:    [[BIN_RDX66:%.*]] = fadd fast <2 x double> [[TMP133]], [[BIN_RDX65]]
+; CHECK-NEXT:    [[BIN_RDX67:%.*]] = fadd fast <2 x double> [[TMP134]], [[BIN_RDX66]]
+; CHECK-NEXT:    [[TMP156:%.*]] = fadd fast <2 x double> [[TMP135]], [[BIN_RDX67]]
 ; CHECK-NEXT:    [[TMP158:%.*]] = call fast double @llvm.vector.reduce.fadd.v2f64(double 0.000000e+00, <2 x double> [[TMP156]])
 ; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[CMP_N]], label %[[FOR_COND_FOR_END13_CRIT_EDGE:.*]], label %[[SCALAR_PH]]
-; CHECK:       [[SCALAR_PH]]:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[FOR_COND1_PREHEADER_PREHEADER]] ]
-; CHECK-NEXT:    [[BC_MERGE_RDX:%.*]] = phi double [ [[TMP158]], %[[MIDDLE_BLOCK]] ], [ 0.000000e+00, %[[FOR_COND1_PREHEADER_PREHEADER]] ]
+; CHECK-NEXT:    br i1 [[CMP_N]], label %[[FOR_COND_FOR_END13_CRIT_EDGE:.*]], label %[[VEC_EPILOG_ITER_CHECK:.*]]
+; CHECK:       [[VEC_EPILOG_ITER_CHECK]]:
+; CHECK-NEXT:    [[N_VEC_REMAINING:%.*]] = sub i64 [[TMP0]], [[N_VEC]]
+; CHECK-NEXT:    [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], 2
+; CHECK-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH]], label %[[VEC_EPILOG_PH]]
+; CHECK:       [[VEC_EPILOG_PH]]:
+; CHECK-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; CHECK-NEXT:    [[BC_MERGE_RDX:%.*]] = phi double [ [[TMP158]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0.000000e+00, %[[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; CHECK-NEXT:    [[N_MOD_VF69:%.*]] = urem i64 [[TMP0]], 2
+; CHECK-NEXT:    [[N_VEC70:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF69]]
+; CHECK-NEXT:    [[TMP138:%.*]] = insertelement <2 x double> zeroinitializer, double [[BC_MERGE_RDX]], i32 0
+; CHECK-NEXT:    br label %[[VEC_EPILOG_VECTOR_BODY:.*]]
+; CHECK:       [[VEC_EPILOG_VECTOR_BODY]]:
+; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT80:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI72:%.*]] = phi <2 x double> [ [[TMP138]], %[[VEC_EPILOG_PH]] ], [ [[TMP176:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT:    [[ARRAYIDX5_REALP:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV]], i64 0, i32 0
+; CHECK-NEXT:    [[WIDE_VEC73:%.*]] = load <12 x float>, ptr [[ARRAYIDX5_REALP]], align 8
+; CHECK-NEXT:    [[STRIDED_VEC74:%.*]] = shufflevector <12 x float> [[WIDE_VEC73]], <12 x float> poison, <2 x i32> <i32 0, i32 6>
+; CHECK-NEXT:    [[STRIDED_VEC75:%.*]] = shufflevector <12 x float> [[WIDE_VEC73]], <12 x float> poison, <2 x i32> <i32 1, i32 7>
+; CHECK-NEXT:    [[STRIDED_VEC76:%.*]] = shufflevector <12 x float> [[WIDE_VEC73]], <12 x float> poison, <2 x i32> <i32 2, i32 8>
+; CHECK-NEXT:    [[STRIDED_VEC77:%.*]] = shufflevector <12 x float> [[WIDE_VEC73]], <12 x float> poison, <2 x i32> <i32 3, i32 9>
+; CHECK-NEXT:    [[STRIDED_VEC78:%.*]] = shufflevector <12 x float> [[WIDE_VEC73]], <12 x float> poison, <2 x i32> <i32 4, i32 10>
+; CHECK-NEXT:    [[STRIDED_VEC79:%.*]] = shufflevector <12 x float> [[WIDE_VEC73]], <12 x float> poison, <2 x i32> <i32 5, i32 11>
+; CHECK-NEXT:    [[TMP140:%.*]] = fmul fast <2 x float> [[STRIDED_VEC74]], [[STRIDED_VEC74]]
+; CHECK-NEXT:    [[TMP141:%.*]] = fmul fast <2 x float> [[STRIDED_VEC75]], [[STRIDED_VEC75]]
+; CHECK-NEXT:    [[TMP177:%.*]] = fadd fast <2 x float> [[TMP141]], [[TMP140]]
+; CHECK-NEXT:    [[TMP178:%.*]] = fpext <2 x float> [[TMP177]] to <2 x double>
+; CHECK-NEXT:    [[TMP179:%.*]] = fadd fast <2 x double> [[TMP178]], [[VEC_PHI72]]
+; CHECK-NEXT:    [[TMP145:%.*]] = fmul fast <2 x float> [[STRIDED_VEC76]], [[STRIDED_VEC76]]
+; CHECK-NEXT:    [[TMP146:%.*]] = fmul fast <2 x float> [[STRIDED_VEC77]], [[STRIDED_VEC77]]
+; CHECK-NEXT:    [[TMP180:%.*]] = fadd fast <2 x float> [[TMP146]], [[TMP145]]
+; CHECK-NEXT:    [[TMP181:%.*]] = fpext <2 x float> [[TMP180]] to <2 x double>
+; CHECK-NEXT:    [[TMP182:%.*]] = fadd fast <2 x double> [[TMP181]], [[TMP179]]
+; CHECK-NEXT:    [[TMP150:%.*]] = fmul fast <2 x float> [[STRIDED_VEC78]], [[STRIDED_VEC78]]
+; CHECK-NEXT:    [[TMP151:%.*]] = fmul fast <2 x float> [[STRIDED_VEC79]], [[STRIDED_VEC79]]
+; CHECK-NEXT:    [[TMP183:%.*]] = fadd fast <2 x float> [[TMP151]], [[TMP150]]
+; CHECK-NEXT:    [[TMP184:%.*]] = fpext <2 x float> [[TMP183]] to <2 x double>
+; CHECK-NEXT:    [[TMP176]] = fadd fast <2 x double> [[TMP184]], [[TMP182]]
+; CHECK-NEXT:    [[INDEX_NEXT80]] = add nuw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT:    [[TMP185:%.*]] = icmp eq i64 [[INDEX_NEXT80]], [[N_VEC70]]
+; CHECK-NEXT:    br i1 [[TMP185]], label %[[VEC_EPILOG_MIDDLE_BLOCK:.*]], label %[[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK:       [[VEC_EPILOG_MIDDLE_BLOCK]]:
+; CHECK-NEXT:    [[TMP186:%.*]] = call fast double @llvm.vector.reduce.fadd.v2f64(double 0.000000e+00, <2 x double> [[TMP176]])
+; CHECK-NEXT:    [[CMP_N81:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC70]]
+; CHECK-NEXT:    br i1 [[CMP_N81]], label %[[FOR_COND_FOR_END13_CRIT_EDGE]], label %[[VEC_EPILOG_SCALAR_PH]]
+; CHECK:       [[VEC_EPILOG_SCALAR_PH]]:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC70]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[ITER_CHECK]] ]
+; CHECK-NEXT:    [[BC_MERGE_RDX82:%.*]] = phi double [ [[TMP186]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[TMP158]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0.000000e+00, %[[ITER_CHECK]] ]
 ; CHECK-NEXT:    br label %[[FOR_COND1_PREHEADER:.*]]
 ; CHECK:       [[FOR_COND1_PREHEADER]]:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_COND1_PREHEADER]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
-; CHECK-NEXT:    [[SUM_026:%.*]] = phi double [ [[ADD10_2:%.*]], %[[FOR_COND1_PREHEADER]] ], [ [[BC_MERGE_RDX]], %[[SCALAR_PH]] ]
-; CHECK-NEXT:    [[ARRAYIDX5_REALP:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV]], i64 0, i32 0
-; CHECK-NEXT:    [[ARRAYIDX5_REAL:%.*]] = load float, ptr [[ARRAYIDX5_REALP]], align 8
-; CHECK-NEXT:    [[ARRAYIDX5_IMAGP:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV]], i64 0, i32 1
+; CHECK-NEXT:    [[INDVARS_IV1:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_COND1_PREHEADER]] ], [ [[BC_RESUME_VAL]], %[[VEC_EPILOG_SCALAR_PH]] ]
+; CHECK-NEXT:    [[SUM_026:%.*]] = phi double [ [[ADD10_2:%.*]], %[[FOR_COND1_PREHEADER]] ], [ [[BC_MERGE_RDX82]], %[[VEC_EPILOG_SCALAR_PH]] ]
+; CHECK-NEXT:    [[ARRAYIDX5_REALP1:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV1]], i64 0, i32 0
+; CHECK-NEXT:    [[ARRAYIDX5_REAL:%.*]] = load float, ptr [[ARRAYIDX5_REALP1]], align 8
+; CHECK-NEXT:    [[ARRAYIDX5_IMAGP:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV1]], i64 0, i32 1
 ; CHECK-NEXT:    [[ARRAYIDX5_IMAG:%.*]] = load float, ptr [[ARRAYIDX5_IMAGP]], align 8
 ; CHECK-NEXT:    [[MUL:%.*]] = fmul fast float [[ARRAYIDX5_REAL]], [[ARRAYIDX5_REAL]]
 ; CHECK-NEXT:    [[MUL9:%.*]] = fmul fast float [[ARRAYIDX5_IMAG]], [[ARRAYIDX5_IMAG]]
 ; CHECK-NEXT:    [[ADD:%.*]] = fadd fast float [[MUL9]], [[MUL]]
 ; CHECK-NEXT:    [[CONV:%.*]] = fpext float [[ADD]] to double
 ; CHECK-NEXT:    [[ADD10:%.*]] = fadd fast double [[CONV]], [[SUM_026]]
-; CHECK-NEXT:    [[ARRAYIDX5_REALP_1:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV]], i64 1, i32 0
+; CHECK-NEXT:    [[ARRAYIDX5_REALP_1:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV1]], i64 1, i32 0
 ; CHECK-NEXT:    [[ARRAYIDX5_REAL_1:%.*]] = load float, ptr [[ARRAYIDX5_REALP_1]], align 8
-; CHECK-NEXT:    [[ARRAYIDX5_IMAGP_1:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV]], i64 1, i32 1
+; CHECK-NEXT:    [[ARRAYIDX5_IMAGP_1:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV1]], i64 1, i32 1
 ; CHECK-NEXT:    [[ARRAYIDX5_IMAG_1:%.*]] = load float, ptr [[ARRAYIDX5_IMAGP_1]], align 8
 ; CHECK-NEXT:    [[MUL_1:%.*]] = fmul fast float [[ARRAYIDX5_REAL_1]], [[ARRAYIDX5_REAL_1]]
 ; CHECK-NEXT:    [[MUL9_1:%.*]] = fmul fast float [[ARRAYIDX5_IMAG_1]], [[ARRAYIDX5_IMAG_1]]
 ; CHECK-NEXT:    [[ADD_1:%.*]] = fadd fast float [[MUL9_1]], [[MUL_1]]
 ; CHECK-NEXT:    [[CONV_1:%.*]] = fpext float [[ADD_1]] to double
 ; CHECK-NEXT:    [[ADD10_1:%.*]] = fadd fast double [[CONV_1]], [[ADD10]]
-; CHECK-NEXT:    [[ARRAYIDX5_REALP_2:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV]], i64 2, i32 0
+; CHECK-NEXT:    [[ARRAYIDX5_REALP_2:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV1]], i64 2, i32 0
 ; CHECK-NEXT:    [[ARRAYIDX5_REAL_2:%.*]] = load float, ptr [[ARRAYIDX5_REALP_2]], align 8
-; CHECK-NEXT:    [[ARRAYIDX5_IMAGP_2:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV]], i64 2, i32 1
+; CHECK-NEXT:    [[ARRAYIDX5_IMAGP_2:%.*]] = getelementptr inbounds [3 x { float, float }], ptr [[A]], i64 [[INDVARS_IV1]], i64 2, i32 1
 ; CHECK-NEXT:    [[ARRAYIDX5_IMAG_2:%.*]] = load float, ptr [[ARRAYIDX5_IMAGP_2]], align 8
 ; CHECK-NEXT:    [[MUL_2:%.*]] = fmul fast float [[ARRAYIDX5_REAL_2]], [[ARRAYIDX5_REAL_2]]
 ; CHECK-NEXT:    [[MUL9_2:%.*]] = fmul fast float [[ARRAYIDX5_IMAG_2]], [[ARRAYIDX5_IMAG_2]]
 ; CHECK-NEXT:    [[ADD_2:%.*]] = fadd fast float [[MUL9_2]], [[MUL_2]]
 ; CHECK-NEXT:    [[CONV_2:%.*]] = fpext float [[ADD_2]] to double
 ; CHECK-NEXT:    [[ADD10_2]] = fadd fast double [[CONV_2]], [[ADD10_1]]
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
+; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV1]], 1
 ; CHECK-NEXT:    [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
 ; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label %[[FOR_COND_FOR_END13_CRIT_EDGE]], label %[[FOR_COND1_PREHEADER]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label %[[FOR_COND_FOR_END13_CRIT_EDGE]], label %[[FOR_COND1_PREHEADER]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       [[FOR_COND_FOR_END13_CRIT_EDGE]]:
-; CHECK-NEXT:    [[ADD10_2_LCSSA:%.*]] = phi double [ [[ADD10_2]], %[[FOR_COND1_PREHEADER]] ], [ [[TMP158]], %[[MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    [[ADD10_2_LCSSA:%.*]] = phi double [ [[ADD10_2]], %[[FOR_COND1_PREHEADER]] ], [ [[TMP158]], %[[MIDDLE_BLOCK]] ], [ [[TMP186]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ]
 ; CHECK-NEXT:    [[PHITMP:%.*]] = fptrunc double [[ADD10_2_LCSSA]] to float
 ; CHECK-NEXT:    br label %[[FOR_END13]]
 ; CHECK:       [[FOR_END13]]:
@@ -234,5 +385,6 @@ for.end13:                                        ; preds = %for.cond.for.end13_
 ; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
 ; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
 ; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
-; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META2]], [[META1]]}
+; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META1]], [[META2]]}
+; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META2]], [[META1]]}
 ;.
diff --git a/llvm/test/Transforms/LoopVectorize/X86/pr47437.ll b/llvm/test/Transforms/LoopVectorize/X86/pr47437.ll
index d2f8f2203b724..5d16ce5346bbf 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/pr47437.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/pr47437.ll
@@ -169,64 +169,140 @@ define void @test_muladd(ptr noalias nocapture %d1, ptr noalias nocapture readon
 ; AVX1-NEXT:  entry:
 ; AVX1-NEXT:    [[CMP30:%.*]] = icmp sgt i32 [[N:%.*]], 0
 ; AVX1-NEXT:    br i1 [[CMP30]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
-; AVX1:       for.body.preheader:
+; AVX1:       iter.check:
 ; AVX1-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[N]] to i64
-; AVX1-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 8
+; AVX1-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 4
 ; AVX1-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; AVX1:       vector.main.loop.iter.check:
+; AVX1-NEXT:    [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 16
+; AVX1-NEXT:    br i1 [[MIN_ITERS_CHECK1]], label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH1:%.*]]
 ; AVX1:       vector.ph:
-; AVX1-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 8
+; AVX1-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 16
 ; AVX1-NEXT:    [[N_VEC:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_MOD_VF]]
 ; AVX1-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; AVX1:       vector.body:
-; AVX1-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX1-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH1]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; AVX1-NEXT:    [[TMP1:%.*]] = add i64 [[INDEX]], 4
+; AVX1-NEXT:    [[TMP3:%.*]] = add i64 [[INDEX]], 8
+; AVX1-NEXT:    [[TMP8:%.*]] = add i64 [[INDEX]], 12
 ; AVX1-NEXT:    [[TMP7:%.*]] = shl nuw nsw i64 [[INDEX]], 1
 ; AVX1-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 1
+; AVX1-NEXT:    [[TMP5:%.*]] = shl nuw nsw i64 [[TMP3]], 1
+; AVX1-NEXT:    [[TMP6:%.*]] = shl nuw nsw i64 [[TMP8]], 1
 ; AVX1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i16, ptr [[S1:%.*]], i64 [[TMP7]]
 ; AVX1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP2]]
+; AVX1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP5]]
+; AVX1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP6]]
 ; AVX1-NEXT:    [[WIDE_VEC2:%.*]] = load <8 x i16>, ptr [[TMP11]], align 2
 ; AVX1-NEXT:    [[STRIDED_VEC5:%.*]] = shufflevector <8 x i16> [[WIDE_VEC2]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
 ; AVX1-NEXT:    [[STRIDED_VEC9:%.*]] = shufflevector <8 x i16> [[WIDE_VEC2]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
 ; AVX1-NEXT:    [[WIDE_VEC3:%.*]] = load <8 x i16>, ptr [[TMP4]], align 2
 ; AVX1-NEXT:    [[STRIDED_VEC6:%.*]] = shufflevector <8 x i16> [[WIDE_VEC3]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
 ; AVX1-NEXT:    [[STRIDED_VEC10:%.*]] = shufflevector <8 x i16> [[WIDE_VEC3]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; AVX1-NEXT:    [[WIDE_VEC6:%.*]] = load <8 x i16>, ptr [[TMP9]], align 2
+; AVX1-NEXT:    [[STRIDED_VEC7:%.*]] = shufflevector <8 x i16> [[WIDE_VEC6]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; AVX1-NEXT:    [[STRIDED_VEC8:%.*]] = shufflevector <8 x i16> [[WIDE_VEC6]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; AVX1-NEXT:    [[WIDE_VEC9:%.*]] = load <8 x i16>, ptr [[TMP10]], align 2
+; AVX1-NEXT:    [[STRIDED_VEC12:%.*]] = shufflevector <8 x i16> [[WIDE_VEC9]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; AVX1-NEXT:    [[STRIDED_VEC11:%.*]] = shufflevector <8 x i16> [[WIDE_VEC9]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
 ; AVX1-NEXT:    [[TMP36:%.*]] = sext <4 x i16> [[STRIDED_VEC5]] to <4 x i32>
 ; AVX1-NEXT:    [[TMP37:%.*]] = sext <4 x i16> [[STRIDED_VEC6]] to <4 x i32>
+; AVX1-NEXT:    [[TMP38:%.*]] = sext <4 x i16> [[STRIDED_VEC7]] to <4 x i32>
+; AVX1-NEXT:    [[TMP39:%.*]] = sext <4 x i16> [[STRIDED_VEC12]] to <4 x i32>
 ; AVX1-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i16, ptr [[S2:%.*]], i64 [[TMP7]]
 ; AVX1-NEXT:    [[TMP23:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP2]]
+; AVX1-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP5]]
+; AVX1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP6]]
 ; AVX1-NEXT:    [[WIDE_VEC13:%.*]] = load <8 x i16>, ptr [[TMP22]], align 2
 ; AVX1-NEXT:    [[STRIDED_VEC17:%.*]] = shufflevector <8 x i16> [[WIDE_VEC13]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
 ; AVX1-NEXT:    [[STRIDED_VEC21:%.*]] = shufflevector <8 x i16> [[WIDE_VEC13]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
 ; AVX1-NEXT:    [[WIDE_VEC14:%.*]] = load <8 x i16>, ptr [[TMP23]], align 2
 ; AVX1-NEXT:    [[STRIDED_VEC18:%.*]] = shufflevector <8 x i16> [[WIDE_VEC14]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
 ; AVX1-NEXT:    [[STRIDED_VEC22:%.*]] = shufflevector <8 x i16> [[WIDE_VEC14]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; AVX1-NEXT:    [[WIDE_VEC18:%.*]] = load <8 x i16>, ptr [[TMP17]], align 2
+; AVX1-NEXT:    [[STRIDED_VEC19:%.*]] = shufflevector <8 x i16> [[WIDE_VEC18]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; AVX1-NEXT:    [[STRIDED_VEC20:%.*]] = shufflevector <8 x i16> [[WIDE_VEC18]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; AVX1-NEXT:    [[WIDE_VEC21:%.*]] = load <8 x i16>, ptr [[TMP18]], align 2
+; AVX1-NEXT:    [[STRIDED_VEC24:%.*]] = shufflevector <8 x i16> [[WIDE_VEC21]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; AVX1-NEXT:    [[STRIDED_VEC23:%.*]] = shufflevector <8 x i16> [[WIDE_VEC21]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
 ; AVX1-NEXT:    [[TMP40:%.*]] = sext <4 x i16> [[STRIDED_VEC17]] to <4 x i32>
 ; AVX1-NEXT:    [[TMP41:%.*]] = sext <4 x i16> [[STRIDED_VEC18]] to <4 x i32>
+; AVX1-NEXT:    [[TMP42:%.*]] = sext <4 x i16> [[STRIDED_VEC19]] to <4 x i32>
+; AVX1-NEXT:    [[TMP43:%.*]] = sext <4 x i16> [[STRIDED_VEC24]] to <4 x i32>
 ; AVX1-NEXT:    [[TMP44:%.*]] = mul nsw <4 x i32> [[TMP40]], [[TMP36]]
 ; AVX1-NEXT:    [[TMP45:%.*]] = mul nsw <4 x i32> [[TMP41]], [[TMP37]]
-; AVX1-NEXT:    [[TMP38:%.*]] = sext <4 x i16> [[STRIDED_VEC9]] to <4 x i32>
-; AVX1-NEXT:    [[TMP39:%.*]] = sext <4 x i16> [[STRIDED_VEC10]] to <4 x i32>
-; AVX1-NEXT:    [[TMP42:%.*]] = sext <4 x i16> [[STRIDED_VEC21]] to <4 x i32>
-; AVX1-NEXT:    [[TMP43:%.*]] = sext <4 x i16> [[STRIDED_VEC22]] to <4 x i32>
 ; AVX1-NEXT:    [[TMP46:%.*]] = mul nsw <4 x i32> [[TMP42]], [[TMP38]]
 ; AVX1-NEXT:    [[TMP47:%.*]] = mul nsw <4 x i32> [[TMP43]], [[TMP39]]
-; AVX1-NEXT:    [[TMP19:%.*]] = add nsw <4 x i32> [[TMP46]], [[TMP44]]
-; AVX1-NEXT:    [[TMP20:%.*]] = add nsw <4 x i32> [[TMP47]], [[TMP45]]
+; AVX1-NEXT:    [[TMP27:%.*]] = sext <4 x i16> [[STRIDED_VEC9]] to <4 x i32>
+; AVX1-NEXT:    [[TMP28:%.*]] = sext <4 x i16> [[STRIDED_VEC10]] to <4 x i32>
+; AVX1-NEXT:    [[TMP29:%.*]] = sext <4 x i16> [[STRIDED_VEC8]] to <4 x i32>
+; AVX1-NEXT:    [[TMP30:%.*]] = sext <4 x i16> [[STRIDED_VEC11]] to <4 x i32>
+; AVX1-NEXT:    [[TMP31:%.*]] = sext <4 x i16> [[STRIDED_VEC21]] to <4 x i32>
+; AVX1-NEXT:    [[TMP32:%.*]] = sext <4 x i16> [[STRIDED_VEC22]] to <4 x i32>
+; AVX1-NEXT:    [[TMP33:%.*]] = sext <4 x i16> [[STRIDED_VEC20]] to <4 x i32>
+; AVX1-NEXT:    [[TMP34:%.*]] = sext <4 x i16> [[STRIDED_VEC23]] to <4 x i32>
+; AVX1-NEXT:    [[TMP35:%.*]] = mul nsw <4 x i32> [[TMP31]], [[TMP27]]
+; AVX1-NEXT:    [[TMP60:%.*]] = mul nsw <4 x i32> [[TMP32]], [[TMP28]]
+; AVX1-NEXT:    [[TMP67:%.*]] = mul nsw <4 x i32> [[TMP33]], [[TMP29]]
+; AVX1-NEXT:    [[TMP68:%.*]] = mul nsw <4 x i32> [[TMP34]], [[TMP30]]
+; AVX1-NEXT:    [[TMP19:%.*]] = add nsw <4 x i32> [[TMP35]], [[TMP44]]
+; AVX1-NEXT:    [[TMP20:%.*]] = add nsw <4 x i32> [[TMP60]], [[TMP45]]
+; AVX1-NEXT:    [[TMP69:%.*]] = add nsw <4 x i32> [[TMP67]], [[TMP46]]
+; AVX1-NEXT:    [[TMP70:%.*]] = add nsw <4 x i32> [[TMP68]], [[TMP47]]
 ; AVX1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[D1:%.*]], i64 [[INDEX]]
 ; AVX1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i32 4
+; AVX1-NEXT:    [[TMP71:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i32 8
+; AVX1-NEXT:    [[TMP72:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i32 12
 ; AVX1-NEXT:    store <4 x i32> [[TMP19]], ptr [[TMP21]], align 4
 ; AVX1-NEXT:    store <4 x i32> [[TMP20]], ptr [[TMP26]], align 4
-; AVX1-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
+; AVX1-NEXT:    store <4 x i32> [[TMP69]], ptr [[TMP71]], align 4
+; AVX1-NEXT:    store <4 x i32> [[TMP70]], ptr [[TMP72]], align 4
+; AVX1-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
 ; AVX1-NEXT:    [[TMP24:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; AVX1-NEXT:    br i1 [[TMP24]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; AVX1:       middle.block:
 ; AVX1-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[WIDE_TRIP_COUNT]], [[N_VEC]]
-; AVX1-NEXT:    br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[SCALAR_PH]]
-; AVX1:       scalar.ph:
-; AVX1-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
+; AVX1-NEXT:    br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
+; AVX1:       vec.epilog.iter.check:
+; AVX1-NEXT:    [[N_VEC_REMAINING:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_VEC]]
+; AVX1-NEXT:    [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], 4
+; AVX1-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[SCALAR_PH]], label [[VEC_EPILOG_PH]]
+; AVX1:       vec.epilog.ph:
+; AVX1-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_PH]] ]
+; AVX1-NEXT:    [[N_MOD_VF24:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 4
+; AVX1-NEXT:    [[N_VEC25:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_MOD_VF24]]
 ; AVX1-NEXT:    br label [[FOR_BODY1:%.*]]
+; AVX1:       vec.epilog.vector.body:
+; AVX1-NEXT:    [[INDEX26:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT33:%.*]], [[FOR_BODY1]] ]
+; AVX1-NEXT:    [[TMP48:%.*]] = shl nuw nsw i64 [[INDEX26]], 1
+; AVX1-NEXT:    [[TMP49:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP48]]
+; AVX1-NEXT:    [[WIDE_VEC27:%.*]] = load <8 x i16>, ptr [[TMP49]], align 2
+; AVX1-NEXT:    [[STRIDED_VEC28:%.*]] = shufflevector <8 x i16> [[WIDE_VEC27]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; AVX1-NEXT:    [[STRIDED_VEC29:%.*]] = shufflevector <8 x i16> [[WIDE_VEC27]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; AVX1-NEXT:    [[TMP50:%.*]] = sext <4 x i16> [[STRIDED_VEC28]] to <4 x i32>
+; AVX1-NEXT:    [[TMP51:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP48]]
+; AVX1-NEXT:    [[WIDE_VEC30:%.*]] = load <8 x i16>, ptr [[TMP51]], align 2
+; AVX1-NEXT:    [[STRIDED_VEC31:%.*]] = shufflevector <8 x i16> [[WIDE_VEC30]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; AVX1-NEXT:    [[STRIDED_VEC32:%.*]] = shufflevector <8 x i16> [[WIDE_VEC30]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; AVX1-NEXT:    [[TMP52:%.*]] = sext <4 x i16> [[STRIDED_VEC31]] to <4 x i32>
+; AVX1-NEXT:    [[TMP53:%.*]] = mul nsw <4 x i32> [[TMP52]], [[TMP50]]
+; AVX1-NEXT:    [[TMP54:%.*]] = sext <4 x i16> [[STRIDED_VEC29]] to <4 x i32>
+; AVX1-NEXT:    [[TMP55:%.*]] = sext <4 x i16> [[STRIDED_VEC32]] to <4 x i32>
+; AVX1-NEXT:    [[TMP56:%.*]] = mul nsw <4 x i32> [[TMP55]], [[TMP54]]
+; AVX1-NEXT:    [[TMP57:%.*]] = add nsw <4 x i32> [[TMP56]], [[TMP53]]
+; AVX1-NEXT:    [[TMP58:%.*]] = getelementptr inbounds i32, ptr [[D1]], i64 [[INDEX26]]
+; AVX1-NEXT:    store <4 x i32> [[TMP57]], ptr [[TMP58]], align 4
+; AVX1-NEXT:    [[INDEX_NEXT33]] = add nuw i64 [[INDEX26]], 4
+; AVX1-NEXT:    [[TMP59:%.*]] = icmp eq i64 [[INDEX_NEXT33]], [[N_VEC25]]
+; AVX1-NEXT:    br i1 [[TMP59]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[FOR_BODY1]], !llvm.loop [[LOOP3:![0-9]+]]
+; AVX1:       vec.epilog.middle.block:
+; AVX1-NEXT:    [[CMP_N34:%.*]] = icmp eq i64 [[WIDE_TRIP_COUNT]], [[N_VEC25]]
+; AVX1-NEXT:    br i1 [[CMP_N34]], label [[FOR_END_LOOPEXIT]], label [[SCALAR_PH]]
+; AVX1:       vec.epilog.scalar.ph:
+; AVX1-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC25]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
+; AVX1-NEXT:    br label [[FOR_BODY:%.*]]
 ; AVX1:       for.body:
-; AVX1-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY1]] ]
+; AVX1-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
 ; AVX1-NEXT:    [[TMP61:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
 ; AVX1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP61]]
 ; AVX1-NEXT:    [[TMP62:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
@@ -248,7 +324,7 @@ define void @test_muladd(ptr noalias nocapture %d1, ptr noalias nocapture readon
 ; AVX1-NEXT:    store i32 [[ADD18]], ptr [[ARRAYIDX20]], align 4
 ; AVX1-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
 ; AVX1-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
-; AVX1-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_END_LOOPEXIT]], label [[FOR_BODY1]], !llvm.loop [[LOOP3:![0-9]+]]
+; AVX1-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_END_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; AVX1:       for.end.loopexit:
 ; AVX1-NEXT:    br label [[FOR_END]]
 ; AVX1:       for.end:

>From 7c900b2ed295e832bdf5ba81debfab0f768a0aea Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Fri, 29 Aug 2025 11:38:46 +0100
Subject: [PATCH 2/3] Don't reset canonical IV start value

---
 .../Transforms/Vectorize/LoopVectorize.cpp    | 166 +++++++++---------
 .../Transforms/Vectorize/VPlanPatternMatch.h  |   6 +
 .../Transforms/Vectorize/VPlanTransforms.cpp  |  11 ++
 3 files changed, 101 insertions(+), 82 deletions(-)

diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index 1bfd70ee53c7c..d09e54dc4212c 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -9719,26 +9719,22 @@ preparePlanForEpilogueVectorLoop(VPlan &Plan, Loop *L,
   VPBasicBlock *Header = VectorLoop->getEntryBasicBlock();
   Header->setName("vec.epilog.vector.body");
 
-  DenseMap<Value *, Value *> ToFrozen;
-  // Ensure that the start values for all header phi recipes are updated before
-  // vectorizing the epilogue loop.
-  for (VPRecipeBase &R : Header->phis()) {
-    if (auto *IV = dyn_cast<VPCanonicalIVPHIRecipe>(&R)) {
-      // When vectorizing the epilogue loop, the canonical induction start
-      // value needs to be changed from zero to the value after the main
-      // vector loop. Find the resume value created during execution of the main
-      // VPlan. It must be the first phi in the loop preheader.
-      // FIXME: Improve modeling for canonical IV start values in the epilogue
-      // loop.
-      using namespace llvm::PatternMatch;
-      PHINode *EPResumeVal = &*L->getLoopPreheader()->phis().begin();
-      for (Value *Inc : EPResumeVal->incoming_values()) {
-        if (match(Inc, m_SpecificInt(0)))
-          continue;
-        assert(!EPI.VectorTripCount &&
-               "Must only have a single non-zero incoming value");
-        EPI.VectorTripCount = Inc;
-      }
+  auto *IV = cast<VPCanonicalIVPHIRecipe>(&*Header->begin());
+  // When vectorizing the epilogue loop, the canonical induction start
+  // value needs to be changed from zero to the value after the main
+  // vector loop. Find the resume value created during execution of the main
+  // VPlan. It must be the first phi in the loop preheader.
+  // FIXME: Improve modeling for canonical IV start values in the epilogue
+  // loop.
+  using namespace llvm::PatternMatch;
+  PHINode *EPResumeVal = &*L->getLoopPreheader()->phis().begin();
+  for (Value *Inc : EPResumeVal->incoming_values()) {
+    if (match(Inc, m_SpecificInt(0)))
+      continue;
+    assert(!EPI.VectorTripCount &&
+           "Must only have a single non-zero incoming value");
+    EPI.VectorTripCount = Inc;
+  }
       // If we didn't find a non-zero vector trip count, all incoming values
       // must be zero, which also means the vector trip count is zero. Pick the
       // first zero as vector trip count.
@@ -9763,70 +9759,76 @@ preparePlanForEpilogueVectorLoop(VPlan &Plan, Loop *L,
                     }) &&
              "the canonical IV should only be used by its increment or "
              "ScalarIVSteps when resetting the start value");
-      IV->setOperand(0, VPV);
-      continue;
-    }
-
-    Value *ResumeV = nullptr;
-    // TODO: Move setting of resume values to prepareToExecute.
-    if (auto *ReductionPhi = dyn_cast<VPReductionPHIRecipe>(&R)) {
-      auto *RdxResult =
-          cast<VPInstruction>(*find_if(ReductionPhi->users(), [](VPUser *U) {
-            auto *VPI = dyn_cast<VPInstruction>(U);
-            return VPI &&
-                   (VPI->getOpcode() == VPInstruction::ComputeAnyOfResult ||
-                    VPI->getOpcode() == VPInstruction::ComputeReductionResult ||
-                    VPI->getOpcode() == VPInstruction::ComputeFindIVResult);
-          }));
-      ResumeV = cast<PHINode>(ReductionPhi->getUnderlyingInstr())
-                    ->getIncomingValueForBlock(L->getLoopPreheader());
-      RecurKind RK = ReductionPhi->getRecurrenceKind();
-      if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) {
-        Value *StartV = RdxResult->getOperand(1)->getLiveInIRValue();
-        // VPReductionPHIRecipes for AnyOf reductions expect a boolean as
-        // start value; compare the final value from the main vector loop
-        // to the start value.
-        BasicBlock *PBB = cast<Instruction>(ResumeV)->getParent();
-        IRBuilder<> Builder(PBB, PBB->getFirstNonPHIIt());
-        ResumeV = Builder.CreateICmpNE(ResumeV, StartV);
-      } else if (RecurrenceDescriptor::isFindIVRecurrenceKind(RK)) {
-        Value *StartV = getStartValueFromReductionResult(RdxResult);
-        ToFrozen[StartV] = cast<PHINode>(ResumeV)->getIncomingValueForBlock(
-            EPI.MainLoopIterationCountCheck);
-
-        // VPReductionPHIRecipe for FindFirstIV/FindLastIV reductions requires
-        // an adjustment to the resume value. The resume value is adjusted to
-        // the sentinel value when the final value from the main vector loop
-        // equals the start value. This ensures correctness when the start value
-        // might not be less than the minimum value of a monotonically
-        // increasing induction variable.
-        BasicBlock *ResumeBB = cast<Instruction>(ResumeV)->getParent();
-        IRBuilder<> Builder(ResumeBB, ResumeBB->getFirstNonPHIIt());
-        Value *Cmp = Builder.CreateICmpEQ(ResumeV, ToFrozen[StartV]);
-        Value *Sentinel = RdxResult->getOperand(2)->getLiveInIRValue();
-        ResumeV = Builder.CreateSelect(Cmp, Sentinel, ResumeV);
-      } else {
-        VPValue *StartVal = Plan.getOrAddLiveIn(ResumeV);
-        auto *PhiR = dyn_cast<VPReductionPHIRecipe>(&R);
-        if (auto *VPI = dyn_cast<VPInstruction>(PhiR->getStartValue())) {
-          assert(VPI->getOpcode() == VPInstruction::ReductionStartVector &&
-                 "unexpected start value");
-          VPI->setOperand(0, StartVal);
-          continue;
+      VPBuilder Builder(Header, Header->getFirstNonPhi());
+      VPInstruction *Add = Builder.createNaryOp(Instruction::Add, {IV, VPV});
+      IV->replaceAllUsesWith(Add);
+      Add->setOperand(0, IV);
+
+      DenseMap<Value *, Value *> ToFrozen;
+      // Ensure that the start values for all header phi recipes are updated
+      // before vectorizing the epilogue loop.
+      for (VPRecipeBase &R : drop_begin(Header->phis())) {
+        Value *ResumeV = nullptr;
+        // TODO: Move setting of resume values to prepareToExecute.
+        if (auto *ReductionPhi = dyn_cast<VPReductionPHIRecipe>(&R)) {
+          auto *RdxResult = cast<VPInstruction>(
+              *find_if(ReductionPhi->users(), [](VPUser *U) {
+                auto *VPI = dyn_cast<VPInstruction>(U);
+                return VPI &&
+                       (VPI->getOpcode() == VPInstruction::ComputeAnyOfResult ||
+                        VPI->getOpcode() ==
+                            VPInstruction::ComputeReductionResult ||
+                        VPI->getOpcode() == VPInstruction::ComputeFindIVResult);
+              }));
+          ResumeV = cast<PHINode>(ReductionPhi->getUnderlyingInstr())
+                        ->getIncomingValueForBlock(L->getLoopPreheader());
+          RecurKind RK = ReductionPhi->getRecurrenceKind();
+          if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) {
+            Value *StartV = RdxResult->getOperand(1)->getLiveInIRValue();
+            // VPReductionPHIRecipes for AnyOf reductions expect a boolean as
+            // start value; compare the final value from the main vector loop
+            // to the start value.
+            BasicBlock *PBB = cast<Instruction>(ResumeV)->getParent();
+            IRBuilder<> Builder(PBB, PBB->getFirstNonPHIIt());
+            ResumeV = Builder.CreateICmpNE(ResumeV, StartV);
+          } else if (RecurrenceDescriptor::isFindIVRecurrenceKind(RK)) {
+            Value *StartV = getStartValueFromReductionResult(RdxResult);
+            ToFrozen[StartV] = cast<PHINode>(ResumeV)->getIncomingValueForBlock(
+                EPI.MainLoopIterationCountCheck);
+
+            // VPReductionPHIRecipe for FindFirstIV/FindLastIV reductions
+            // requires an adjustment to the resume value. The resume value is
+            // adjusted to the sentinel value when the final value from the main
+            // vector loop equals the start value. This ensures correctness when
+            // the start value might not be less than the minimum value of a
+            // monotonically increasing induction variable.
+            BasicBlock *ResumeBB = cast<Instruction>(ResumeV)->getParent();
+            IRBuilder<> Builder(ResumeBB, ResumeBB->getFirstNonPHIIt());
+            Value *Cmp = Builder.CreateICmpEQ(ResumeV, ToFrozen[StartV]);
+            Value *Sentinel = RdxResult->getOperand(2)->getLiveInIRValue();
+            ResumeV = Builder.CreateSelect(Cmp, Sentinel, ResumeV);
+          } else {
+            VPValue *StartVal = Plan.getOrAddLiveIn(ResumeV);
+            auto *PhiR = dyn_cast<VPReductionPHIRecipe>(&R);
+            if (auto *VPI = dyn_cast<VPInstruction>(PhiR->getStartValue())) {
+              assert(VPI->getOpcode() == VPInstruction::ReductionStartVector &&
+                     "unexpected start value");
+              VPI->setOperand(0, StartVal);
+              continue;
+            }
+          }
+        } else {
+          // Retrieve the induction resume values for wide inductions from
+          // their original phi nodes in the scalar loop.
+          PHINode *IndPhi = cast<VPWidenInductionRecipe>(&R)->getPHINode();
+          // Hook up to the PHINode generated by a ResumePhi recipe of main
+          // loop VPlan, which feeds the scalar loop.
+          ResumeV = IndPhi->getIncomingValueForBlock(L->getLoopPreheader());
         }
+        assert(ResumeV && "Must have a resume value");
+        VPValue *StartVal = Plan.getOrAddLiveIn(ResumeV);
+        cast<VPHeaderPHIRecipe>(&R)->setStartValue(StartVal);
       }
-    } else {
-      // Retrieve the induction resume values for wide inductions from
-      // their original phi nodes in the scalar loop.
-      PHINode *IndPhi = cast<VPWidenInductionRecipe>(&R)->getPHINode();
-      // Hook up to the PHINode generated by a ResumePhi recipe of main
-      // loop VPlan, which feeds the scalar loop.
-      ResumeV = IndPhi->getIncomingValueForBlock(L->getLoopPreheader());
-    }
-    assert(ResumeV && "Must have a resume value");
-    VPValue *StartVal = Plan.getOrAddLiveIn(ResumeV);
-    cast<VPHeaderPHIRecipe>(&R)->setStartValue(StartVal);
-  }
 
   // For some VPValues in the epilogue plan we must re-use the generated IR
   // values from the main plan. Replace them with live-in VPValues.
diff --git a/llvm/lib/Transforms/Vectorize/VPlanPatternMatch.h b/llvm/lib/Transforms/Vectorize/VPlanPatternMatch.h
index 18c2bef90f08f..fd5c3e4651beb 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanPatternMatch.h
+++ b/llvm/lib/Transforms/Vectorize/VPlanPatternMatch.h
@@ -383,6 +383,12 @@ m_c_Binary(const Op0_t &Op0, const Op1_t &Op1) {
   return AllRecipe_commutative_match<Opcode, Op0_t, Op1_t>(Op0, Op1);
 }
 
+template <typename Op0_t, typename Op1_t>
+inline AllRecipe_match<Instruction::Add, Op0_t, Op1_t> m_Add(const Op0_t &Op0,
+                                                             const Op1_t &Op1) {
+  return m_Binary<Instruction::Add, Op0_t, Op1_t>(Op0, Op1);
+}
+
 template <typename Op0_t, typename Op1_t>
 inline AllRecipe_commutative_match<Instruction::Add, Op0_t, Op1_t>
 m_c_Add(const Op0_t &Op0, const Op1_t &Op1) {
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index 8d376557553c0..4e00a39fd0858 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -1207,6 +1207,17 @@ static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo) {
   if (!Plan->isUnrolled())
     return;
 
+  if (match(Def, m_Add(m_VPValue(X), m_VPValue(Y))) && Y->isLiveIn() &&
+      isa<VPPhi>(X)) {
+    auto *Phi = cast<VPPhi>(X);
+    if (Phi->getOperand(1) != Def && match(Phi->getOperand(0), m_ZeroInt()) &&
+        Phi->getNumUsers() == 1 && (*Phi->user_begin() == &R)) {
+      Phi->setOperand(0, Y);
+      Def->replaceAllUsesWith(Phi);
+      return;
+    }
+  }
+
   // VPVectorPointer for part 0 can be replaced by their start pointer.
   if (auto *VecPtr = dyn_cast<VPVectorPointerRecipe>(&R)) {
     if (VecPtr->isFirstPart()) {

>From bc336d246b152fe48499f3dca2b4bfe84b6450f8 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Thu, 5 Jun 2025 22:17:46 +0100
Subject: [PATCH 3/3] [VPlan] Make canonical IV part of the region

---
 .../Transforms/Vectorize/LoopVectorize.cpp    | 210 +++++++++---------
 llvm/lib/Transforms/Vectorize/VPlan.cpp       |  82 +++++--
 llvm/lib/Transforms/Vectorize/VPlan.h         | 156 +++++--------
 .../Transforms/Vectorize/VPlanAnalysis.cpp    |  43 ++--
 .../Vectorize/VPlanConstruction.cpp           |  28 +--
 .../Transforms/Vectorize/VPlanPatternMatch.h  |   1 -
 .../lib/Transforms/Vectorize/VPlanRecipes.cpp |  16 +-
 .../Transforms/Vectorize/VPlanTransforms.cpp  |  96 ++++----
 llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp |   4 +-
 llvm/lib/Transforms/Vectorize/VPlanUtils.cpp  |   4 +-
 llvm/lib/Transforms/Vectorize/VPlanValue.h    |  25 ++-
 .../Transforms/Vectorize/VPlanVerifier.cpp    |   6 -
 .../AArch64/conditional-branches-cost.ll      |  14 +-
 .../AArch64/divs-with-scalable-vfs.ll         |   4 +-
 .../AArch64/fully-unrolled-cost.ll            |   4 +-
 .../AArch64/induction-costs-sve.ll            |   8 +-
 .../AArch64/masked-call-scalarize.ll          |   4 +-
 .../LoopVectorize/AArch64/masked-call.ll      |   2 +-
 .../LoopVectorize/AArch64/optsize_minsize.ll  |   8 +-
 .../AArch64/partial-reduce-dot-product.ll     |   6 +-
 .../AArch64/reduction-recurrence-costs-sve.ll |   4 +-
 .../AArch64/scalable-strict-fadd.ll           |  14 +-
 .../LoopVectorize/AArch64/store-costs-sve.ll  |   2 +-
 .../sve-interleaved-masked-accesses.ll        |   8 +-
 .../AArch64/sve-tail-folding-forced.ll        |   2 +-
 .../sve-tail-folding-overflow-checks.ll       |   2 +-
 .../AArch64/sve-tail-folding-reductions.ll    |  12 +-
 .../AArch64/sve-tail-folding-unroll.ll        |   4 +-
 .../LoopVectorize/AArch64/sve-tail-folding.ll |  20 +-
 .../LoopVectorize/AArch64/sve2-histcnt.ll     |   2 +-
 .../AArch64/tail-fold-uniform-memops.ll       |   4 +-
 .../AArch64/tail-folding-styles.ll            |   2 +-
 .../AArch64/uniform-args-call-variants.ll     |  10 +-
 .../LoopVectorize/ARM/mve-icmpcost.ll         |  15 +-
 .../LoopVectorize/ARM/optsize_minsize.ll      |   2 +-
 .../X86/CostModel/vpinstruction-cost.ll       |   2 -
 .../LoopVectorize/iv-select-cmp-decreasing.ll |   4 +-
 ...eref-pred-poison-ub-ops-feeding-pointer.ll |  10 +-
 .../LoopVectorize/vplan-printing.ll           |  68 +++---
 .../Transforms/Vectorize/VPlanHCFGTest.cpp    |   5 +-
 .../Vectorize/VPlanPatternMatchTest.cpp       |  29 ++-
 .../Vectorize/VPlanVerifierTest.cpp           |  27 +--
 42 files changed, 486 insertions(+), 483 deletions(-)

diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index d09e54dc4212c..c4e24b3b19942 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -4064,7 +4064,6 @@ static bool willGenerateVectors(VPlan &Plan, ElementCount VF,
       case VPDef::VPScalarIVStepsSC:
       case VPDef::VPReplicateSC:
       case VPDef::VPInstructionSC:
-      case VPDef::VPCanonicalIVPHISC:
       case VPDef::VPVectorPointerSC:
       case VPDef::VPVectorEndPointerSC:
       case VPDef::VPExpandSCEVSC:
@@ -8594,6 +8593,7 @@ VPlanPtr LoopVectorizationPlanner::tryToBuildVPlanWithVPRecipes(
                             m_Specific(Plan->getCanonicalIV()), m_VPValue())) &&
            "Did not find the canonical IV increment");
     cast<VPRecipeWithIRFlags>(IVInc)->dropPoisonGeneratingFlags();
+    Plan->getCanonicalIVInfo().HasNUW = false;
   }
 
   // ---------------------------------------------------------------------------
@@ -8657,8 +8657,7 @@ VPlanPtr LoopVectorizationPlanner::tryToBuildVPlanWithVPRecipes(
       // latter are added above for masking.
       // FIXME: Migrate code relying on the underlying instruction from VPlan0
       // to construct recipes below to not use the underlying instruction.
-      if (isa<VPCanonicalIVPHIRecipe, VPWidenCanonicalIVRecipe, VPBlendRecipe>(
-              &R) ||
+      if (isa<VPWidenCanonicalIVRecipe, VPBlendRecipe>(&R) ||
           (isa<VPInstruction>(&R) && !UnderlyingValue))
         continue;
 
@@ -8879,8 +8878,6 @@ VPlanPtr LoopVectorizationPlanner::tryToBuildVPlan(VFRange &Range) {
   VPRecipeBuilder RecipeBuilder(*Plan, OrigLoop, TLI, &TTI, Legal, CM, PSE,
                                 Builder, BlockMaskCache, nullptr /*LVer*/);
   for (auto &R : Plan->getVectorLoopRegion()->getEntryBasicBlock()->phis()) {
-    if (isa<VPCanonicalIVPHIRecipe>(&R))
-      continue;
     auto *HeaderR = cast<VPHeaderPHIRecipe>(&R);
     RecipeBuilder.setRecipe(HeaderR->getUnderlyingInstr(), HeaderR);
   }
@@ -9632,8 +9629,6 @@ static void preparePlanForMainVectorLoop(VPlan &MainPlan, VPlan &EpiPlan) {
   SmallPtrSet<PHINode *, 2> EpiWidenedPhis;
   for (VPRecipeBase &R :
        EpiPlan.getVectorLoopRegion()->getEntryBasicBlock()->phis()) {
-    if (isa<VPCanonicalIVPHIRecipe>(&R))
-      continue;
     EpiWidenedPhis.insert(
         cast<PHINode>(R.getVPSingleValue()->getUnderlyingValue()));
   }
@@ -9685,7 +9680,7 @@ static void preparePlanForMainVectorLoop(VPlan &MainPlan, VPlan &EpiPlan) {
   VPValue *VectorTC = &MainPlan.getVectorTripCount();
   // If there is a suitable resume value for the canonical induction in the
   // scalar (which will become vector) epilogue loop, use it and move it to the
-  // beginning of the scalar preheader. Otherwise create it below.
+  // beginning of the scalar prMeheader. Otherwise create it below.
   auto ResumePhiIter =
       find_if(MainScalarPH->phis(), [VectorTC](VPRecipeBase &R) {
         return match(&R, m_VPInstruction<Instruction::PHI>(m_Specific(VectorTC),
@@ -9694,8 +9689,9 @@ static void preparePlanForMainVectorLoop(VPlan &MainPlan, VPlan &EpiPlan) {
   VPPhi *ResumePhi = nullptr;
   if (ResumePhiIter == MainScalarPH->phis().end()) {
     VPBuilder ScalarPHBuilder(MainScalarPH, MainScalarPH->begin());
+    Type *Ty = VPTypeAnalysis(MainPlan).inferScalarType(VectorTC);
     ResumePhi = ScalarPHBuilder.createScalarPhi(
-        {VectorTC, MainPlan.getCanonicalIV()->getStartValue()}, {},
+        {VectorTC, MainPlan.getOrAddLiveIn(Constant::getNullValue(Ty))}, {},
         "vec.epilog.resume.val");
   } else {
     ResumePhi = cast<VPPhi>(&*ResumePhiIter);
@@ -9719,13 +9715,14 @@ preparePlanForEpilogueVectorLoop(VPlan &Plan, Loop *L,
   VPBasicBlock *Header = VectorLoop->getEntryBasicBlock();
   Header->setName("vec.epilog.vector.body");
 
-  auto *IV = cast<VPCanonicalIVPHIRecipe>(&*Header->begin());
-  // When vectorizing the epilogue loop, the canonical induction start
-  // value needs to be changed from zero to the value after the main
-  // vector loop. Find the resume value created during execution of the main
-  // VPlan. It must be the first phi in the loop preheader.
-  // FIXME: Improve modeling for canonical IV start values in the epilogue
-  // loop.
+  DenseMap<Value *, Value *> ToFrozen;
+  // Ensure that the start values for all header phi recipes are updated before
+  // vectorizing the epilogue loop.
+  auto *IV = Plan.getCanonicalIV();
+  // When vectorizing the epilogue loop, the canonical induction start value
+  // needs to be changed from zero to the value after the main vector loop. Find
+  // the resume value created during execution of the main VPlan.
+  // FIXME: Improve modeling for canonical IV start values in the epilogue loop.
   using namespace llvm::PatternMatch;
   PHINode *EPResumeVal = &*L->getLoopPreheader()->phis().begin();
   for (Value *Inc : EPResumeVal->incoming_values()) {
@@ -9735,100 +9732,97 @@ preparePlanForEpilogueVectorLoop(VPlan &Plan, Loop *L,
            "Must only have a single non-zero incoming value");
     EPI.VectorTripCount = Inc;
   }
-      // If we didn't find a non-zero vector trip count, all incoming values
-      // must be zero, which also means the vector trip count is zero. Pick the
-      // first zero as vector trip count.
-      // TODO: We should not choose VF * UF so the main vector loop is known to
-      // be dead.
-      if (!EPI.VectorTripCount) {
-        assert(
-            EPResumeVal->getNumIncomingValues() > 0 &&
-            all_of(EPResumeVal->incoming_values(),
-                   [](Value *Inc) { return match(Inc, m_SpecificInt(0)); }) &&
-            "all incoming values must be 0");
-        EPI.VectorTripCount = EPResumeVal->getOperand(0);
-      }
-      VPValue *VPV = Plan.getOrAddLiveIn(EPResumeVal);
-      assert(all_of(IV->users(),
-                    [](const VPUser *U) {
-                      return isa<VPScalarIVStepsRecipe>(U) ||
-                             isa<VPDerivedIVRecipe>(U) ||
-                             cast<VPRecipeBase>(U)->isScalarCast() ||
-                             cast<VPInstruction>(U)->getOpcode() ==
-                                 Instruction::Add;
-                    }) &&
-             "the canonical IV should only be used by its increment or "
-             "ScalarIVSteps when resetting the start value");
-      VPBuilder Builder(Header, Header->getFirstNonPhi());
-      VPInstruction *Add = Builder.createNaryOp(Instruction::Add, {IV, VPV});
-      IV->replaceAllUsesWith(Add);
-      Add->setOperand(0, IV);
-
-      DenseMap<Value *, Value *> ToFrozen;
-      // Ensure that the start values for all header phi recipes are updated
-      // before vectorizing the epilogue loop.
-      for (VPRecipeBase &R : drop_begin(Header->phis())) {
-        Value *ResumeV = nullptr;
-        // TODO: Move setting of resume values to prepareToExecute.
-        if (auto *ReductionPhi = dyn_cast<VPReductionPHIRecipe>(&R)) {
-          auto *RdxResult = cast<VPInstruction>(
-              *find_if(ReductionPhi->users(), [](VPUser *U) {
-                auto *VPI = dyn_cast<VPInstruction>(U);
-                return VPI &&
-                       (VPI->getOpcode() == VPInstruction::ComputeAnyOfResult ||
-                        VPI->getOpcode() ==
-                            VPInstruction::ComputeReductionResult ||
-                        VPI->getOpcode() == VPInstruction::ComputeFindIVResult);
-              }));
-          ResumeV = cast<PHINode>(ReductionPhi->getUnderlyingInstr())
-                        ->getIncomingValueForBlock(L->getLoopPreheader());
-          RecurKind RK = ReductionPhi->getRecurrenceKind();
-          if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) {
-            Value *StartV = RdxResult->getOperand(1)->getLiveInIRValue();
-            // VPReductionPHIRecipes for AnyOf reductions expect a boolean as
-            // start value; compare the final value from the main vector loop
-            // to the start value.
-            BasicBlock *PBB = cast<Instruction>(ResumeV)->getParent();
-            IRBuilder<> Builder(PBB, PBB->getFirstNonPHIIt());
-            ResumeV = Builder.CreateICmpNE(ResumeV, StartV);
-          } else if (RecurrenceDescriptor::isFindIVRecurrenceKind(RK)) {
-            Value *StartV = getStartValueFromReductionResult(RdxResult);
-            ToFrozen[StartV] = cast<PHINode>(ResumeV)->getIncomingValueForBlock(
-                EPI.MainLoopIterationCountCheck);
-
-            // VPReductionPHIRecipe for FindFirstIV/FindLastIV reductions
-            // requires an adjustment to the resume value. The resume value is
-            // adjusted to the sentinel value when the final value from the main
-            // vector loop equals the start value. This ensures correctness when
-            // the start value might not be less than the minimum value of a
-            // monotonically increasing induction variable.
-            BasicBlock *ResumeBB = cast<Instruction>(ResumeV)->getParent();
-            IRBuilder<> Builder(ResumeBB, ResumeBB->getFirstNonPHIIt());
-            Value *Cmp = Builder.CreateICmpEQ(ResumeV, ToFrozen[StartV]);
-            Value *Sentinel = RdxResult->getOperand(2)->getLiveInIRValue();
-            ResumeV = Builder.CreateSelect(Cmp, Sentinel, ResumeV);
-          } else {
-            VPValue *StartVal = Plan.getOrAddLiveIn(ResumeV);
-            auto *PhiR = dyn_cast<VPReductionPHIRecipe>(&R);
-            if (auto *VPI = dyn_cast<VPInstruction>(PhiR->getStartValue())) {
-              assert(VPI->getOpcode() == VPInstruction::ReductionStartVector &&
-                     "unexpected start value");
-              VPI->setOperand(0, StartVal);
-              continue;
-            }
-          }
-        } else {
-          // Retrieve the induction resume values for wide inductions from
-          // their original phi nodes in the scalar loop.
-          PHINode *IndPhi = cast<VPWidenInductionRecipe>(&R)->getPHINode();
-          // Hook up to the PHINode generated by a ResumePhi recipe of main
-          // loop VPlan, which feeds the scalar loop.
-          ResumeV = IndPhi->getIncomingValueForBlock(L->getLoopPreheader());
-        }
-        assert(ResumeV && "Must have a resume value");
+  // If we didn't find a non-zero vector trip count, all incoming values
+  // must be zero, which also means the vector trip count is zero. Pick the
+  // first zero as vector trip count.
+  // TODO: We should not choose VF * UF so the main vector loop is known to
+  // be dead.
+  if (!EPI.VectorTripCount) {
+    assert(EPResumeVal->getNumIncomingValues() > 0 &&
+           all_of(EPResumeVal->incoming_values(),
+                  [](Value *Inc) { return match(Inc, m_SpecificInt(0)); }) &&
+           "all incoming values must be 0");
+    EPI.VectorTripCount = EPResumeVal->getOperand(0);
+  }
+  VPValue *VPV = Plan.getOrAddLiveIn(EPResumeVal);
+  assert(all_of(IV->users(),
+                [](const VPUser *U) {
+                  return isa<VPScalarIVStepsRecipe>(U) ||
+                         isa<VPDerivedIVRecipe>(U) ||
+                         cast<VPRecipeBase>(U)->isScalarCast() ||
+                         cast<VPInstruction>(U)->getOpcode() ==
+                             Instruction::Add;
+                }) &&
+         "the canonical IV should only be used by its increment or "
+         "ScalarIVSteps when resetting the start value");
+  VPBuilder Builder(Header, Header->getFirstNonPhi());
+  VPInstruction *Add = Builder.createNaryOp(Instruction::Add, {IV, VPV});
+  IV->replaceAllUsesWith(Add);
+  Add->setOperand(0, IV);
+
+  // Ensure that the start values for all header phi recipes are updated before
+  // vectorizing the epilogue loop.
+  for (VPRecipeBase &R : Header->phis()) {
+    Value *ResumeV = nullptr;
+    // TODO: Move setting of resume values to prepareToExecute.
+    if (auto *ReductionPhi = dyn_cast<VPReductionPHIRecipe>(&R)) {
+      auto *RdxResult =
+          cast<VPInstruction>(*find_if(ReductionPhi->users(), [](VPUser *U) {
+            auto *VPI = dyn_cast<VPInstruction>(U);
+            return VPI &&
+                   (VPI->getOpcode() == VPInstruction::ComputeAnyOfResult ||
+                    VPI->getOpcode() == VPInstruction::ComputeReductionResult ||
+                    VPI->getOpcode() == VPInstruction::ComputeFindIVResult);
+          }));
+      ResumeV = cast<PHINode>(ReductionPhi->getUnderlyingInstr())
+                    ->getIncomingValueForBlock(L->getLoopPreheader());
+      RecurKind RK = ReductionPhi->getRecurrenceKind();
+      if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) {
+        Value *StartV = RdxResult->getOperand(1)->getLiveInIRValue();
+        // VPReductionPHIRecipes for AnyOf reductions expect a boolean as
+        // start value; compare the final value from the main vector loop
+        // to the start value.
+        BasicBlock *PBB = cast<Instruction>(ResumeV)->getParent();
+        IRBuilder<> Builder(PBB, PBB->getFirstNonPHIIt());
+        ResumeV = Builder.CreateICmpNE(ResumeV, StartV);
+      } else if (RecurrenceDescriptor::isFindIVRecurrenceKind(RK)) {
+        Value *StartV = getStartValueFromReductionResult(RdxResult);
+        ToFrozen[StartV] = cast<PHINode>(ResumeV)->getIncomingValueForBlock(
+            EPI.MainLoopIterationCountCheck);
+
+        // VPReductionPHIRecipe for FindFirstIV/FindLastIV reductions
+        // requires an adjustment to the resume value. The resume value is
+        // adjusted to the sentinel value when the final value from the main
+        // vector loop equals the start value. This ensures correctness when
+        // the start value might not be less than the minimum value of a
+        // monotonically increasing induction variable.
+        BasicBlock *ResumeBB = cast<Instruction>(ResumeV)->getParent();
+        IRBuilder<> Builder(ResumeBB, ResumeBB->getFirstNonPHIIt());
+        Value *Cmp = Builder.CreateICmpEQ(ResumeV, ToFrozen[StartV]);
+        Value *Sentinel = RdxResult->getOperand(2)->getLiveInIRValue();
+        ResumeV = Builder.CreateSelect(Cmp, Sentinel, ResumeV);
+      } else {
         VPValue *StartVal = Plan.getOrAddLiveIn(ResumeV);
-        cast<VPHeaderPHIRecipe>(&R)->setStartValue(StartVal);
+        auto *PhiR = dyn_cast<VPReductionPHIRecipe>(&R);
+        if (auto *VPI = dyn_cast<VPInstruction>(PhiR->getStartValue())) {
+          assert(VPI->getOpcode() == VPInstruction::ReductionStartVector &&
+                 "unexpected start value");
+          VPI->setOperand(0, StartVal);
+          continue;
+        }
       }
+    } else {
+      // Retrieve the induction resume values for wide inductions from
+      // their original phi nodes in the scalar loop.
+      PHINode *IndPhi = cast<VPWidenInductionRecipe>(&R)->getPHINode();
+      // Hook up to the PHINode generated by a ResumePhi recipe of main
+      // loop VPlan, which feeds the scalar loop.
+      ResumeV = IndPhi->getIncomingValueForBlock(L->getLoopPreheader());
+    }
+    assert(ResumeV && "Must have a resume value");
+    VPValue *StartVal = Plan.getOrAddLiveIn(ResumeV);
+    cast<VPHeaderPHIRecipe>(&R)->setStartValue(StartVal);
+  }
 
   // For some VPValues in the epilogue plan we must re-use the generated IR
   // values from the main plan. Replace them with live-in VPValues.
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.cpp b/llvm/lib/Transforms/Vectorize/VPlan.cpp
index 1438dc366b55d..e3c42029ca6c3 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlan.cpp
@@ -771,10 +771,16 @@ static std::pair<VPBlockBase *, VPBlockBase *> cloneFrom(VPBlockBase *Entry) {
 
 VPRegionBlock *VPRegionBlock::clone() {
   const auto &[NewEntry, NewExiting] = cloneFrom(getEntry());
-  auto *NewRegion = getPlan()->createVPRegionBlock(NewEntry, NewExiting,
-                                                   getName(), isReplicator());
+  auto *NewRegion =
+      getPlan()->createVPRegionBlock(NewEntry, NewExiting, getName());
   for (VPBlockBase *Block : vp_depth_first_shallow(NewEntry))
     Block->setParent(NewRegion);
+
+  if (CanIV.CanIV) {
+    NewRegion->CanIV.CanIV = new VPRegionValue();
+    NewRegion->CanIV.HasNUW = CanIV.HasNUW;
+    NewRegion->CanIV.DL = CanIV.DL;
+  }
   return NewRegion;
 }
 
@@ -868,6 +874,11 @@ void VPRegionBlock::print(raw_ostream &O, const Twine &Indent,
                           VPSlotTracker &SlotTracker) const {
   O << Indent << (isReplicator() ? "<xVFxUF> " : "<x1> ") << getName() << ": {";
   auto NewIndent = Indent + "  ";
+  if (CanIV.CanIV) {
+    O << '\n';
+    CanIV.CanIV->print(O, SlotTracker);
+    O << '\n';
+  }
   for (auto *BlockBase : vp_depth_first_shallow(Entry)) {
     O << '\n';
     BlockBase->print(O, NewIndent, SlotTracker);
@@ -880,18 +891,48 @@ void VPRegionBlock::print(raw_ostream &O, const Twine &Indent,
 
 void VPRegionBlock::dissolveToCFGLoop() {
   auto *Header = cast<VPBasicBlock>(getEntry());
-  if (auto *CanIV = dyn_cast<VPCanonicalIVPHIRecipe>(&Header->front())) {
-    assert(this == getPlan()->getVectorLoopRegion() &&
-           "Canonical IV must be in the entry of the top-level loop region");
-    auto *ScalarR = VPBuilder(CanIV).createScalarPhi(
-        {CanIV->getStartValue(), CanIV->getBackedgeValue()},
-        CanIV->getDebugLoc(), "index");
-    CanIV->replaceAllUsesWith(ScalarR);
-    CanIV->eraseFromParent();
+  auto *ExitingLatch = cast<VPBasicBlock>(getExiting());
+  if (CanIV.CanIV && CanIV.CanIV->getNumUsers() > 0) {
+    VPlan &Plan = *getPlan();
+    auto *ExitingTerm = ExitingLatch->getTerminator();
+    VPValue *IncCand;
+    VPInstruction *CanIVInc = nullptr;
+    if (match(ExitingTerm, m_BranchOnCount(m_VPValue(IncCand), m_VPValue()))) {
+      assert(!CanIVInc && "Expecting at most one canonical IV increment");
+      CanIVInc = cast<VPInstruction>(IncCand);
+    }
+    if (!CanIVInc) {
+      for (VPUser *U : CanIV.CanIV->users()) {
+        VPValue *Inc;
+        if (!match(U, m_VPInstruction<Instruction::Add>(m_Specific(CanIV.CanIV),
+                                                        m_VPValue(Inc))))
+          continue;
+        auto *IncCand = cast<VPInstruction>(U);
+        if (Inc != &Plan.getVFxUF() &&
+            (IncCand->getNumUsers() != 1 ||
+             !match(*IncCand->user_begin(),
+                    m_BranchOnCount(m_Specific(IncCand), m_VPValue()))))
+          continue;
+        assert(!CanIVInc && "Expecting at most one canonical IV increment");
+        CanIVInc = cast<VPInstruction>(U);
+      }
+    }
+    if (!CanIVInc) {
+      CanIVInc = VPBuilder(ExitingTerm)
+                     .createOverflowingOp(
+                         Instruction::Add, {CanIV.CanIV, &Plan.getVFxUF()},
+                         {CanIV.HasNUW, false}, CanIV.DL, "index.next");
+    }
+    Type *CanIVTy = VPTypeAnalysis(Plan).inferScalarType(CanIV.CanIV);
+    auto *ScalarR =
+        VPBuilder(Header, Header->begin())
+            .createScalarPhi(
+                {Plan.getOrAddLiveIn(ConstantInt::get(CanIVTy, 0)), CanIVInc},
+                CanIV.DL, "index");
+    CanIV.CanIV->replaceAllUsesWith(ScalarR);
   }
 
   VPBlockBase *Preheader = getSinglePredecessor();
-  auto *ExitingLatch = cast<VPBasicBlock>(getExiting());
   VPBlockBase *Middle = getSingleSuccessor();
   VPBlockUtils::disconnectBlocks(Preheader, this);
   VPBlockUtils::disconnectBlocks(this, Middle);
@@ -928,7 +969,10 @@ VPlan::~VPlan() {
         for (unsigned I = 0, E = R.getNumOperands(); I != E; I++)
           R.setOperand(I, &DummyValue);
       }
+    } else if (auto *CanIV = cast<VPRegionBlock>(VPB)->getCanonicalIV()) {
+      CanIV->replaceAllUsesWith(&DummyValue);
     }
+
     delete VPB;
   }
   for (VPValue *VPV : getLiveIns())
@@ -1212,6 +1256,11 @@ VPlan *VPlan::duplicate() {
   // else NewTripCount will be created and inserted into Old2NewVPValues when
   // TripCount is cloned. In any case NewPlan->TripCount is updated below.
 
+  if (auto *LoopRegion = getVectorLoopRegion()) {
+    Old2NewVPValues[LoopRegion->getCanonicalIV()] =
+        NewPlan->getVectorLoopRegion()->getCanonicalIV();
+  }
+
   remapOperands(Entry, NewEntry, Old2NewVPValues);
 
   // Initialize remaining fields of cloned VPlan.
@@ -1392,6 +1441,8 @@ void VPlanPrinter::dumpRegion(const VPRegionBlock *Region) {
 /// Returns true if there is a vector loop region and \p VPV is defined in a
 /// loop region.
 static bool isDefinedInsideLoopRegions(const VPValue *VPV) {
+  if (isa<VPRegionValue>(VPV))
+    return true;
   const VPRecipeBase *DefR = VPV->getDefiningRecipe();
   return DefR && (!DefR->getParent()->getPlan()->getVectorLoopRegion() ||
                   DefR->getParent()->getEnclosingLoopRegion());
@@ -1501,9 +1552,12 @@ void VPSlotTracker::assignNames(const VPlan &Plan) {
 
   ReversePostOrderTraversal<VPBlockDeepTraversalWrapper<const VPBlockBase *>>
       RPOT(VPBlockDeepTraversalWrapper<const VPBlockBase *>(Plan.getEntry()));
-  for (const VPBasicBlock *VPBB :
-       VPBlockUtils::blocksOnly<const VPBasicBlock>(RPOT))
-    assignNames(VPBB);
+  for (const VPBlockBase *VPB : RPOT) {
+    if (auto *VPBB = dyn_cast<VPBasicBlock>(VPB)) {
+      assignNames(VPBB);
+    } else if (auto *CanIV = cast<VPRegionBlock>(VPB)->getCanonicalIV())
+      assignName(CanIV);
+  }
 }
 
 void VPSlotTracker::assignNames(const VPBasicBlock *VPBB) {
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
index 055db2b9adb95..fae8c8b87d074 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.h
+++ b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -547,7 +547,6 @@ class VPSingleDefRecipe : public VPRecipeBase, public VPValue {
     case VPRecipeBase::VPWidenSelectSC:
     case VPRecipeBase::VPBlendSC:
     case VPRecipeBase::VPPredInstPHISC:
-    case VPRecipeBase::VPCanonicalIVPHISC:
     case VPRecipeBase::VPActiveLaneMaskPHISC:
     case VPRecipeBase::VPFirstOrderRecurrencePHISC:
     case VPRecipeBase::VPWidenPHISC:
@@ -1930,12 +1929,6 @@ class VPVectorPointerRecipe : public VPRecipeWithIRFlags,
 /// the backedge is the second operand.
 ///
 /// Inductions are modeled using the following sub-classes:
-///  * VPCanonicalIVPHIRecipe: Canonical scalar induction of the vector loop,
-///    starting at a specified value (zero for the main vector loop, the resume
-///    value for the epilogue vector loop) and stepping by 1. The induction
-///    controls exiting of the vector loop by comparing against the vector trip
-///    count. Produces a single scalar PHI for the induction value per
-///    iteration.
 ///  * VPWidenIntOrFpInductionRecipe: Generates vector values for integer and
 ///    floating point inductions with arbitrary start and step values. Produces
 ///    a vector PHI per-part.
@@ -3288,63 +3281,6 @@ class VPExpandSCEVRecipe : public VPSingleDefRecipe {
   const SCEV *getSCEV() const { return Expr; }
 };
 
-/// Canonical scalar induction phi of the vector loop. Starting at the specified
-/// start value (either 0 or the resume value when vectorizing the epilogue
-/// loop). VPWidenCanonicalIVRecipe represents the vector version of the
-/// canonical induction variable.
-class VPCanonicalIVPHIRecipe : public VPHeaderPHIRecipe {
-public:
-  VPCanonicalIVPHIRecipe(VPValue *StartV, DebugLoc DL)
-      : VPHeaderPHIRecipe(VPDef::VPCanonicalIVPHISC, nullptr, StartV, DL) {}
-
-  ~VPCanonicalIVPHIRecipe() override = default;
-
-  VPCanonicalIVPHIRecipe *clone() override {
-    auto *R = new VPCanonicalIVPHIRecipe(getOperand(0), getDebugLoc());
-    R->addOperand(getBackedgeValue());
-    return R;
-  }
-
-  VP_CLASSOF_IMPL(VPDef::VPCanonicalIVPHISC)
-
-  void execute(VPTransformState &State) override {
-    llvm_unreachable("cannot execute this recipe, should be replaced by a "
-                     "scalar phi recipe");
-  }
-
-#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
-  /// Print the recipe.
-  void print(raw_ostream &O, const Twine &Indent,
-             VPSlotTracker &SlotTracker) const override;
-#endif
-
-  /// Returns the scalar type of the induction.
-  Type *getScalarType() const {
-    return getStartValue()->getLiveInIRValue()->getType();
-  }
-
-  /// Returns true if the recipe only uses the first lane of operand \p Op.
-  bool onlyFirstLaneUsed(const VPValue *Op) const override {
-    assert(is_contained(operands(), Op) &&
-           "Op must be an operand of the recipe");
-    return true;
-  }
-
-  /// Returns true if the recipe only uses the first part of operand \p Op.
-  bool onlyFirstPartUsed(const VPValue *Op) const override {
-    assert(is_contained(operands(), Op) &&
-           "Op must be an operand of the recipe");
-    return true;
-  }
-
-  /// Return the cost of this VPCanonicalIVPHIRecipe.
-  InstructionCost computeCost(ElementCount VF,
-                              VPCostContext &Ctx) const override {
-    // For now, match the behavior of the legacy cost model.
-    return 0;
-  }
-};
-
 /// A recipe for generating the active lane mask for the vector loop that is
 /// used to predicate the vector operations.
 /// TODO: It would be good to use the existing VPWidenPHIRecipe instead and
@@ -3423,14 +3359,13 @@ class VPEVLBasedIVPHIRecipe : public VPHeaderPHIRecipe {
 class VPWidenCanonicalIVRecipe : public VPSingleDefRecipe,
                                  public VPUnrollPartAccessor<1> {
 public:
-  VPWidenCanonicalIVRecipe(VPCanonicalIVPHIRecipe *CanonicalIV)
+  VPWidenCanonicalIVRecipe(VPValue *CanonicalIV)
       : VPSingleDefRecipe(VPDef::VPWidenCanonicalIVSC, {CanonicalIV}) {}
 
   ~VPWidenCanonicalIVRecipe() override = default;
 
   VPWidenCanonicalIVRecipe *clone() override {
-    return new VPWidenCanonicalIVRecipe(
-        cast<VPCanonicalIVPHIRecipe>(getOperand(0)));
+    return new VPWidenCanonicalIVRecipe(getOperand(0));
   }
 
   VP_CLASSOF_IMPL(VPDef::VPWidenCanonicalIVSC)
@@ -3469,8 +3404,7 @@ class VPDerivedIVRecipe : public VPSingleDefRecipe {
 
 public:
   VPDerivedIVRecipe(const InductionDescriptor &IndDesc, VPValue *Start,
-                    VPCanonicalIVPHIRecipe *CanonicalIV, VPValue *Step,
-                    const Twine &Name = "")
+                    VPValue *CanonicalIV, VPValue *Step, const Twine &Name = "")
       : VPDerivedIVRecipe(
             IndDesc.getKind(),
             dyn_cast_or_null<FPMathOperator>(IndDesc.getInductionBinOp()),
@@ -3816,6 +3750,17 @@ class VPIRBasicBlock : public VPBasicBlock {
   BasicBlock *getIRBasicBlock() const { return IRBB; }
 };
 
+struct VPCanonicalIVS {
+  VPValue *CanIV = nullptr;
+  bool HasNUW = true;
+  DebugLoc DL = DebugLoc::getUnknown();
+
+  VPCanonicalIVS(VPValue *CanIV, bool HasNUW, DebugLoc DL)
+      : CanIV(CanIV), HasNUW(HasNUW), DL(DL) {}
+
+  VPCanonicalIVS() {}
+};
+
 /// VPRegionBlock represents a collection of VPBasicBlocks and VPRegionBlocks
 /// which form a Single-Entry-Single-Exiting subgraph of the output IR CFG.
 /// A VPRegionBlock may indicate that its contents are to be replicated several
@@ -3834,26 +3779,40 @@ class LLVM_ABI_FOR_TEST VPRegionBlock : public VPBlockBase {
   /// VPRegionBlock.
   VPBlockBase *Exiting;
 
-  /// An indicator whether this region is to generate multiple replicated
-  /// instances of output IR corresponding to its VPBlockBases.
-  bool IsReplicator;
+  /// Canonical IV of the loop region. If nullptr, the region is a replication
+  /// region.
+  VPCanonicalIVS CanIV;
 
   /// Use VPlan::createVPRegionBlock to create VPRegionBlocks.
   VPRegionBlock(VPBlockBase *Entry, VPBlockBase *Exiting,
-                const std::string &Name = "", bool IsReplicator = false)
+                const std::string &Name = "")
+      : VPBlockBase(VPRegionBlockSC, Name), Entry(Entry), Exiting(Exiting),
+        CanIV() {
+    assert(Entry->getPredecessors().empty() && "Entry block has predecessors.");
+    assert(Exiting->getSuccessors().empty() && "Exit block has successors.");
+    Entry->setParent(this);
+    Exiting->setParent(this);
+  }
+
+  VPRegionBlock(VPBlockBase *Entry, VPBlockBase *Exiting, VPCanonicalIVS CanIV,
+                const std::string &Name = "")
       : VPBlockBase(VPRegionBlockSC, Name), Entry(Entry), Exiting(Exiting),
-        IsReplicator(IsReplicator) {
+        CanIV(CanIV) {
     assert(Entry->getPredecessors().empty() && "Entry block has predecessors.");
     assert(Exiting->getSuccessors().empty() && "Exit block has successors.");
     Entry->setParent(this);
     Exiting->setParent(this);
   }
-  VPRegionBlock(const std::string &Name = "", bool IsReplicator = false)
+
+  VPRegionBlock(DebugLoc DL, const std::string &Name = "")
       : VPBlockBase(VPRegionBlockSC, Name), Entry(nullptr), Exiting(nullptr),
-        IsReplicator(IsReplicator) {}
+        CanIV(new VPRegionValue(), true, DL) {}
 
 public:
-  ~VPRegionBlock() override {}
+  ~VPRegionBlock() override {
+    if (getCanonicalIV())
+      delete CanIV.CanIV;
+  }
 
   /// Method to support type inquiry through isa, cast, and dyn_cast.
   static inline bool classof(const VPBlockBase *V) {
@@ -3892,7 +3851,7 @@ class LLVM_ABI_FOR_TEST VPRegionBlock : public VPBlockBase {
 
   /// An indicator whether this region is to generate multiple replicated
   /// instances of output IR corresponding to its VPBlockBases.
-  bool isReplicator() const { return IsReplicator; }
+  bool isReplicator() const { return !getCanonicalIV(); }
 
   /// The method which generates the output IR instructions that correspond to
   /// this VPRegionBlock, thereby "executing" the VPlan.
@@ -3920,6 +3879,13 @@ class LLVM_ABI_FOR_TEST VPRegionBlock : public VPBlockBase {
   /// Remove the current region from its VPlan, connecting its predecessor to
   /// its entry, and its exiting block to its successor.
   void dissolveToCFGLoop();
+
+  /// Return the canonical induction variable of the region, null for
+  /// replicating regions.
+  VPValue *getCanonicalIV() { return CanIV.CanIV; }
+  const VPValue *getCanonicalIV() const { return CanIV.CanIV; }
+
+  VPCanonicalIVS &getCanonicalIVInfo() { return CanIV; }
 };
 
 /// VPlan models a candidate for vectorization, encoding various decisions take
@@ -4231,14 +4197,10 @@ class VPlan {
   LLVM_DUMP_METHOD void dump() const;
 #endif
 
-  /// Returns the canonical induction recipe of the vector loop.
-  VPCanonicalIVPHIRecipe *getCanonicalIV() {
-    VPBasicBlock *EntryVPBB = getVectorLoopRegion()->getEntryBasicBlock();
-    if (EntryVPBB->empty()) {
-      // VPlan native path.
-      EntryVPBB = cast<VPBasicBlock>(EntryVPBB->getSingleSuccessor());
-    }
-    return cast<VPCanonicalIVPHIRecipe>(&*EntryVPBB->begin());
+  /// Returns the canonical induction VPValue of the vector loop.
+  VPValue *getCanonicalIV() { return getVectorLoopRegion()->getCanonicalIV(); }
+  VPCanonicalIVS &getCanonicalIVInfo() {
+    return getVectorLoopRegion()->getCanonicalIVInfo();
   }
 
   VPValue *getSCEVExpansion(const SCEV *S) const {
@@ -4264,22 +4226,22 @@ class VPlan {
     return VPB;
   }
 
-  /// Create a new VPRegionBlock with \p Entry, \p Exiting and \p Name. If \p
-  /// IsReplicator is true, the region is a replicate region. The returned block
-  /// is owned by the VPlan and deleted once the VPlan is destroyed.
+  /// Create a new replicate VPRegionBlock with \p Entry, \p Exiting and \p
+  /// Name. The returned block is owned by the VPlan and deleted once the VPlan
+  /// is destroyed.
   VPRegionBlock *createVPRegionBlock(VPBlockBase *Entry, VPBlockBase *Exiting,
-                                     const std::string &Name = "",
-                                     bool IsReplicator = false) {
-    auto *VPB = new VPRegionBlock(Entry, Exiting, Name, IsReplicator);
+                                     const std::string &Name = "") {
+    auto *VPB = new VPRegionBlock(Entry, Exiting, Name);
     CreatedBlocks.push_back(VPB);
     return VPB;
   }
 
-  /// Create a new loop VPRegionBlock with \p Name and entry and exiting blocks set
-  /// to nullptr. The returned block is owned by the VPlan and deleted once the
-  /// VPlan is destroyed.
-  VPRegionBlock *createVPRegionBlock(const std::string &Name = "") {
-    auto *VPB = new VPRegionBlock(Name);
+  /// Create a new loop VPRegionBlock with \p StartV and \p Name, and entry and
+  /// exiting blocks set to nullptr. The returned block is owned by the VPlan
+  /// and deleted once the VPlan is destroyed.
+  VPRegionBlock *createVPRegionBlock(DebugLoc DL,
+                                     const std::string &Name = "") {
+    auto *VPB = new VPRegionBlock(DL, Name);
     CreatedBlocks.push_back(VPB);
     return VPB;
   }
diff --git a/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp b/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
index 8e5997fa8befe..73b89e3251557 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
@@ -23,14 +23,6 @@ using namespace llvm;
 #define DEBUG_TYPE "vplan"
 
 VPTypeAnalysis::VPTypeAnalysis(const VPlan &Plan) : Ctx(Plan.getContext()) {
-  if (auto LoopRegion = Plan.getVectorLoopRegion()) {
-    if (const auto *CanIV = dyn_cast<VPCanonicalIVPHIRecipe>(
-            &LoopRegion->getEntryBasicBlock()->front())) {
-      CanonicalIVTy = CanIV->getScalarType();
-      return;
-    }
-  }
-
   // If there's no canonical IV, retrieve the type from the trip count
   // expression.
   auto *TC = Plan.getTripCount();
@@ -270,18 +262,20 @@ Type *VPTypeAnalysis::inferScalarType(const VPValue *V) {
     return CanonicalIVTy;
   }
 
+  if (auto *CanIV = dyn_cast<VPRegionValue>(V))
+    return CanonicalIVTy;
+
   Type *ResultTy =
       TypeSwitch<const VPRecipeBase *, Type *>(V->getDefiningRecipe())
-          .Case<VPActiveLaneMaskPHIRecipe, VPCanonicalIVPHIRecipe,
-                VPFirstOrderRecurrencePHIRecipe, VPReductionPHIRecipe,
-                VPWidenPointerInductionRecipe, VPEVLBasedIVPHIRecipe>(
-              [this](const auto *R) {
-                // Handle header phi recipes, except VPWidenIntOrFpInduction
-                // which needs special handling due it being possibly truncated.
-                // TODO: consider inferring/caching type of siblings, e.g.,
-                // backedge value, here and in cases below.
-                return inferScalarType(R->getStartValue());
-              })
+          .Case<VPActiveLaneMaskPHIRecipe, VPFirstOrderRecurrencePHIRecipe,
+                VPReductionPHIRecipe, VPWidenPointerInductionRecipe,
+                VPEVLBasedIVPHIRecipe>([this](const auto *R) {
+            // Handle header phi recipes, except VPWidenIntOrFpInduction
+            // which needs special handling due it being possibly truncated.
+            // TODO: consider inferring/caching type of siblings, e.g.,
+            // backedge value, here and in cases below.
+            return inferScalarType(R->getStartValue());
+          })
           .Case<VPWidenIntOrFpInductionRecipe, VPDerivedIVRecipe>(
               [](const auto *R) { return R->getScalarType(); })
           .Case<VPReductionRecipe, VPPredInstPHIRecipe, VPWidenPHIRecipe,
@@ -460,12 +454,13 @@ SmallVector<VPRegisterUsage, 8> llvm::calculateRegisterUsageForPlan(
         // FIXME: Might need some motivation why these values are ignored. If
         // for example an argument is used inside the loop it will increase the
         // register pressure (so shouldn't we add it to LoopInvariants).
-        if (!DefR && (!U->getLiveInIRValue() ||
-                      !isa<Instruction>(U->getLiveInIRValue())))
+        if (!isa<VPRegionValue>(U) && !DefR &&
+            (!U->getLiveInIRValue() ||
+             !isa<Instruction>(U->getLiveInIRValue())))
           continue;
 
         // If this recipe is outside the loop then record it and continue.
-        if (!DefR) {
+        if (!DefR && !isa<VPRegionValue>(U)) {
           LoopInvariants.insert(U);
           continue;
         }
@@ -513,6 +508,10 @@ SmallVector<VPRegisterUsage, 8> llvm::calculateRegisterUsageForPlan(
     return TTICapture.getRegUsageForType(VectorType::get(Ty, VF));
   };
 
+  if (auto *CanIV = LoopRegion->getCanonicalIV())
+    if (CanIV->getNumUsers() != 0)
+      OpenIntervals.insert(CanIV);
+
   // We scan the instructions linearly and record each time that a new interval
   // starts, by placing it in a set. If we find this value in TransposEnds then
   // we remove it from the set. The max register usage is the maximum register
@@ -558,7 +557,7 @@ SmallVector<VPRegisterUsage, 8> llvm::calculateRegisterUsageForPlan(
           continue;
 
         if (VFs[J].isScalar() ||
-            isa<VPCanonicalIVPHIRecipe, VPReplicateRecipe, VPDerivedIVRecipe,
+            isa<VPRegionValue, VPReplicateRecipe, VPDerivedIVRecipe,
                 VPEVLBasedIVPHIRecipe, VPScalarIVStepsRecipe>(VPV) ||
             (isa<VPInstruction>(VPV) && vputils::onlyScalarValuesUsed(VPV)) ||
             (isa<VPReductionPHIRecipe>(VPV) &&
diff --git a/llvm/lib/Transforms/Vectorize/VPlanConstruction.cpp b/llvm/lib/Transforms/Vectorize/VPlanConstruction.cpp
index c1c525ee23122..6be031b0d39a0 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanConstruction.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanConstruction.cpp
@@ -392,7 +392,7 @@ static bool canonicalHeaderAndLatch(VPBlockBase *HeaderVPB,
 /// Create a new VPRegionBlock for the loop starting at \p HeaderVPB.
 static void createLoopRegion(VPlan &Plan, VPBlockBase *HeaderVPB) {
   auto *PreheaderVPBB = HeaderVPB->getPredecessors()[0];
-  auto *LatchVPBB = HeaderVPB->getPredecessors()[1];
+  auto *LatchVPBB = cast<VPBasicBlock>(HeaderVPB->getPredecessors()[1]);
 
   VPBlockUtils::disconnectBlocks(PreheaderVPBB, HeaderVPB);
   VPBlockUtils::disconnectBlocks(LatchVPBB, HeaderVPB);
@@ -403,13 +403,22 @@ static void createLoopRegion(VPlan &Plan, VPBlockBase *HeaderVPB) {
   // LatchExitVPB, taking care to preserve the original predecessor & successor
   // order of blocks. Set region entry and exiting after both HeaderVPB and
   // LatchVPBB have been disconnected from their predecessors/successors.
-  auto *R = Plan.createVPRegionBlock();
+  VPPhi *ScalarCanIV = nullptr;
+  if (PreheaderVPBB->getSinglePredecessor() == Plan.getEntry())
+    ScalarCanIV = cast<VPPhi>(&*cast<VPBasicBlock>(HeaderVPB)->begin());
+  auto *R =
+      Plan.createVPRegionBlock(ScalarCanIV ? ScalarCanIV->getDebugLoc()
+                                           : DebugLoc::getCompilerGenerated());
   VPBlockUtils::insertOnEdge(LatchVPBB, LatchExitVPB, R);
   VPBlockUtils::disconnectBlocks(LatchVPBB, R);
   VPBlockUtils::connectBlocks(PreheaderVPBB, R);
   R->setEntry(HeaderVPB);
   R->setExiting(LatchVPBB);
 
+  if (ScalarCanIV) {
+    ScalarCanIV->replaceAllUsesWith(R->getCanonicalIV());
+    ScalarCanIV->eraseFromParent();
+  }
   // All VPBB's reachable shallowly from HeaderVPB belong to the current region.
   for (VPBlockBase *VPBB : vp_depth_first_shallow(HeaderVPB))
     VPBB->setParent(R);
@@ -422,9 +431,7 @@ static void addCanonicalIVRecipes(VPlan &Plan, VPBasicBlock *HeaderVPBB,
                                   DebugLoc DL) {
   Value *StartIdx = ConstantInt::get(IdxTy, 0);
   auto *StartV = Plan.getOrAddLiveIn(StartIdx);
-
-  // Add a VPCanonicalIVPHIRecipe starting at 0 to the header.
-  auto *CanonicalIVPHI = new VPCanonicalIVPHIRecipe(StartV, DL);
+  auto *CanonicalIVPHI = new VPPhi(StartV, DL);
   HeaderVPBB->insert(CanonicalIVPHI, HeaderVPBB->begin());
 
   // We are about to replace the branch to exit the region. Remove the original
@@ -437,14 +444,9 @@ static void addCanonicalIVRecipes(VPlan &Plan, VPBasicBlock *HeaderVPBB,
   }
 
   VPBuilder Builder(LatchVPBB);
-  // Add a VPInstruction to increment the scalar canonical IV by VF * UF.
-  // Initially the induction increment is guaranteed to not wrap, but that may
-  // change later, e.g. when tail-folding, when the flags need to be dropped.
-  auto *CanonicalIVIncrement = Builder.createOverflowingOp(
+  auto CanonicalIVIncrement = Builder.createOverflowingOp(
       Instruction::Add, {CanonicalIVPHI, &Plan.getVFxUF()}, {true, false}, DL,
       "index.next");
-  CanonicalIVPHI->addOperand(CanonicalIVIncrement);
-
   // Add the BranchOnCount VPInstruction to the latch.
   Builder.createNaryOp(VPInstruction::BranchOnCount,
                        {CanonicalIVIncrement, &Plan.getVectorTripCount()},
@@ -660,7 +662,7 @@ void VPlanTransforms::attachCheckBlock(VPlan &Plan, Value *Cond,
   VPIRMetadata VPBranchWeights;
   auto *Term = VPBuilder(CheckBlockVPBB)
                    .createNaryOp(VPInstruction::BranchOnCond, {CondVPV},
-                                 Plan.getCanonicalIV()->getDebugLoc());
+                                 Plan.getCanonicalIVInfo().DL);
   if (AddBranchWeights) {
     MDBuilder MDB(Plan.getContext());
     MDNode *BranchWeights =
@@ -789,7 +791,7 @@ bool VPlanTransforms::handleMaxMinNumReductions(VPlan &Plan) {
   VPReductionPHIRecipe *RedPhiR = nullptr;
   bool HasUnsupportedPhi = false;
   for (auto &R : LoopRegion->getEntryBasicBlock()->phis()) {
-    if (isa<VPCanonicalIVPHIRecipe, VPWidenIntOrFpInductionRecipe>(&R))
+    if (isa<VPWidenIntOrFpInductionRecipe>(&R))
       continue;
     auto *Cur = dyn_cast<VPReductionPHIRecipe>(&R);
     if (!Cur) {
diff --git a/llvm/lib/Transforms/Vectorize/VPlanPatternMatch.h b/llvm/lib/Transforms/Vectorize/VPlanPatternMatch.h
index fd5c3e4651beb..90199d7a246fd 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanPatternMatch.h
+++ b/llvm/lib/Transforms/Vectorize/VPlanPatternMatch.h
@@ -253,7 +253,6 @@ struct Recipe_match {
     auto *DefR = dyn_cast<RecipeTy>(R);
     // Check for recipes that do not have opcodes.
     if constexpr (std::is_same<RecipeTy, VPScalarIVStepsRecipe>::value ||
-                  std::is_same<RecipeTy, VPCanonicalIVPHIRecipe>::value ||
                   std::is_same<RecipeTy, VPDerivedIVRecipe>::value ||
                   std::is_same<RecipeTy, VPWidenGEPRecipe>::value)
       return DefR;
diff --git a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
index bd9a93ed57b8a..da2f5dba02448 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
@@ -66,7 +66,6 @@ bool VPRecipeBase::mayWriteToMemory() const {
                 ->onlyReadsMemory();
   case VPWidenIntrinsicSC:
     return cast<VPWidenIntrinsicRecipe>(this)->mayWriteToMemory();
-  case VPCanonicalIVPHISC:
   case VPBranchOnMaskSC:
   case VPFirstOrderRecurrencePHISC:
   case VPReductionPHISC:
@@ -2309,9 +2308,10 @@ bool VPWidenIntOrFpInductionRecipe::isCanonical() const {
     return false;
   auto *StepC = dyn_cast<ConstantInt>(getStepValue()->getLiveInIRValue());
   auto *StartC = dyn_cast<ConstantInt>(getStartValue()->getLiveInIRValue());
-  auto *CanIV = cast<VPCanonicalIVPHIRecipe>(&*getParent()->begin());
+  auto *CanIV = getParent()->getParent()->getCanonicalIV();
   return StartC && StartC->isZero() && StepC && StepC->isOne() &&
-         getScalarType() == CanIV->getScalarType();
+         getScalarType() ==
+             VPTypeAnalysis(*getParent()->getPlan()).inferScalarType(CanIV);
 }
 
 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
@@ -3877,16 +3877,6 @@ InstructionCost VPInterleaveRecipe::computeCost(ElementCount VF,
                                            0);
 }
 
-#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
-void VPCanonicalIVPHIRecipe::print(raw_ostream &O, const Twine &Indent,
-                                   VPSlotTracker &SlotTracker) const {
-  O << Indent << "EMIT ";
-  printAsOperand(O, SlotTracker);
-  O << " = CANONICAL-INDUCTION ";
-  printOperands(O, SlotTracker);
-}
-#endif
-
 bool VPWidenPointerInductionRecipe::onlyScalarsGenerated(bool IsScalable) {
   return IsScalarAfterVectorization &&
          (!IsScalable || vputils::onlyFirstLaneUsed(this));
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index 4e00a39fd0858..abc28f2dffdc9 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -367,8 +367,7 @@ static VPRegionBlock *createReplicateRegion(VPReplicateRecipe *PredRecipe,
   PredRecipe->eraseFromParent();
   auto *Exiting =
       Plan.createVPBasicBlock(Twine(RegionName) + ".continue", PHIRecipe);
-  VPRegionBlock *Region =
-      Plan.createVPRegionBlock(Entry, Exiting, RegionName, true);
+  VPRegionBlock *Region = Plan.createVPRegionBlock(Entry, Exiting, RegionName);
 
   // Note: first set Entry as region entry and then connect successors starting
   // from it in order, to propagate the "parent" of each VPBasicBlock.
@@ -496,7 +495,7 @@ static void removeRedundantInductionCasts(VPlan &Plan) {
 /// Try to replace VPWidenCanonicalIVRecipes with a widened canonical IV
 /// recipe, if it exists.
 static void removeRedundantCanonicalIVs(VPlan &Plan) {
-  VPCanonicalIVPHIRecipe *CanonicalIV = Plan.getCanonicalIV();
+  VPValue *CanonicalIV = Plan.getCanonicalIV();
   VPWidenCanonicalIVRecipe *WidenNewIV = nullptr;
   for (VPUser *U : CanonicalIV->users()) {
     WidenNewIV = dyn_cast<VPWidenCanonicalIVRecipe>(U);
@@ -578,7 +577,7 @@ createScalarIVSteps(VPlan &Plan, InductionDescriptor::InductionKind Kind,
                     VPValue *StartV, VPValue *Step, DebugLoc DL,
                     VPBuilder &Builder) {
   VPBasicBlock *HeaderVPBB = Plan.getVectorLoopRegion()->getEntryBasicBlock();
-  VPCanonicalIVPHIRecipe *CanonicalIV = Plan.getCanonicalIV();
+  VPValue *CanonicalIV = Plan.getCanonicalIV();
   VPSingleDefRecipe *BaseIV = Builder.createDerivedIV(
       Kind, FPBinOp, StartV, CanonicalIV, Step, "offset.idx");
 
@@ -796,7 +795,7 @@ static VPValue *optimizeEarlyExitInductionUser(VPlan &Plan,
 
   // Calculate the final index.
   VPValue *EndValue = Plan.getCanonicalIV();
-  auto CanonicalIVType = Plan.getCanonicalIV()->getScalarType();
+  auto CanonicalIVType = TypeInfo.inferScalarType(EndValue);
   VPBuilder B(cast<VPBasicBlock>(PredVPBB));
 
   DebugLoc DL = cast<VPInstruction>(Op)->getDebugLoc();
@@ -1465,9 +1464,11 @@ static bool isConditionTrueViaVFAndUF(VPValue *Cond, VPlan &Plan,
     });
 
   auto *CanIV = Plan.getCanonicalIV();
-  if (!match(Cond, m_SpecificICmp(CmpInst::ICMP_EQ,
-                                  m_Specific(CanIV->getBackedgeValue()),
-                                  m_Specific(&Plan.getVectorTripCount()))))
+  if (!match(Cond, m_Binary<Instruction::ICmp>(
+                       m_c_Add(m_Specific(CanIV), m_Specific(&Plan.getVFxUF())),
+                       m_Specific(&Plan.getVectorTripCount()))) ||
+      cast<VPRecipeWithIRFlags>(Cond->getDefiningRecipe())->getPredicate() !=
+          CmpInst::ICMP_EQ)
     return false;
 
   // The compare checks CanIV + VFxUF == vector trip count. The vector trip
@@ -1529,8 +1530,8 @@ static bool simplifyBranchConditionForVFAndUF(VPlan &Plan, ElementCount BestVF,
   if (all_of(Header->phis(), [](VPRecipeBase &Phi) {
         if (auto *R = dyn_cast<VPWidenIntOrFpInductionRecipe>(&Phi))
           return R->isCanonical();
-        return isa<VPCanonicalIVPHIRecipe, VPEVLBasedIVPHIRecipe,
-                   VPFirstOrderRecurrencePHIRecipe, VPPhi>(&Phi);
+        return isa<VPEVLBasedIVPHIRecipe, VPFirstOrderRecurrencePHIRecipe,
+                   VPPhi>(&Phi);
       })) {
     for (VPRecipeBase &HeaderR : make_early_inc_range(Header->phis())) {
       if (auto *R = dyn_cast<VPWidenIntOrFpInductionRecipe>(&HeaderR)) {
@@ -1545,6 +1546,9 @@ static bool simplifyBranchConditionForVFAndUF(VPlan &Plan, ElementCount BestVF,
       HeaderR.getVPSingleValue()->replaceAllUsesWith(Phi->getIncomingValue(0));
       HeaderR.eraseFromParent();
     }
+    Plan.getCanonicalIV()->replaceAllUsesWith(
+        Plan.getOrAddLiveIn(ConstantInt::getNullValue(
+            VPTypeAnalysis(Plan).inferScalarType(Plan.getCanonicalIV()))));
 
     VPBlockBase *Preheader = VectorRegion->getSinglePredecessor();
     VPBlockBase *Exit = VectorRegion->getSingleSuccessor();
@@ -2022,15 +2026,18 @@ static VPActiveLaneMaskPHIRecipe *addVPLaneMaskPhiAndUpdateExitBranch(
     VPlan &Plan, bool DataAndControlFlowWithoutRuntimeCheck) {
   VPRegionBlock *TopRegion = Plan.getVectorLoopRegion();
   VPBasicBlock *EB = TopRegion->getExitingBasicBlock();
-  auto *CanonicalIVPHI = Plan.getCanonicalIV();
-  VPValue *StartV = CanonicalIVPHI->getStartValue();
+  VPValue *CanonicalIV = Plan.getCanonicalIV();
+  VPValue *StartV = Plan.getOrAddLiveIn(Constant::getNullValue(
+      VPTypeAnalysis(Plan).inferScalarType(CanonicalIV)));
 
   auto *CanonicalIVIncrement =
-      cast<VPInstruction>(CanonicalIVPHI->getBackedgeValue());
+      cast<VPInstruction>(EB->getTerminator()->getOperand(0));
   // TODO: Check if dropping the flags is needed if
   // !DataAndControlFlowWithoutRuntimeCheck.
   CanonicalIVIncrement->dropPoisonGeneratingFlags();
-  DebugLoc DL = CanonicalIVIncrement->getDebugLoc();
+  auto &CanIVInfo = Plan.getCanonicalIVInfo();
+  CanIVInfo.HasNUW = false;
+  DebugLoc DL = CanIVInfo.DL;
   // We can't use StartV directly in the ActiveLaneMask VPInstruction, since
   // we have to take unrolling into account. Each part needs to start at
   //   Part * VF
@@ -2051,7 +2058,7 @@ static VPActiveLaneMaskPHIRecipe *addVPLaneMaskPhiAndUpdateExitBranch(
     // When avoiding a runtime check, the active.lane.mask inside the loop
     // uses a modified trip count and the induction variable increment is
     // done after the active.lane.mask intrinsic is called.
-    IncrementValue = CanonicalIVPHI;
+    IncrementValue = CanonicalIV;
     TripCount = Builder.createNaryOp(VPInstruction::CalculateTripCountMinusVF,
                                      {TC}, DL);
   }
@@ -2067,7 +2074,8 @@ static VPActiveLaneMaskPHIRecipe *addVPLaneMaskPhiAndUpdateExitBranch(
   // Now create the ActiveLaneMaskPhi recipe in the main loop using the
   // preheader ActiveLaneMask instruction.
   auto *LaneMaskPhi = new VPActiveLaneMaskPHIRecipe(EntryALM, DebugLoc());
-  LaneMaskPhi->insertAfter(CanonicalIVPHI);
+  auto *HeaderVPBB = TopRegion->getEntryBasicBlock();
+  LaneMaskPhi->insertBefore(*HeaderVPBB, HeaderVPBB->begin());
 
   // Create the active lane mask for the next iteration of the loop before the
   // original terminator.
@@ -2372,7 +2380,7 @@ static void transformRecipestoEVLRecipes(VPlan &Plan, VPValue &EVL) {
 
 /// Add a VPEVLBasedIVPHIRecipe and related recipes to \p Plan and
 /// replaces all uses except the canonical IV increment of
-/// VPCanonicalIVPHIRecipe with a VPEVLBasedIVPHIRecipe. VPCanonicalIVPHIRecipe
+/// VPCanonicalIV with a VPEVLBasedIVPHIRecipe. VPCanonicalIV
 /// is used only for loop iterations counting after this transformation.
 ///
 /// The function uses the following definitions:
@@ -2417,13 +2425,18 @@ void VPlanTransforms::addExplicitVectorLength(
     VPlan &Plan, const std::optional<unsigned> &MaxSafeElements) {
   VPBasicBlock *Header = Plan.getVectorLoopRegion()->getEntryBasicBlock();
 
-  auto *CanonicalIVPHI = Plan.getCanonicalIV();
-  auto *CanIVTy = CanonicalIVPHI->getScalarType();
-  VPValue *StartV = CanonicalIVPHI->getStartValue();
+  auto *CanonicalIV = Plan.getCanonicalIV();
+  auto &CanIVInfo = Plan.getCanonicalIVInfo();
+  auto *CanIVTy = VPTypeAnalysis(Plan).inferScalarType(CanonicalIV);
+  VPValue *StartV = Plan.getOrAddLiveIn(ConstantInt::getNullValue(CanIVTy));
+  auto *CanonicalIVIncrement = cast<VPInstruction>(Plan.getVectorLoopRegion()
+                                                       ->getExitingBasicBlock()
+                                                       ->getTerminator()
+                                                       ->getOperand(0));
 
   // Create the ExplicitVectorLengthPhi recipe in the main loop.
   auto *EVLPhi = new VPEVLBasedIVPHIRecipe(StartV, DebugLoc());
-  EVLPhi->insertAfter(CanonicalIVPHI);
+  EVLPhi->insertBefore(*Header, Header->begin());
   VPBuilder Builder(Header, Header->getFirstNonPhi());
   // Create the AVL (application vector length), starting from TC -> 0 in steps
   // of EVL.
@@ -2441,8 +2454,6 @@ void VPlanTransforms::addExplicitVectorLength(
   auto *VPEVL = Builder.createNaryOp(VPInstruction::ExplicitVectorLength, AVL,
                                      DebugLoc());
 
-  auto *CanonicalIVIncrement =
-      cast<VPInstruction>(CanonicalIVPHI->getBackedgeValue());
   Builder.setInsertPoint(CanonicalIVIncrement);
   VPValue *OpVPEVL = VPEVL;
 
@@ -2451,9 +2462,7 @@ void VPlanTransforms::addExplicitVectorLength(
       OpVPEVL, CanIVTy, I32Ty, CanonicalIVIncrement->getDebugLoc());
 
   auto *NextEVLIV = Builder.createOverflowingOp(
-      Instruction::Add, {OpVPEVL, EVLPhi},
-      {CanonicalIVIncrement->hasNoUnsignedWrap(),
-       CanonicalIVIncrement->hasNoSignedWrap()},
+      Instruction::Add, {OpVPEVL, EVLPhi}, {CanIVInfo.HasNUW, false},
       CanonicalIVIncrement->getDebugLoc(), "index.evl.next");
   EVLPhi->addOperand(NextEVLIV);
 
@@ -2464,10 +2473,10 @@ void VPlanTransforms::addExplicitVectorLength(
 
   transformRecipestoEVLRecipes(Plan, *VPEVL);
 
-  // Replace all uses of VPCanonicalIVPHIRecipe by
+  // Replace all uses of VPCanonicalIV by
   // VPEVLBasedIVPHIRecipe except for the canonical IV increment.
-  CanonicalIVPHI->replaceAllUsesWith(EVLPhi);
-  CanonicalIVIncrement->setOperand(0, CanonicalIVPHI);
+  CanonicalIV->replaceAllUsesWith(EVLPhi);
+  CanonicalIVIncrement->setOperand(0, CanonicalIV);
   // TODO: support unroll factor > 1.
   Plan.setUF(1);
 }
@@ -2518,15 +2527,15 @@ void VPlanTransforms::canonicalizeEVLLoops(VPlan &Plan) {
   // Replace CanonicalIVInc with EVL-PHI increment.
   auto *CanonicalIV = cast<VPPhi>(&*HeaderVPBB->begin());
   VPValue *Backedge = CanonicalIV->getIncomingValue(1);
-  assert(match(Backedge, m_c_Add(m_Specific(CanonicalIV),
-                                 m_Specific(&Plan.getVFxUF()))) &&
-         "Unexpected canonical iv");
-  Backedge->replaceAllUsesWith(EVLIncrement);
+  if (match(Backedge,
+            m_c_Add(m_Specific(CanonicalIV), m_Specific(&Plan.getVFxUF())))) {
+    Backedge->replaceAllUsesWith(EVLIncrement);
 
-  // Remove unused phi and increment.
-  VPRecipeBase *CanonicalIVIncrement = Backedge->getDefiningRecipe();
-  CanonicalIVIncrement->eraseFromParent();
-  CanonicalIV->eraseFromParent();
+    // Remove unused phi and increment.
+    VPRecipeBase *CanonicalIVIncrement = Backedge->getDefiningRecipe();
+    CanonicalIVIncrement->eraseFromParent();
+    CanonicalIV->eraseFromParent();
+  }
 
   // Replace the use of VectorTripCount in the latch-exiting block.
   // Before: (branch-on-count EVLIVInc, VectorTripCount)
@@ -3634,8 +3643,7 @@ void VPlanTransforms::narrowInterleaveGroups(VPlan &Plan, ElementCount VF,
   unsigned VFMinVal = VF.getKnownMinValue();
   SmallVector<VPInterleaveRecipe *> StoreGroups;
   for (auto &R : *VectorLoop->getEntryBasicBlock()) {
-    if (isa<VPCanonicalIVPHIRecipe>(&R) ||
-        match(&R, m_BranchOnCount(m_VPValue(), m_VPValue())))
+    if (match(&R, m_BranchOnCount(m_VPValue(), m_VPValue())))
       continue;
 
     if (isa<VPDerivedIVRecipe, VPScalarIVStepsRecipe>(&R) &&
@@ -3785,21 +3793,23 @@ void VPlanTransforms::narrowInterleaveGroups(VPlan &Plan, ElementCount VF,
   // Adjust induction to reflect that the transformed plan only processes one
   // original iteration.
   auto *CanIV = Plan.getCanonicalIV();
-  auto *Inc = cast<VPInstruction>(CanIV->getBackedgeValue());
+  Type *CanIVTy = TypeInfo.inferScalarType(CanIV);
+  auto *Inc = cast<VPInstruction>(
+      VectorLoop->getExitingBasicBlock()->getTerminator()->getOperand(0));
   VPBuilder PHBuilder(Plan.getVectorPreheader());
 
   VPValue *UF = Plan.getOrAddLiveIn(
-      ConstantInt::get(CanIV->getScalarType(), 1 * Plan.getUF()));
+      ConstantInt::get(CanIVTy, 1 * Plan.getUF()));
   if (VF.isScalable()) {
     VPValue *VScale = PHBuilder.createElementCount(
-        CanIV->getScalarType(), ElementCount::getScalable(1));
+        CanIVTy, ElementCount::getScalable(1));
     VPValue *VScaleUF = PHBuilder.createNaryOp(Instruction::Mul, {VScale, UF});
     Inc->setOperand(1, VScaleUF);
     Plan.getVF().replaceAllUsesWith(VScale);
   } else {
     Inc->setOperand(1, UF);
     Plan.getVF().replaceAllUsesWith(
-        Plan.getOrAddLiveIn(ConstantInt::get(CanIV->getScalarType(), 1)));
+        Plan.getOrAddLiveIn(ConstantInt::get(CanIVTy, 1)));
   }
   removeDeadRecipes(Plan);
 }
diff --git a/llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp b/llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp
index 7a63d20825a31..3ac45f327507d 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp
@@ -69,7 +69,7 @@ class UnrollState {
                                 VPBasicBlock::iterator InsertPtForPhi);
 
   VPValue *getConstantVPV(unsigned Part) {
-    Type *CanIVIntTy = Plan.getCanonicalIV()->getScalarType();
+    Type *CanIVIntTy = TypeInfo.inferScalarType(Plan.getCanonicalIV());
     return Plan.getOrAddLiveIn(ConstantInt::get(CanIVIntTy, Part));
   }
 
@@ -79,7 +79,7 @@ class UnrollState {
   void unrollBlock(VPBlockBase *VPB);
 
   VPValue *getValueForPart(VPValue *V, unsigned Part) {
-    if (Part == 0 || V->isLiveIn())
+    if (Part == 0 || V->isLiveIn() || isa<VPRegionValue>(V))
       return V;
     assert((VPV2Parts.contains(V) && VPV2Parts[V].size() >= Part) &&
            "accessed value does not exist");
diff --git a/llvm/lib/Transforms/Vectorize/VPlanUtils.cpp b/llvm/lib/Transforms/Vectorize/VPlanUtils.cpp
index 700a733bf9f2c..6e83ba03ff35e 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanUtils.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanUtils.cpp
@@ -105,9 +105,7 @@ bool vputils::isUniformAcrossVFsAndUFs(VPValue *V) {
     return all_of(R->operands(), isUniformAcrossVFsAndUFs);
   }
 
-  auto *CanonicalIV = R->getParent()->getPlan()->getCanonicalIV();
-  // Canonical IV chain is uniform.
-  if (V == CanonicalIV || V == CanonicalIV->getBackedgeValue())
+  if (isa<VPRegionValue>(V))
     return true;
 
   return TypeSwitch<const VPRecipeBase *, bool>(R)
diff --git a/llvm/lib/Transforms/Vectorize/VPlanValue.h b/llvm/lib/Transforms/Vectorize/VPlanValue.h
index 24f6d61512ef6..7bb739dbc3261 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanValue.h
+++ b/llvm/lib/Transforms/Vectorize/VPlanValue.h
@@ -40,6 +40,7 @@ class VPUser;
 class VPRecipeBase;
 class VPInterleaveRecipe;
 class VPPhiAccessors;
+class VPRegionValue;
 
 // This is the base class of the VPlan Def/Use graph, used for modeling the data
 // flow into, within and out of the VPlan. VPValues can stand for live-ins
@@ -51,6 +52,7 @@ class LLVM_ABI_FOR_TEST VPValue {
   friend class VPInterleaveRecipe;
   friend class VPlan;
   friend class VPExpressionRecipe;
+  friend class VPRegionValue;
 
   const unsigned char SubclassID; ///< Subclass identifier (for isa/dyn_cast).
 
@@ -89,7 +91,9 @@ class LLVM_ABI_FOR_TEST VPValue {
   enum {
     VPValueSC, /// A generic VPValue, like live-in values or defined by a recipe
                /// that defines multiple values.
-    VPVRecipeSC /// A VPValue sub-class that is a VPRecipeBase.
+    VPVRecipeSC,     /// A VPValue sub-class that is a VPRecipeBase.
+    VPRegionValueSC, /// A VPValue sub-class defines the canonical IV of a loop
+                     /// region.
   };
 
   VPValue(const VPValue &) = delete;
@@ -166,7 +170,9 @@ class LLVM_ABI_FOR_TEST VPValue {
   bool hasDefiningRecipe() const { return getDefiningRecipe(); }
 
   /// Returns true if this VPValue is a live-in, i.e. defined outside the VPlan.
-  bool isLiveIn() const { return !hasDefiningRecipe(); }
+  bool isLiveIn() const {
+    return !hasDefiningRecipe() && SubclassID != VPRegionValueSC;
+  }
 
   /// Returns the underlying IR value, if this VPValue is defined outside the
   /// scope of VPlan. Returns nullptr if the VPValue is defined by a VPDef
@@ -187,6 +193,18 @@ class LLVM_ABI_FOR_TEST VPValue {
   }
 };
 
+/// VPValues defined by a VPRegionBlock, like the canonical IV.
+class VPRegionValue : public VPValue {
+public:
+  VPRegionValue() : VPValue(VPValue::VPRegionValueSC) {}
+
+  ~VPRegionValue() override = default;
+
+  static inline bool classof(const VPValue *V) {
+    return V->getVPValueID() == VPValue::VPRegionValueSC;
+  }
+};
+
 typedef DenseMap<Value *, VPValue *> Value2VPValueTy;
 typedef DenseMap<VPValue *, Value *> VPValue2ValueTy;
 
@@ -361,7 +379,6 @@ class VPDef {
     VPPredInstPHISC,
     // START: SubclassID for recipes that inherit VPHeaderPHIRecipe.
     // VPHeaderPHIRecipe need to be kept together.
-    VPCanonicalIVPHISC,
     VPActiveLaneMaskPHISC,
     VPEVLBasedIVPHISC,
     VPFirstOrderRecurrencePHISC,
@@ -371,7 +388,7 @@ class VPDef {
     // END: SubclassID for recipes that inherit VPHeaderPHIRecipe
     // END: Phi-like recipes
     VPFirstPHISC = VPWidenPHISC,
-    VPFirstHeaderPHISC = VPCanonicalIVPHISC,
+    VPFirstHeaderPHISC = VPActiveLaneMaskPHISC,
     VPLastHeaderPHISC = VPReductionPHISC,
     VPLastPHISC = VPReductionPHISC,
   };
diff --git a/llvm/lib/Transforms/Vectorize/VPlanVerifier.cpp b/llvm/lib/Transforms/Vectorize/VPlanVerifier.cpp
index e25ffe135418e..dc0696a5ce844 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanVerifier.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanVerifier.cpp
@@ -458,12 +458,6 @@ bool VPlanVerifier::verify(const VPlan &Plan) {
     return false;
   }
 
-  if (!isa<VPCanonicalIVPHIRecipe>(&*Entry->begin())) {
-    errs() << "VPlan vector loop header does not start with a "
-              "VPCanonicalIVPHIRecipe\n";
-    return false;
-  }
-
   const VPBasicBlock *Exiting = dyn_cast<VPBasicBlock>(TopRegion->getExiting());
   if (!Exiting) {
     errs() << "VPlan exiting block is not a VPBasicBlock\n";
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll b/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll
index aa3d81a19a6d2..4710fa197b871 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll
@@ -579,10 +579,10 @@ define void @multiple_exit_conditions(ptr %src, ptr noalias %dst) #1 {
 ; PRED-NEXT:    [[TMP13:%.*]] = or <vscale x 2 x i16> [[BROADCAST_SPLAT]], splat (i16 1)
 ; PRED-NEXT:    [[TMP14:%.*]] = uitofp <vscale x 2 x i16> [[TMP13]] to <vscale x 2 x double>
 ; PRED-NEXT:    call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP14]], ptr [[NEXT_GEP]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX]], i64 [[TMP10]])
 ; PRED-NEXT:    [[TMP15:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; PRED-NEXT:    [[TMP16:%.*]] = xor i1 [[TMP15]], true
+; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; PRED-NEXT:    br i1 [[TMP16]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    br [[EXIT:label %.*]]
@@ -658,15 +658,15 @@ define void @low_trip_count_fold_tail_scalarized_store(ptr %dst) {
 ; COMMON-NEXT:    store i8 6, ptr [[TMP6]], align 1
 ; COMMON-NEXT:    br label %[[PRED_STORE_CONTINUE12]]
 ; COMMON:       [[PRED_STORE_CONTINUE12]]:
-; COMMON-NEXT:    br i1 false, label %[[PRED_STORE_IF13:.*]], label %[[PRED_STORE_CONTINUE14:.*]]
+; COMMON-NEXT:    br i1 false, label %[[PRED_STORE_IF13:.*]], label %[[EXIT:.*]]
 ; COMMON:       [[PRED_STORE_IF13]]:
 ; COMMON-NEXT:    [[TMP7:%.*]] = getelementptr i8, ptr [[DST]], i64 7
 ; COMMON-NEXT:    store i8 7, ptr [[TMP7]], align 1
-; COMMON-NEXT:    br label %[[PRED_STORE_CONTINUE14]]
-; COMMON:       [[PRED_STORE_CONTINUE14]]:
+; COMMON-NEXT:    br label %[[EXIT]]
+; COMMON:       [[EXIT]]:
 ; COMMON-NEXT:    br label %[[MIDDLE_BLOCK:.*]]
 ; COMMON:       [[MIDDLE_BLOCK]]:
-; COMMON-NEXT:    br [[EXIT:label %.*]]
+; COMMON-NEXT:    br [[EXIT1:label %.*]]
 ; COMMON:       [[SCALAR_PH]]:
 ;
 entry:
@@ -1052,11 +1052,11 @@ define void @test_conditional_interleave_group (ptr noalias %src.1, ptr noalias
 ; PRED-NEXT:    store float 0.000000e+00, ptr [[TMP83]], align 4
 ; PRED-NEXT:    br label %[[PRED_STORE_CONTINUE27]]
 ; PRED:       [[PRED_STORE_CONTINUE27]]:
-; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 8
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <8 x i1> @llvm.get.active.lane.mask.v8i1.i64(i64 [[INDEX]], i64 [[TMP17]])
 ; PRED-NEXT:    [[TMP84:%.*]] = extractelement <8 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; PRED-NEXT:    [[TMP85:%.*]] = xor i1 [[TMP84]], true
 ; PRED-NEXT:    [[VEC_IND_NEXT]] = add <8 x i64> [[VEC_IND]], splat (i64 8)
+; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 8
 ; PRED-NEXT:    br i1 [[TMP85]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    br [[EXIT:label %.*]]
@@ -1347,10 +1347,10 @@ define void @pred_udiv_select_cost(ptr %A, ptr %B, ptr %C, i64 %n, i8 %y) #1 {
 ; PRED-NEXT:    [[TMP26:%.*]] = fptoui <vscale x 16 x float> [[TMP25]] to <vscale x 16 x i8>
 ; PRED-NEXT:    [[TMP27:%.*]] = getelementptr i8, ptr [[C]], i64 [[INDEX]]
 ; PRED-NEXT:    call void @llvm.masked.store.nxv16i8.p0(<vscale x 16 x i8> [[TMP26]], ptr [[TMP27]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]])
-; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP6]]
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 [[INDEX]], i64 [[TMP11]])
 ; PRED-NEXT:    [[TMP28:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; PRED-NEXT:    [[TMP29:%.*]] = xor i1 [[TMP28]], true
+; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP6]]
 ; PRED-NEXT:    br i1 [[TMP29]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    br [[EXIT:label %.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/divs-with-scalable-vfs.ll b/llvm/test/Transforms/LoopVectorize/AArch64/divs-with-scalable-vfs.ll
index a44cc09b8a8ea..65b2c348607ca 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/divs-with-scalable-vfs.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/divs-with-scalable-vfs.ll
@@ -140,11 +140,11 @@ define void @sdiv_feeding_gep_predicated(ptr %dst, i32 %x, i64 %M, i64 %conv6, i
 ; CHECK-NEXT:    [[TMP33:%.*]] = sext i32 [[TMP32]] to i64
 ; CHECK-NEXT:    [[TMP34:%.*]] = getelementptr double, ptr [[DST]], i64 [[TMP33]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> zeroinitializer, ptr [[TMP34]], i32 8, <vscale x 2 x i1> [[TMP23]])
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP6]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX]], i64 [[TMP14]])
 ; CHECK-NEXT:    [[TMP35:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP36:%.*]] = xor i1 [[TMP35]], true
 ; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 2 x i64> [[VEC_IND]], [[DOTSPLAT]]
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP6]]
 ; CHECK-NEXT:    br i1 [[TMP36]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       [[MIDDLE_BLOCK]]:
 ; CHECK-NEXT:    br label %[[EXIT:.*]]
@@ -258,11 +258,11 @@ define void @udiv_urem_feeding_gep(i64 %x, ptr %dst, i64 %N) {
 ; CHECK-NEXT:    [[TMP37:%.*]] = ashr i64 [[TMP36]], 32
 ; CHECK-NEXT:    [[TMP38:%.*]] = getelementptr i64, ptr [[DST]], i64 [[TMP37]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv2i64.p0(<vscale x 2 x i64> [[TMP23]], ptr [[TMP38]], i32 4, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP6]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX]], i64 [[TMP14]])
 ; CHECK-NEXT:    [[TMP39:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP40:%.*]] = xor i1 [[TMP39]], true
 ; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 2 x i64> [[VEC_IND]], [[BROADCAST_SPLAT4]]
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP6]]
 ; CHECK-NEXT:    br i1 [[TMP40]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; CHECK:       [[MIDDLE_BLOCK]]:
 ; CHECK-NEXT:    br label %[[EXIT:.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/fully-unrolled-cost.ll b/llvm/test/Transforms/LoopVectorize/AArch64/fully-unrolled-cost.ll
index c3b0bc8c00a74..bc99d50b7b761 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/fully-unrolled-cost.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/fully-unrolled-cost.ll
@@ -10,10 +10,10 @@ define i64 @test(ptr %a, ptr %b) #0 {
 ; CHECK: Cost of 1 for VF 8: induction instruction   %i.iv.next = add nuw nsw i64 %i.iv, 1
 ; CHECK-NEXT: Cost of 0 for VF 8: induction instruction   %i.iv = phi i64 [ 0, %entry ], [ %i.iv.next, %for.body ]
 ; CHECK-NEXT: Cost of 1 for VF 8: exit condition instruction   %exitcond.not = icmp eq i64 %i.iv.next, 16
-; CHECK-NEXT: Cost of 0 for VF 8: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK-NEXT: Cost of 0 for VF 8: WIDEN-REDUCTION-PHI ir<{{.+}}> = phi vp<{{.+}}>, ir<{{.+}}>
 ; CHECK: Cost for VF 8: 30
 ; CHECK-NEXT: Cost of 0 for VF 16: induction instruction   %i.iv = phi i64 [ 0, %entry ], [ %i.iv.next, %for.body ]
-; CHECK-NEXT: Cost of 0 for VF 16: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK-NEXT: Cost of 0 for VF 16: WIDEN-REDUCTION-PHI ir<{{.+}}> = phi vp<{{.+}}>, ir<{{.+}}>
 ; CHECK: Cost for VF 16: 56
 ; CHECK: LV: Selecting VF: 16
 entry:
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll b/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll
index 137e07336fd50..fa11706a12b0b 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll
@@ -124,10 +124,10 @@ define void @iv_casts(ptr %dst, ptr %src, i32 %x, i64 %N) #0 {
 ; PRED-NEXT:    [[TMP23:%.*]] = trunc <vscale x 16 x i16> [[TMP21]] to <vscale x 16 x i8>
 ; PRED-NEXT:    [[TMP26:%.*]] = getelementptr i8, ptr [[DST]], i64 [[INDEX]]
 ; PRED-NEXT:    call void @llvm.masked.store.nxv16i8.p0(<vscale x 16 x i8> [[TMP23]], ptr [[TMP26]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]])
-; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP5]]
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 [[INDEX]], i64 [[TMP15]])
 ; PRED-NEXT:    [[TMP25:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; PRED-NEXT:    [[TMP27:%.*]] = xor i1 [[TMP25]], true
+; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP5]]
 ; PRED-NEXT:    br i1 [[TMP27]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    br label %[[EXIT:.*]]
@@ -291,11 +291,11 @@ define void @iv_trunc(i32 %x, ptr %dst, i64 %N) #0 {
 ; PRED-NEXT:    store i32 1, ptr [[TMP23]], align 4
 ; PRED-NEXT:    br label %[[PRED_STORE_CONTINUE2]]
 ; PRED:       [[PRED_STORE_CONTINUE2]]:
-; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 2
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <2 x i1> @llvm.get.active.lane.mask.v2i1.i64(i64 [[INDEX]], i64 [[TMP15]])
 ; PRED-NEXT:    [[TMP24:%.*]] = extractelement <2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; PRED-NEXT:    [[TMP25:%.*]] = xor i1 [[TMP24]], true
 ; PRED-NEXT:    [[VEC_IND_NEXT]] = add <2 x i32> [[VEC_IND]], splat (i32 2)
+; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 2
 ; PRED-NEXT:    br i1 [[TMP25]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    br label %[[EXIT:.*]]
@@ -480,11 +480,11 @@ define void @trunc_ivs_and_store(i32 %x, ptr %dst, i64 %N) #0 {
 ; PRED-NEXT:    store i32 [[TMP34]], ptr [[TMP33]], align 4
 ; PRED-NEXT:    br label %[[PRED_STORE_CONTINUE7]]
 ; PRED:       [[PRED_STORE_CONTINUE7]]:
-; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 4
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i64(i64 [[INDEX]], i64 [[TMP16]])
 ; PRED-NEXT:    [[TMP35:%.*]] = extractelement <4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; PRED-NEXT:    [[TMP36:%.*]] = xor i1 [[TMP35]], true
 ; PRED-NEXT:    [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 4)
+; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 4
 ; PRED-NEXT:    br i1 [[TMP36]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    br label %[[EXIT:.*]]
@@ -671,11 +671,11 @@ define void @ivs_trunc_and_ext(i32 %x, ptr %dst, i64 %N) #0 {
 ; PRED-NEXT:    store i32 [[TMP33]], ptr [[TMP32]], align 4
 ; PRED-NEXT:    br label %[[PRED_STORE_CONTINUE6]]
 ; PRED:       [[PRED_STORE_CONTINUE6]]:
-; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 4
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i64(i64 [[INDEX]], i64 [[TMP15]])
 ; PRED-NEXT:    [[TMP34:%.*]] = extractelement <4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; PRED-NEXT:    [[TMP35:%.*]] = xor i1 [[TMP34]], true
 ; PRED-NEXT:    [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 4)
+; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 4
 ; PRED-NEXT:    br i1 [[TMP35]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    br label %[[EXIT:.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/masked-call-scalarize.ll b/llvm/test/Transforms/LoopVectorize/AArch64/masked-call-scalarize.ll
index aed1c3d9fcc4c..53f477e828136 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/masked-call-scalarize.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/masked-call-scalarize.ll
@@ -85,10 +85,10 @@ define void @test_widen_exp_v2(ptr noalias %p2, ptr noalias %p, i64 %n) #5 {
 ; TFCOMMON-NEXT:    store double [[TMP19]], ptr [[P]], align 8
 ; TFCOMMON-NEXT:    br label [[PRED_STORE_CONTINUE6]]
 ; TFCOMMON:       pred.store.continue2:
-; TFCOMMON-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 2
 ; TFCOMMON-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <2 x i1> @llvm.get.active.lane.mask.v2i1.i64(i64 [[INDEX]], i64 [[TMP3]])
 ; TFCOMMON-NEXT:    [[TMP15:%.*]] = extractelement <2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; TFCOMMON-NEXT:    [[TMP17:%.*]] = xor i1 [[TMP15]], true
+; TFCOMMON-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 2
 ; TFCOMMON-NEXT:    br i1 [[TMP17]], label [[END:%.*]], label [[LOOP]], !llvm.loop [[LOOP0:![0-9]+]]
 ; TFCOMMON:       end:
 ; TFCOMMON-NEXT:    ret void
@@ -149,12 +149,12 @@ define void @test_widen_exp_v2(ptr noalias %p2, ptr noalias %p, i64 %n) #5 {
 ; TFA_INTERLEAVE-NEXT:    store double [[TMP34]], ptr [[P]], align 8
 ; TFA_INTERLEAVE-NEXT:    br label [[PRED_STORE_CONTINUE9]]
 ; TFA_INTERLEAVE:       pred.store.continue9:
-; TFA_INTERLEAVE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 4
 ; TFA_INTERLEAVE-NEXT:    [[TMP27:%.*]] = add i64 [[INDEX]], 2
 ; TFA_INTERLEAVE-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <2 x i1> @llvm.get.active.lane.mask.v2i1.i64(i64 [[INDEX]], i64 [[TMP3]])
 ; TFA_INTERLEAVE-NEXT:    [[ACTIVE_LANE_MASK_NEXT10]] = call <2 x i1> @llvm.get.active.lane.mask.v2i1.i64(i64 [[TMP27]], i64 [[TMP3]])
 ; TFA_INTERLEAVE-NEXT:    [[TMP26:%.*]] = extractelement <2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; TFA_INTERLEAVE-NEXT:    [[TMP28:%.*]] = xor i1 [[TMP26]], true
+; TFA_INTERLEAVE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 4
 ; TFA_INTERLEAVE-NEXT:    br i1 [[TMP28]], label [[END:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; TFA_INTERLEAVE:       end:
 ; TFA_INTERLEAVE-NEXT:    ret void
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/masked-call.ll b/llvm/test/Transforms/LoopVectorize/AArch64/masked-call.ll
index 666057b18ccd0..939c2d5225397 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/masked-call.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/masked-call.ll
@@ -992,11 +992,11 @@ define void @test_widen_exp_v2(ptr noalias %p2, ptr noalias %p, i64 %n) #5 {
 ; TFA_INTERLEAVE-NEXT:    store double [[SPEC_SELECT]], ptr [[P]], align 8
 ; TFA_INTERLEAVE-NEXT:    br label %[[TMP13]]
 ; TFA_INTERLEAVE:       [[TMP13]]:
-; TFA_INTERLEAVE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 2
 ; TFA_INTERLEAVE-NEXT:    [[TMP20:%.*]] = add i64 [[INDEX]], 1
 ; TFA_INTERLEAVE-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = icmp ult i64 [[INDEX]], [[TMP3]]
 ; TFA_INTERLEAVE-NEXT:    [[ACTIVE_LANE_MASK_NEXT6]] = icmp ult i64 [[TMP20]], [[TMP3]]
 ; TFA_INTERLEAVE-NEXT:    [[TMP21:%.*]] = xor i1 [[ACTIVE_LANE_MASK_NEXT]], true
+; TFA_INTERLEAVE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 2
 ; TFA_INTERLEAVE-NEXT:    br i1 [[TMP21]], label %[[END:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
 ; TFA_INTERLEAVE:       [[END]]:
 ; TFA_INTERLEAVE-NEXT:    ret void
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/optsize_minsize.ll b/llvm/test/Transforms/LoopVectorize/AArch64/optsize_minsize.ll
index 67e6902b5d32a..2aef013662ad8 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/optsize_minsize.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/optsize_minsize.ll
@@ -384,9 +384,9 @@ define void @tail_predicate_without_optsize(ptr %p, i8 %a, i8 %b, i8 %c, i32 %n)
 ; DEFAULT-NEXT:    store i8 [[TMP71]], ptr [[TMP70]], align 1
 ; DEFAULT-NEXT:    br label %[[PRED_STORE_CONTINUE35]]
 ; DEFAULT:       [[PRED_STORE_CONTINUE35]]:
-; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
 ; DEFAULT-NEXT:    [[VEC_IND_NEXT]] = add <16 x i8> [[VEC_IND]], splat (i8 16)
 ; DEFAULT-NEXT:    [[VEC_IND_NEXT2]] = add <16 x i8> [[VEC_IND1]], splat (i8 16)
+; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
 ; DEFAULT-NEXT:    br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    br label %[[FOR_COND_CLEANUP:.*]]
@@ -523,11 +523,11 @@ define void @sve_tail_predicate_without_minsize(ptr %p, i8 %a, i8 %b, i8 %c, i32
 ; DEFAULT-NEXT:    [[TMP21:%.*]] = add <vscale x 16 x i8> [[TMP18]], [[TMP20]]
 ; DEFAULT-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[INDEX]]
 ; DEFAULT-NEXT:    call void @llvm.masked.store.nxv16i8.p0(<vscale x 16 x i8> [[TMP21]], ptr [[TMP22]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]])
-; DEFAULT-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; DEFAULT-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; DEFAULT-NEXT:    [[TMP24:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; DEFAULT-NEXT:    [[TMP23:%.*]] = xor i1 [[TMP24]], true
 ; DEFAULT-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 16 x i8> [[VEC_IND]], [[DOTSPLAT]]
+; DEFAULT-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; DEFAULT-NEXT:    br i1 [[TMP23]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    br label %[[FOR_COND_CLEANUP:.*]]
@@ -590,11 +590,11 @@ define void @sve_tail_predicate_without_minsize(ptr %p, i8 %a, i8 %b, i8 %c, i32
 ; OPTSIZE-NEXT:    [[TMP21:%.*]] = add <vscale x 16 x i8> [[TMP18]], [[TMP20]]
 ; OPTSIZE-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[INDEX]]
 ; OPTSIZE-NEXT:    call void @llvm.masked.store.nxv16i8.p0(<vscale x 16 x i8> [[TMP21]], ptr [[TMP22]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]])
-; OPTSIZE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; OPTSIZE-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; OPTSIZE-NEXT:    [[TMP24:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; OPTSIZE-NEXT:    [[TMP23:%.*]] = xor i1 [[TMP24]], true
 ; OPTSIZE-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 16 x i8> [[VEC_IND]], [[DOTSPLAT]]
+; OPTSIZE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; OPTSIZE-NEXT:    br i1 [[TMP23]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
 ; OPTSIZE:       [[MIDDLE_BLOCK]]:
 ; OPTSIZE-NEXT:    br label %[[FOR_COND_CLEANUP:.*]]
@@ -657,11 +657,11 @@ define void @sve_tail_predicate_without_minsize(ptr %p, i8 %a, i8 %b, i8 %c, i32
 ; MINSIZE-NEXT:    [[TMP21:%.*]] = add <vscale x 16 x i8> [[TMP18]], [[TMP20]]
 ; MINSIZE-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[INDEX]]
 ; MINSIZE-NEXT:    call void @llvm.masked.store.nxv16i8.p0(<vscale x 16 x i8> [[TMP21]], ptr [[TMP22]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]])
-; MINSIZE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; MINSIZE-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; MINSIZE-NEXT:    [[TMP24:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; MINSIZE-NEXT:    [[TMP23:%.*]] = xor i1 [[TMP24]], true
 ; MINSIZE-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 16 x i8> [[VEC_IND]], [[DOTSPLAT]]
+; MINSIZE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; MINSIZE-NEXT:    br i1 [[TMP23]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
 ; MINSIZE:       [[MIDDLE_BLOCK]]:
 ; MINSIZE-NEXT:    br label %[[FOR_COND_CLEANUP:.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll b/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll
index 792249d7829b0..22215f4c89ebf 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll
@@ -1401,10 +1401,10 @@ define i32 @dotp_predicated(i64 %N, ptr %a, ptr %b) #0 {
 ; CHECK-INTERLEAVE1-NEXT:    [[TMP17:%.*]] = mul nsw <vscale x 4 x i32> [[TMP16]], [[TMP13]]
 ; CHECK-INTERLEAVE1-NEXT:    [[TMP18:%.*]] = add <vscale x 4 x i32> [[TMP17]], [[VEC_PHI]]
 ; CHECK-INTERLEAVE1-NEXT:    [[TMP19]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> [[TMP18]], <vscale x 4 x i32> [[VEC_PHI]]
-; CHECK-INTERLEAVE1-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
 ; CHECK-INTERLEAVE1-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; CHECK-INTERLEAVE1-NEXT:    [[TMP20:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-INTERLEAVE1-NEXT:    [[TMP21:%.*]] = xor i1 [[TMP20]], true
+; CHECK-INTERLEAVE1-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
 ; CHECK-INTERLEAVE1-NEXT:    br i1 [[TMP21]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
 ; CHECK-INTERLEAVE1:       middle.block:
 ; CHECK-INTERLEAVE1-NEXT:    [[TMP22:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP19]])
@@ -1438,10 +1438,10 @@ define i32 @dotp_predicated(i64 %N, ptr %a, ptr %b) #0 {
 ; CHECK-INTERLEAVED-NEXT:    [[TMP17:%.*]] = mul nsw <vscale x 4 x i32> [[TMP16]], [[TMP13]]
 ; CHECK-INTERLEAVED-NEXT:    [[TMP18:%.*]] = add <vscale x 4 x i32> [[TMP17]], [[VEC_PHI]]
 ; CHECK-INTERLEAVED-NEXT:    [[TMP19]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> [[TMP18]], <vscale x 4 x i32> [[VEC_PHI]]
-; CHECK-INTERLEAVED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
 ; CHECK-INTERLEAVED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; CHECK-INTERLEAVED-NEXT:    [[TMP20:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-INTERLEAVED-NEXT:    [[TMP21:%.*]] = xor i1 [[TMP20]], true
+; CHECK-INTERLEAVED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
 ; CHECK-INTERLEAVED-NEXT:    br i1 [[TMP21]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
 ; CHECK-INTERLEAVED:       middle.block:
 ; CHECK-INTERLEAVED-NEXT:    [[TMP22:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP19]])
@@ -1475,10 +1475,10 @@ define i32 @dotp_predicated(i64 %N, ptr %a, ptr %b) #0 {
 ; CHECK-MAXBW-NEXT:    [[TMP17:%.*]] = mul nsw <vscale x 16 x i32> [[TMP16]], [[TMP13]]
 ; CHECK-MAXBW-NEXT:    [[TMP18:%.*]] = select <vscale x 16 x i1> [[ACTIVE_LANE_MASK]], <vscale x 16 x i32> [[TMP17]], <vscale x 16 x i32> zeroinitializer
 ; CHECK-MAXBW-NEXT:    [[PARTIAL_REDUCE]] = call <vscale x 4 x i32> @llvm.experimental.vector.partial.reduce.add.nxv4i32.nxv16i32(<vscale x 4 x i32> [[VEC_PHI]], <vscale x 16 x i32> [[TMP18]])
-; CHECK-MAXBW-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
 ; CHECK-MAXBW-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; CHECK-MAXBW-NEXT:    [[TMP19:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-MAXBW-NEXT:    [[TMP20:%.*]] = xor i1 [[TMP19]], true
+; CHECK-MAXBW-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
 ; CHECK-MAXBW-NEXT:    br i1 [[TMP20]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
 ; CHECK-MAXBW:       middle.block:
 ; CHECK-MAXBW-NEXT:    [[TMP21:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[PARTIAL_REDUCE]])
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/reduction-recurrence-costs-sve.ll b/llvm/test/Transforms/LoopVectorize/AArch64/reduction-recurrence-costs-sve.ll
index 8b2da8c4a7047..6161de23ec432 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/reduction-recurrence-costs-sve.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/reduction-recurrence-costs-sve.ll
@@ -233,10 +233,10 @@ define i32 @chained_recurrences(i32 %x, i64 %y, ptr %src.1, i32 %z, ptr %src.2)
 ; PRED-NEXT:    [[TMP39:%.*]] = or <vscale x 4 x i32> [[WIDE_MASKED_GATHER]], [[VEC_PHI]]
 ; PRED-NEXT:    [[TMP40:%.*]] = or <vscale x 4 x i32> [[TMP39]], [[WIDE_MASKED_GATHER7]]
 ; PRED-NEXT:    [[TMP41]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> [[TMP40]], <vscale x 4 x i32> [[VEC_PHI]]
-; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[IV]], [[TMP2]]
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[IV]], i64 [[TMP10]])
 ; PRED-NEXT:    [[TMP43:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; PRED-NEXT:    [[TMP42:%.*]] = xor i1 [[TMP43]], true
+; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[IV]], [[TMP2]]
 ; PRED-NEXT:    br i1 [[TMP42]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    [[TMP44:%.*]] = call i32 @llvm.vector.reduce.or.nxv4i32(<vscale x 4 x i32> [[TMP41]])
@@ -456,10 +456,10 @@ define i16 @reduce_udiv(ptr %src, i16 %x, i64 %N) #0 {
 ; PRED-NEXT:    [[TMP20:%.*]] = udiv <vscale x 8 x i16> [[WIDE_MASKED_LOAD]], [[BROADCAST_SPLAT]]
 ; PRED-NEXT:    [[TMP21:%.*]] = or <vscale x 8 x i16> [[TMP20]], [[VEC_PHI]]
 ; PRED-NEXT:    [[TMP16]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x i16> [[TMP21]], <vscale x 8 x i16> [[VEC_PHI]]
-; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP2]]
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX]], i64 [[TMP12]])
 ; PRED-NEXT:    [[TMP15:%.*]] = extractelement <vscale x 8 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; PRED-NEXT:    [[TMP17:%.*]] = xor i1 [[TMP15]], true
+; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP2]]
 ; PRED-NEXT:    br i1 [[TMP17]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    [[TMP19:%.*]] = call i16 @llvm.vector.reduce.or.nxv8i16(<vscale x 8 x i16> [[TMP16]])
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/scalable-strict-fadd.ll b/llvm/test/Transforms/LoopVectorize/AArch64/scalable-strict-fadd.ll
index 375e412861777..6310ab4dca889 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/scalable-strict-fadd.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/scalable-strict-fadd.ll
@@ -137,10 +137,10 @@ define float @fadd_strict(ptr noalias nocapture readonly %a, i64 %n) #0 {
 ; CHECK-ORDERED-TF-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 8 x float> @llvm.masked.load.nxv8f32.p0(ptr [[TMP7]], i32 4, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> poison)
 ; CHECK-ORDERED-TF-NEXT:    [[TMP8:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK]], <vscale x 8 x float> [[WIDE_MASKED_LOAD]], <vscale x 8 x float> splat (float -0.000000e+00)
 ; CHECK-ORDERED-TF-NEXT:    [[TMP9]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[VEC_PHI]], <vscale x 8 x float> [[TMP8]])
-; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-ORDERED-TF-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX]], i64 [[TMP6]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP10:%.*]] = extractelement <vscale x 8 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-ORDERED-TF-NEXT:    [[TMP11:%.*]] = xor i1 [[TMP10]], true
+; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-ORDERED-TF-NEXT:    br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK-ORDERED-TF:       middle.block:
 ; CHECK-ORDERED-TF-NEXT:    br label [[FOR_END:%.*]]
@@ -374,7 +374,6 @@ define float @fadd_strict_unroll(ptr noalias nocapture readonly %a, i64 %n) #0 {
 ; CHECK-ORDERED-TF-NEXT:    [[TMP28:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP26]], <vscale x 8 x float> [[TMP27]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP29:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> [[WIDE_MASKED_LOAD11]], <vscale x 8 x float> splat (float -0.000000e+00)
 ; CHECK-ORDERED-TF-NEXT:    [[TMP30]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP28]], <vscale x 8 x float> [[TMP29]])
-; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-ORDERED-TF-NEXT:    [[TMP31:%.*]] = call i64 @llvm.vscale.i64()
 ; CHECK-ORDERED-TF-NEXT:    [[TMP32:%.*]] = shl nuw i64 [[TMP31]], 3
 ; CHECK-ORDERED-TF-NEXT:    [[TMP33:%.*]] = add i64 [[INDEX]], [[TMP32]]
@@ -390,6 +389,7 @@ define float @fadd_strict_unroll(ptr noalias nocapture readonly %a, i64 %n) #0 {
 ; CHECK-ORDERED-TF-NEXT:    [[ACTIVE_LANE_MASK_NEXT14]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP39]], i64 [[TMP6]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP40:%.*]] = extractelement <vscale x 8 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-ORDERED-TF-NEXT:    [[TMP41:%.*]] = xor i1 [[TMP40]], true
+; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-ORDERED-TF-NEXT:    br i1 [[TMP41]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK-ORDERED-TF:       middle.block:
 ; CHECK-ORDERED-TF-NEXT:    br label [[FOR_END:%.*]]
@@ -626,10 +626,10 @@ define void @fadd_strict_interleave(ptr noalias nocapture readonly %a, ptr noali
 ; CHECK-ORDERED-TF-NEXT:    [[TMP14]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP13]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP15:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[TMP11]], <vscale x 4 x float> splat (float -0.000000e+00)
 ; CHECK-ORDERED-TF-NEXT:    [[TMP16]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI1]], <vscale x 4 x float> [[TMP15]])
-; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
 ; CHECK-ORDERED-TF-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP17:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-ORDERED-TF-NEXT:    [[TMP18:%.*]] = xor i1 [[TMP17]], true
+; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
 ; CHECK-ORDERED-TF-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; CHECK-ORDERED-TF:       middle.block:
 ; CHECK-ORDERED-TF-NEXT:    br label [[FOR_END:%.*]]
@@ -859,10 +859,10 @@ define float @fadd_of_sum(ptr noalias nocapture readonly %a, ptr noalias nocaptu
 ; CHECK-ORDERED-TF-NEXT:    [[TMP10:%.*]] = fadd <vscale x 4 x float> [[WIDE_MASKED_LOAD]], [[WIDE_MASKED_LOAD1]]
 ; CHECK-ORDERED-TF-NEXT:    [[TMP11:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[TMP10]], <vscale x 4 x float> splat (float -0.000000e+00)
 ; CHECK-ORDERED-TF-NEXT:    [[TMP12]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP11]])
-; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP2]]
 ; CHECK-ORDERED-TF-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP7]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP13:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-ORDERED-TF-NEXT:    [[TMP14:%.*]] = xor i1 [[TMP13]], true
+; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP2]]
 ; CHECK-ORDERED-TF-NEXT:    br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
 ; CHECK-ORDERED-TF:       middle.block:
 ; CHECK-ORDERED-TF-NEXT:    br label [[FOR_END_LOOPEXIT:%.*]]
@@ -1077,10 +1077,10 @@ define float @fadd_conditional(ptr noalias nocapture readonly %a, ptr noalias no
 ; CHECK-ORDERED-TF-NEXT:    [[PREDPHI:%.*]] = select <vscale x 4 x i1> [[TMP9]], <vscale x 4 x float> [[WIDE_MASKED_LOAD1]], <vscale x 4 x float> splat (float 3.000000e+00)
 ; CHECK-ORDERED-TF-NEXT:    [[TMP11:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[PREDPHI]], <vscale x 4 x float> splat (float -0.000000e+00)
 ; CHECK-ORDERED-TF-NEXT:    [[TMP12]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP11]])
-; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-ORDERED-TF-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP6]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP13:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-ORDERED-TF-NEXT:    [[TMP14:%.*]] = xor i1 [[TMP13]], true
+; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-ORDERED-TF-NEXT:    br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
 ; CHECK-ORDERED-TF:       middle.block:
 ; CHECK-ORDERED-TF-NEXT:    br label [[FOR_END:%.*]]
@@ -1526,7 +1526,6 @@ define float @fmuladd_strict(ptr %a, ptr %b, i64 %n) #0 {
 ; CHECK-ORDERED-TF-NEXT:    [[TMP42:%.*]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP40]], <vscale x 8 x float> [[TMP41]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP43:%.*]] = select <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> [[TMP36]], <vscale x 8 x float> splat (float -0.000000e+00)
 ; CHECK-ORDERED-TF-NEXT:    [[TMP44]] = call float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP42]], <vscale x 8 x float> [[TMP43]])
-; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-ORDERED-TF-NEXT:    [[TMP45:%.*]] = call i64 @llvm.vscale.i64()
 ; CHECK-ORDERED-TF-NEXT:    [[TMP46:%.*]] = shl nuw i64 [[TMP45]], 3
 ; CHECK-ORDERED-TF-NEXT:    [[TMP47:%.*]] = add i64 [[INDEX]], [[TMP46]]
@@ -1542,6 +1541,7 @@ define float @fmuladd_strict(ptr %a, ptr %b, i64 %n) #0 {
 ; CHECK-ORDERED-TF-NEXT:    [[ACTIVE_LANE_MASK_NEXT18]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP53]], i64 [[TMP6]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP54:%.*]] = extractelement <vscale x 8 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-ORDERED-TF-NEXT:    [[TMP55:%.*]] = xor i1 [[TMP54]], true
+; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-ORDERED-TF-NEXT:    br i1 [[TMP55]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
 ; CHECK-ORDERED-TF:       middle.block:
 ; CHECK-ORDERED-TF-NEXT:    br label [[FOR_END:%.*]]
@@ -1836,7 +1836,6 @@ define float @fmuladd_strict_fmf(ptr %a, ptr %b, i64 %n) #0 {
 ; CHECK-ORDERED-TF-NEXT:    [[TMP42:%.*]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP40]], <vscale x 8 x float> [[TMP41]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP43:%.*]] = select nnan <vscale x 8 x i1> [[ACTIVE_LANE_MASK8]], <vscale x 8 x float> [[TMP36]], <vscale x 8 x float> splat (float -0.000000e+00)
 ; CHECK-ORDERED-TF-NEXT:    [[TMP44]] = call nnan float @llvm.vector.reduce.fadd.nxv8f32(float [[TMP42]], <vscale x 8 x float> [[TMP43]])
-; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-ORDERED-TF-NEXT:    [[TMP45:%.*]] = call i64 @llvm.vscale.i64()
 ; CHECK-ORDERED-TF-NEXT:    [[TMP46:%.*]] = shl nuw i64 [[TMP45]], 3
 ; CHECK-ORDERED-TF-NEXT:    [[TMP47:%.*]] = add i64 [[INDEX]], [[TMP46]]
@@ -1852,6 +1851,7 @@ define float @fmuladd_strict_fmf(ptr %a, ptr %b, i64 %n) #0 {
 ; CHECK-ORDERED-TF-NEXT:    [[ACTIVE_LANE_MASK_NEXT18]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[TMP53]], i64 [[TMP6]])
 ; CHECK-ORDERED-TF-NEXT:    [[TMP54:%.*]] = extractelement <vscale x 8 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-ORDERED-TF-NEXT:    [[TMP55:%.*]] = xor i1 [[TMP54]], true
+; CHECK-ORDERED-TF-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-ORDERED-TF-NEXT:    br i1 [[TMP55]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
 ; CHECK-ORDERED-TF:       middle.block:
 ; CHECK-ORDERED-TF-NEXT:    br label [[FOR_END:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/store-costs-sve.ll b/llvm/test/Transforms/LoopVectorize/AArch64/store-costs-sve.ll
index f6de370874d12..5d0cedf2263c3 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/store-costs-sve.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/store-costs-sve.ll
@@ -93,10 +93,10 @@ define void @cost_store_i8(ptr %dst) #0 {
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 16 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; PRED-NEXT:    [[TMP13:%.*]] = getelementptr i8, ptr [[DST]], i64 [[INDEX]]
 ; PRED-NEXT:    call void @llvm.masked.store.nxv16i8.p0(<vscale x 16 x i8> zeroinitializer, ptr [[TMP13]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]])
-; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 [[INDEX]], i64 [[TMP11]])
 ; PRED-NEXT:    [[TMP14:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; PRED-NEXT:    [[TMP12:%.*]] = xor i1 [[TMP14]], true
+; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; PRED-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; PRED:       middle.block:
 ; PRED-NEXT:    br label [[EXIT:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-masked-accesses.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-masked-accesses.ll
index 32235860dd9e2..b5c8e26092dcc 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-masked-accesses.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-masked-accesses.ll
@@ -109,10 +109,10 @@ define dso_local void @masked_strided1(ptr noalias nocapture readonly %p, ptr no
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[INTERLEAVED_VEC:%.*]] = call <vscale x 32 x i8> @llvm.vector.interleave2.nxv32i8(<vscale x 16 x i8> [[TMP15]], <vscale x 16 x i8> [[TMP18]])
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[INTERLEAVED_MASK3:%.*]] = call <vscale x 32 x i1> @llvm.vector.interleave2.nxv32i1(<vscale x 16 x i1> [[TMP9]], <vscale x 16 x i1> [[TMP9]])
 ; PREDICATED_TAIL_FOLDING-NEXT:    call void @llvm.masked.store.nxv32i8.p0(<vscale x 32 x i8> [[INTERLEAVED_VEC]], ptr [[TMP17]], i32 1, <vscale x 32 x i1> [[INTERLEAVED_MASK3]])
-; PREDICATED_TAIL_FOLDING-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], [[TMP1]]
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i32(i32 [[INDEX]], i32 [[TMP6]])
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[TMP19:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 16 x i32> [[VEC_IND]], [[BROADCAST_SPLAT2]]
+; PREDICATED_TAIL_FOLDING-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], [[TMP1]]
 ; PREDICATED_TAIL_FOLDING-NEXT:    br i1 [[TMP19]], label [[VECTOR_BODY]], label [[MIDDLE_BLOCK:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
 ; PREDICATED_TAIL_FOLDING:       middle.block:
 ; PREDICATED_TAIL_FOLDING-NEXT:    br label [[FOR_END:%.*]]
@@ -239,10 +239,10 @@ define dso_local void @masked_strided2(ptr noalias nocapture readnone %p, ptr no
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[TMP14:%.*]] = zext nneg <vscale x 16 x i32> [[TMP13]] to <vscale x 16 x i64>
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[Q]], <vscale x 16 x i64> [[TMP14]]
 ; PREDICATED_TAIL_FOLDING-NEXT:    call void @llvm.masked.scatter.nxv16i8.nxv16p0(<vscale x 16 x i8> splat (i8 2), <vscale x 16 x ptr> [[TMP15]], i32 1, <vscale x 16 x i1> [[TMP12]])
-; PREDICATED_TAIL_FOLDING-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], [[TMP1]]
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i32(i32 [[INDEX]], i32 [[TMP6]])
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[TMP16:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 16 x i32> [[VEC_IND]], [[BROADCAST_SPLAT2]]
+; PREDICATED_TAIL_FOLDING-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], [[TMP1]]
 ; PREDICATED_TAIL_FOLDING-NEXT:    br i1 [[TMP16]], label [[VECTOR_BODY]], label [[MIDDLE_BLOCK:%.*]], !llvm.loop [[LOOP4:![0-9]+]]
 ; PREDICATED_TAIL_FOLDING:       middle.block:
 ; PREDICATED_TAIL_FOLDING-NEXT:    br label [[FOR_END:%.*]]
@@ -373,10 +373,10 @@ define dso_local void @masked_strided3(ptr noalias nocapture readnone %p, ptr no
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[TMP16:%.*]] = zext nneg <vscale x 16 x i32> [[TMP15]] to <vscale x 16 x i64>
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[Q]], <vscale x 16 x i64> [[TMP16]]
 ; PREDICATED_TAIL_FOLDING-NEXT:    call void @llvm.masked.scatter.nxv16i8.nxv16p0(<vscale x 16 x i8> splat (i8 2), <vscale x 16 x ptr> [[TMP17]], i32 1, <vscale x 16 x i1> [[TMP14]])
-; PREDICATED_TAIL_FOLDING-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], [[TMP1]]
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i32(i32 [[INDEX]], i32 [[TMP6]])
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[TMP18:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 16 x i32> [[VEC_IND]], [[BROADCAST_SPLAT4]]
+; PREDICATED_TAIL_FOLDING-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], [[TMP1]]
 ; PREDICATED_TAIL_FOLDING-NEXT:    br i1 [[TMP18]], label [[VECTOR_BODY]], label [[MIDDLE_BLOCK:%.*]], !llvm.loop [[LOOP6:![0-9]+]]
 ; PREDICATED_TAIL_FOLDING:       middle.block:
 ; PREDICATED_TAIL_FOLDING-NEXT:    br label [[FOR_END:%.*]]
@@ -533,10 +533,10 @@ define dso_local void @masked_strided_factor4(ptr noalias nocapture readonly %p,
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[INTERLEAVED_VEC:%.*]] = call <vscale x 64 x i8> @llvm.vector.interleave4.nxv64i8(<vscale x 16 x i8> [[TMP17]], <vscale x 16 x i8> [[TMP18]], <vscale x 16 x i8> [[TMP19]], <vscale x 16 x i8> [[TMP20]])
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[INTERLEAVED_MASK3:%.*]] = call <vscale x 64 x i1> @llvm.vector.interleave4.nxv64i1(<vscale x 16 x i1> [[TMP9]], <vscale x 16 x i1> [[TMP9]], <vscale x 16 x i1> [[TMP9]], <vscale x 16 x i1> [[TMP9]])
 ; PREDICATED_TAIL_FOLDING-NEXT:    call void @llvm.masked.store.nxv64i8.p0(<vscale x 64 x i8> [[INTERLEAVED_VEC]], ptr [[TMP22]], i32 1, <vscale x 64 x i1> [[INTERLEAVED_MASK3]])
-; PREDICATED_TAIL_FOLDING-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], [[TMP1]]
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i32(i32 [[INDEX]], i32 [[TMP6]])
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[TMP23:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
 ; PREDICATED_TAIL_FOLDING-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 16 x i32> [[VEC_IND]], [[BROADCAST_SPLAT2]]
+; PREDICATED_TAIL_FOLDING-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], [[TMP1]]
 ; PREDICATED_TAIL_FOLDING-NEXT:    br i1 [[TMP23]], label [[VECTOR_BODY]], label [[MIDDLE_BLOCK:%.*]], !llvm.loop [[LOOP8:![0-9]+]]
 ; PREDICATED_TAIL_FOLDING:       middle.block:
 ; PREDICATED_TAIL_FOLDING-NEXT:    br label [[FOR_END:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-forced.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-forced.ll
index ad9eefd8ee6a1..54c255f961eee 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-forced.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-forced.ll
@@ -62,10 +62,10 @@ define void @simple_memset(i32 %val, ptr %ptr, i64 %n) #0 {
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 [[INDEX1]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP11]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP14]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP12:%.*]] = xor i1 [[TMP15]], true
+; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP14]]
 ; CHECK-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-overflow-checks.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-overflow-checks.ll
index a22f065415307..5d696d1d96d06 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-overflow-checks.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-overflow-checks.ll
@@ -84,9 +84,9 @@ define void @can_overflow_i64_induction_var(ptr noalias %dst, ptr readonly %src,
 ; CHECK-NEXT:    [[TMP4:%.*]] = add nsw <vscale x 4 x i32> [[WIDE_MASKED_LOAD]], splat (i32 42)
 ; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[DST:%.*]], i64 [[INDEX]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP4]], ptr [[TMP5]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP7]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP2]])
 ; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP7]]
 ; CHECK-NEXT:    br i1 [[TMP8]], label [[VECTOR_BODY]], label [[MIDDLE_BLOCK:%.*]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[FOR_COND_CLEANUP_LOOPEXIT:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-reductions.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-reductions.ll
index 821b9fbbda78f..bb7a78f087110 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-reductions.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-reductions.ll
@@ -29,10 +29,10 @@ define i32 @add_reduction_i32(ptr %ptr, i64 %n) #0 {
 ; CHECK-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr [[TMP11]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> poison)
 ; CHECK-NEXT:    [[TMP13:%.*]] = add <vscale x 4 x i32> [[VEC_PHI]], [[WIDE_MASKED_LOAD]]
 ; CHECK-NEXT:    [[TMP14]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> [[TMP13]], <vscale x 4 x i32> [[VEC_PHI]]
-; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP16]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP18:%.*]] = xor i1 [[TMP17]], true
+; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP16]]
 ; CHECK-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    [[TMP19:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP14]])
@@ -75,10 +75,10 @@ define i32 @add_reduction_i32(ptr %ptr, i64 %n) #0 {
 ; CHECK-IN-LOOP-NEXT:    [[TMP13:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> [[WIDE_MASKED_LOAD]], <vscale x 4 x i32> zeroinitializer
 ; CHECK-IN-LOOP-NEXT:    [[TMP14:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP13]])
 ; CHECK-IN-LOOP-NEXT:    [[TMP15]] = add i32 [[VEC_PHI]], [[TMP14]]
-; CHECK-IN-LOOP-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP17]]
 ; CHECK-IN-LOOP-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-IN-LOOP-NEXT:    [[TMP18:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-IN-LOOP-NEXT:    [[TMP19:%.*]] = xor i1 [[TMP18]], true
+; CHECK-IN-LOOP-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP17]]
 ; CHECK-IN-LOOP-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK-IN-LOOP:       middle.block:
 ; CHECK-IN-LOOP-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
@@ -137,10 +137,10 @@ define float @add_reduction_f32(ptr %ptr, i64 %n) #0 {
 ; CHECK-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x float> @llvm.masked.load.nxv4f32.p0(ptr [[TMP11]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> poison)
 ; CHECK-NEXT:    [[TMP13:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[WIDE_MASKED_LOAD]], <vscale x 4 x float> splat (float -0.000000e+00)
 ; CHECK-NEXT:    [[TMP14]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP13]])
-; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP16]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP18:%.*]] = xor i1 [[TMP17]], true
+; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP16]]
 ; CHECK-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
@@ -181,10 +181,10 @@ define float @add_reduction_f32(ptr %ptr, i64 %n) #0 {
 ; CHECK-IN-LOOP-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x float> @llvm.masked.load.nxv4f32.p0(ptr [[TMP11]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> poison)
 ; CHECK-IN-LOOP-NEXT:    [[TMP13:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[WIDE_MASKED_LOAD]], <vscale x 4 x float> splat (float -0.000000e+00)
 ; CHECK-IN-LOOP-NEXT:    [[TMP14]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP13]])
-; CHECK-IN-LOOP-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP16]]
 ; CHECK-IN-LOOP-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-IN-LOOP-NEXT:    [[TMP17:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-IN-LOOP-NEXT:    [[TMP18:%.*]] = xor i1 [[TMP17]], true
+; CHECK-IN-LOOP-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP16]]
 ; CHECK-IN-LOOP-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK-IN-LOOP:       middle.block:
 ; CHECK-IN-LOOP-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
@@ -247,10 +247,10 @@ define i32 @cond_xor_reduction(ptr noalias %a, ptr noalias %cond, i64 %N) #0 {
 ; CHECK-NEXT:    [[TMP17:%.*]] = xor <vscale x 4 x i32> [[VEC_PHI]], [[WIDE_MASKED_LOAD1]]
 ; CHECK-NEXT:    [[PREDPHI:%.*]] = select <vscale x 4 x i1> [[TMP15]], <vscale x 4 x i32> [[TMP17]], <vscale x 4 x i32> [[VEC_PHI]]
 ; CHECK-NEXT:    [[TMP20]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> [[PREDPHI]], <vscale x 4 x i32> [[VEC_PHI]]
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP22]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP18:%.*]] = xor i1 [[TMP16]], true
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP22]]
 ; CHECK-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    [[TMP25:%.*]] = call i32 @llvm.vector.reduce.xor.nxv4i32(<vscale x 4 x i32> [[TMP20]])
@@ -304,10 +304,10 @@ define i32 @cond_xor_reduction(ptr noalias %a, ptr noalias %cond, i64 %N) #0 {
 ; CHECK-IN-LOOP-NEXT:    [[TMP17:%.*]] = select <vscale x 4 x i1> [[TMP15]], <vscale x 4 x i32> [[WIDE_MASKED_LOAD1]], <vscale x 4 x i32> zeroinitializer
 ; CHECK-IN-LOOP-NEXT:    [[TMP18:%.*]] = call i32 @llvm.vector.reduce.xor.nxv4i32(<vscale x 4 x i32> [[TMP17]])
 ; CHECK-IN-LOOP-NEXT:    [[TMP19]] = xor i32 [[VEC_PHI]], [[TMP18]]
-; CHECK-IN-LOOP-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP21]]
 ; CHECK-IN-LOOP-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; CHECK-IN-LOOP-NEXT:    [[TMP22:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-IN-LOOP-NEXT:    [[TMP23:%.*]] = xor i1 [[TMP22]], true
+; CHECK-IN-LOOP-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP21]]
 ; CHECK-IN-LOOP-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; CHECK-IN-LOOP:       middle.block:
 ; CHECK-IN-LOOP-NEXT:    br label [[FOR_END:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-unroll.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-unroll.ll
index 90224caa68cd4..e1d2a4bbe8cde 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-unroll.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-unroll.ll
@@ -53,7 +53,6 @@ define void @simple_memset(i32 %val, ptr %ptr, i64 %n) #0 {
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP54]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK7]])
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP57]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK8]])
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP60]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK9]])
-; CHECK-NEXT:    [[INDEX_NEXT10]] = add i64 [[INDEX6]], [[TMP62]]
 ; CHECK-NEXT:    [[TMP63:%.*]] = call i64 @llvm.vscale.i64()
 ; CHECK-NEXT:    [[TMP64:%.*]] = shl nuw i64 [[TMP63]], 2
 ; CHECK-NEXT:    [[TMP65:%.*]] = add i64 [[INDEX6]], [[TMP64]]
@@ -69,6 +68,7 @@ define void @simple_memset(i32 %val, ptr %ptr, i64 %n) #0 {
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT13]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[TMP71]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP35:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP36:%.*]] = xor i1 [[TMP35]], true
+; CHECK-NEXT:    [[INDEX_NEXT10]] = add i64 [[INDEX6]], [[TMP62]]
 ; CHECK-NEXT:    br i1 [[TMP36]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
@@ -160,7 +160,6 @@ define void @cond_memset(i32 %val, ptr noalias readonly %cond_ptr, ptr noalias %
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP76]], i32 4, <vscale x 4 x i1> [[TMP70]])
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP79]], i32 4, <vscale x 4 x i1> [[TMP71]])
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP82]], i32 4, <vscale x 4 x i1> [[TMP72]])
-; CHECK-NEXT:    [[INDEX_NEXT13]] = add i64 [[INDEX6]], [[TMP6]]
 ; CHECK-NEXT:    [[TMP85:%.*]] = call i64 @llvm.vscale.i64()
 ; CHECK-NEXT:    [[TMP86:%.*]] = shl nuw i64 [[TMP85]], 2
 ; CHECK-NEXT:    [[TMP87:%.*]] = add i64 [[INDEX6]], [[TMP86]]
@@ -176,6 +175,7 @@ define void @cond_memset(i32 %val, ptr noalias readonly %cond_ptr, ptr noalias %
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT16]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[TMP93]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP66:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP67:%.*]] = xor i1 [[TMP66]], true
+; CHECK-NEXT:    [[INDEX_NEXT13]] = add i64 [[INDEX6]], [[TMP6]]
 ; CHECK-NEXT:    br i1 [[TMP67]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding.ll
index 552d9e23c33c5..e487a91cea5c3 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding.ll
@@ -26,10 +26,10 @@ define void @simple_memset(i32 %val, ptr %ptr, i64 %n) #0 {
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 [[INDEX1]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP11]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP13:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP12:%.*]] = xor i1 [[TMP13]], true
+; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
@@ -69,10 +69,10 @@ define void @simple_memset_v4i32(i32 %val, ptr %ptr, i64 %n) #0 {
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK:%.*]] = phi <4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 [[INDEX1]]
 ; CHECK-NEXT:    call void @llvm.masked.store.v4i32.p0(<4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP4]], i32 4, <4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], 4
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i64(i64 [[INDEX1]], i64 [[TMP2]])
 ; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP5:%.*]] = xor i1 [[TMP6]], true
+; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], 4
 ; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
@@ -116,10 +116,10 @@ define void @simple_memcpy(ptr noalias %dst, ptr noalias %src, i64 %n) #0 {
 ; CHECK-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr [[TMP11]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> poison)
 ; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr i32, ptr [[DST:%.*]], i64 [[INDEX1]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[WIDE_MASKED_LOAD]], ptr [[TMP13]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP14:%.*]] = xor i1 [[TMP12]], true
+; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
@@ -175,11 +175,11 @@ define void @copy_stride4(ptr noalias %dst, ptr noalias %src, i64 %n) #0 {
 ; CHECK-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP19]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> poison)
 ; CHECK-NEXT:    [[TMP20:%.*]] = getelementptr i32, ptr [[DST:%.*]], <vscale x 4 x i64> [[VEC_IND]]
 ; CHECK-NEXT:    call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[WIDE_MASKED_GATHER]], <vscale x 4 x ptr> [[TMP20]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP4]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP12]])
 ; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP22:%.*]] = xor i1 [[TMP21]], true
 ; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
+; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP4]]
 ; CHECK-NEXT:    br i1 [[TMP22]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
@@ -227,10 +227,10 @@ define void @simple_gather_scatter(ptr noalias %dst, ptr noalias %src, ptr noali
 ; CHECK-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP13]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> poison)
 ; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr i32, ptr [[DST:%.*]], <vscale x 4 x i32> [[WIDE_MASKED_LOAD]]
 ; CHECK-NEXT:    call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[WIDE_MASKED_GATHER]], <vscale x 4 x ptr> [[TMP14]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP16:%.*]] = xor i1 [[TMP15]], true
+; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
@@ -280,10 +280,10 @@ define void @uniform_load(ptr noalias %dst, ptr noalias readonly %src, i64 %n) #
 ; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 ; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[DST:%.*]], i64 [[INDEX]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP12]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP13:%.*]] = xor i1 [[TMP14]], true
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[FOR_END:%.*]]
@@ -338,10 +338,10 @@ define void @cond_uniform_load(ptr noalias %dst, ptr noalias readonly %src, ptr
 ; CHECK-NEXT:    [[PREDPHI:%.*]] = select <vscale x 4 x i1> [[TMP15]], <vscale x 4 x i32> [[WIDE_MASKED_GATHER]], <vscale x 4 x i32> zeroinitializer
 ; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[DST:%.*]], i64 [[INDEX1]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[PREDPHI]], ptr [[TMP16]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP18:%.*]] = xor i1 [[TMP17]], true
+; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[FOR_END:%.*]]
@@ -399,10 +399,10 @@ define void @uniform_store(ptr noalias %dst, ptr noalias readonly %src, i64 %n)
 ; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
 ; CHECK-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr [[TMP11]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> poison)
 ; CHECK-NEXT:    call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[WIDE_MASKED_LOAD]], <vscale x 4 x ptr> [[BROADCAST_SPLAT]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP13:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP12:%.*]] = xor i1 [[TMP13]], true
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[FOR_END:%.*]]
@@ -450,10 +450,10 @@ define void @simple_fdiv(ptr noalias %dst, ptr noalias %src, i64 %n) #0 {
 ; CHECK-NEXT:    [[WIDE_MASKED_LOAD2:%.*]] = call <vscale x 4 x float> @llvm.masked.load.nxv4f32.p0(ptr [[TMP12]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> poison)
 ; CHECK-NEXT:    [[TMP15:%.*]] = fdiv <vscale x 4 x float> [[WIDE_MASKED_LOAD]], [[WIDE_MASKED_LOAD2]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4f32.p0(<vscale x 4 x float> [[TMP15]], ptr [[TMP12]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT3]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP13:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP14:%.*]] = xor i1 [[TMP13]], true
+; CHECK-NEXT:    [[INDEX_NEXT3]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
@@ -505,10 +505,10 @@ define void @simple_idiv(ptr noalias %dst, ptr noalias %src, i64 %n) #0 {
 ; CHECK-NEXT:    [[TMP15:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> [[WIDE_MASKED_LOAD2]], <vscale x 4 x i32> splat (i32 1)
 ; CHECK-NEXT:    [[TMP16:%.*]] = udiv <vscale x 4 x i32> [[WIDE_MASKED_LOAD]], [[TMP15]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP16]], ptr [[TMP12]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT3]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP17:%.*]] = xor i1 [[TMP14]], true
+; CHECK-NEXT:    [[INDEX_NEXT3]] = add i64 [[INDEX1]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt.ll
index 3b19e9ee1a5a3..0ae1d7ad41aa2 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt.ll
@@ -555,9 +555,9 @@ define void @simple_histogram_tailfold(ptr noalias %buckets, ptr readonly %indic
 ; CHECK-NEXT:    [[TMP9:%.*]] = zext <vscale x 4 x i32> [[WIDE_LOAD]] to <vscale x 4 x i64>
 ; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP9]]
 ; CHECK-NEXT:    call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP10]], i32 1, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP6]])
 ; CHECK-NEXT:    [[TMP11:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP11]], label [[VECTOR_BODY]], label [[MIDDLE_BLOCK:%.*]], !llvm.loop [[LOOP20:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[FOR_EXIT:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/tail-fold-uniform-memops.ll b/llvm/test/Transforms/LoopVectorize/AArch64/tail-fold-uniform-memops.ll
index 150b79c448005..410d167e75c61 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/tail-fold-uniform-memops.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/tail-fold-uniform-memops.ll
@@ -27,10 +27,10 @@ define void @uniform_load(ptr noalias %dst, ptr noalias readonly %src, i64 %n) #
 ; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <4 x i32> [[TMP4]], <4 x i32> poison, <4 x i32> zeroinitializer
 ; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[IDX]]
 ; CHECK-NEXT:    call void @llvm.masked.store.v4i32.p0(<4 x i32> [[TMP5]], ptr [[TMP6]], i32 4, <4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[IDX]], 4
 ; CHECK-NEXT:    [[NEXT_ACTIVE_LANE_MASK]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i64(i64 [[IDX]], i64 [[N2]])
 ; CHECK-NEXT:    [[EXTRACT_FIRST_LANE_MASK:%.*]] = extractelement <4 x i1> [[NEXT_ACTIVE_LANE_MASK]], i32 0
 ; CHECK-NEXT:    [[FIRST_LANE_SET:%.*]] = xor i1 [[EXTRACT_FIRST_LANE_MASK]], true
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[IDX]], 4
 ; CHECK-NEXT:    br i1 [[FIRST_LANE_SET]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       [[MIDDLE_BLOCK]]:
 ; CHECK-NEXT:    br [[FOR_END:label %.*]]
@@ -81,10 +81,10 @@ define void @cond_uniform_load(ptr noalias nocapture %dst, ptr nocapture readonl
 ; CHECK-NEXT:    [[PREDPHI:%.*]] = select <4 x i1> [[MASK]], <4 x i32> [[WIDE_MASKED_GATHER]], <4 x i32> zeroinitializer
 ; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[INDEX6]]
 ; CHECK-NEXT:    call void @llvm.masked.store.v4i32.p0(<4 x i32> [[PREDPHI]], ptr [[TMP7]], i32 4, <4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX6]], 4
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i64(i64 [[INDEX6]], i64 [[TMP3]])
 ; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; CHECK-NEXT:    [[TMP9:%.*]] = xor i1 [[TMP8]], true
+; CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX6]], 4
 ; CHECK-NEXT:    br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       [[MIDDLE_BLOCK]]:
 ; CHECK-NEXT:    br [[FOR_END:label %.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/tail-folding-styles.ll b/llvm/test/Transforms/LoopVectorize/AArch64/tail-folding-styles.ll
index 24d23d81b1f92..9a2d3a4bdc44b 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/tail-folding-styles.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/tail-folding-styles.ll
@@ -184,10 +184,10 @@ define void @simple_memset_tailfold(i32 %val, ptr %ptr, i64 %n) "target-features
 ; DATA_AND_CONTROL_NO_RT_CHECK-NEXT:    [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; DATA_AND_CONTROL_NO_RT_CHECK-NEXT:    [[TMP11:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 [[INDEX1]]
 ; DATA_AND_CONTROL_NO_RT_CHECK-NEXT:    call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP11]], i32 4, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; DATA_AND_CONTROL_NO_RT_CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP1]]
 ; DATA_AND_CONTROL_NO_RT_CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX1]], i64 [[TMP9]])
 ; DATA_AND_CONTROL_NO_RT_CHECK-NEXT:    [[TMP15:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
 ; DATA_AND_CONTROL_NO_RT_CHECK-NEXT:    [[TMP12:%.*]] = xor i1 [[TMP15]], true
+; DATA_AND_CONTROL_NO_RT_CHECK-NEXT:    [[INDEX_NEXT2]] = add i64 [[INDEX1]], [[TMP1]]
 ; DATA_AND_CONTROL_NO_RT_CHECK-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; DATA_AND_CONTROL_NO_RT_CHECK:       middle.block:
 ; DATA_AND_CONTROL_NO_RT_CHECK-NEXT:    br label [[WHILE_END_LOOPEXIT:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/uniform-args-call-variants.ll b/llvm/test/Transforms/LoopVectorize/AArch64/uniform-args-call-variants.ll
index e9de5e21228fd..9c886a4128c62 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/uniform-args-call-variants.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/uniform-args-call-variants.ll
@@ -25,9 +25,9 @@ define void @test_uniform(ptr noalias %dst, ptr readonly %src, i64 %uniform , i6
 ; CHECK-NEXT:    [[TMP6:%.*]] = call <vscale x 2 x double> @foo_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD]], i64 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
 ; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP6]], ptr [[TMP7]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX]], i64 [[TMP4]])
 ; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP8]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       for.cond.cleanup:
 ; CHECK-NEXT:    ret void
@@ -63,13 +63,13 @@ define void @test_uniform(ptr noalias %dst, ptr readonly %src, i64 %uniform , i6
 ; INTERLEAVE-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP12]], i64 [[DOTIDX5]]
 ; INTERLEAVE-NEXT:    call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP10]], ptr [[TMP12]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
 ; INTERLEAVE-NEXT:    call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP11]], ptr [[TMP14]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
-; INTERLEAVE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; INTERLEAVE-NEXT:    [[TMP15:%.*]] = call i64 @llvm.vscale.i64()
 ; INTERLEAVE-NEXT:    [[TMP16:%.*]] = shl nuw i64 [[TMP15]], 1
 ; INTERLEAVE-NEXT:    [[TMP17:%.*]] = add i64 [[INDEX]], [[TMP16]]
 ; INTERLEAVE-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX]], i64 [[TMP4]])
 ; INTERLEAVE-NEXT:    [[ACTIVE_LANE_MASK_NEXT4]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[TMP17]], i64 [[TMP4]])
 ; INTERLEAVE-NEXT:    [[TMP18:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; INTERLEAVE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; INTERLEAVE-NEXT:    br i1 [[TMP18]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
 ; INTERLEAVE:       for.cond.cleanup:
 ; INTERLEAVE-NEXT:    ret void
@@ -111,9 +111,9 @@ define void @test_uniform_smaller_scalar(ptr noalias %dst, ptr readonly %src, i3
 ; CHECK-NEXT:    [[TMP6:%.*]] = call <vscale x 2 x double> @bar_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD]], i32 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
 ; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
 ; CHECK-NEXT:    call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP6]], ptr [[TMP7]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX]], i64 [[TMP4]])
 ; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; CHECK-NEXT:    br i1 [[TMP8]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP3:![0-9]+]]
 ; CHECK:       for.cond.cleanup:
 ; CHECK-NEXT:    ret void
@@ -149,13 +149,13 @@ define void @test_uniform_smaller_scalar(ptr noalias %dst, ptr readonly %src, i3
 ; INTERLEAVE-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP12]], i64 [[DOTIDX5]]
 ; INTERLEAVE-NEXT:    call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP10]], ptr [[TMP12]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
 ; INTERLEAVE-NEXT:    call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP11]], ptr [[TMP14]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
-; INTERLEAVE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; INTERLEAVE-NEXT:    [[TMP15:%.*]] = call i64 @llvm.vscale.i64()
 ; INTERLEAVE-NEXT:    [[TMP16:%.*]] = shl nuw i64 [[TMP15]], 1
 ; INTERLEAVE-NEXT:    [[TMP17:%.*]] = add i64 [[INDEX]], [[TMP16]]
 ; INTERLEAVE-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX]], i64 [[TMP4]])
 ; INTERLEAVE-NEXT:    [[ACTIVE_LANE_MASK_NEXT4]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[TMP17]], i64 [[TMP4]])
 ; INTERLEAVE-NEXT:    [[TMP18:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; INTERLEAVE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
 ; INTERLEAVE-NEXT:    br i1 [[TMP18]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP3:![0-9]+]]
 ; INTERLEAVE:       for.cond.cleanup:
 ; INTERLEAVE-NEXT:    ret void
@@ -227,10 +227,10 @@ define void @test_uniform_not_invariant(ptr noalias %dst, ptr readonly %src, i64
 ; INTERLEAVE-NEXT:    store double [[TMP8]], ptr [[TMP9]], align 8
 ; INTERLEAVE-NEXT:    br label [[PRED_STORE_CONTINUE4]]
 ; INTERLEAVE:       pred.store.continue4:
-; INTERLEAVE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 2
 ; INTERLEAVE-NEXT:    [[TMP10:%.*]] = or disjoint i64 [[INDEX]], 1
 ; INTERLEAVE-NEXT:    [[ACTIVE_LANE_MASK_NEXT:%.*]] = icmp ult i64 [[INDEX]], [[TMP0]]
 ; INTERLEAVE-NEXT:    [[ACTIVE_LANE_MASK_NEXT5]] = icmp ult i64 [[TMP10]], [[TMP0]]
+; INTERLEAVE-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 2
 ; INTERLEAVE-NEXT:    br i1 [[ACTIVE_LANE_MASK_NEXT]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP4:![0-9]+]]
 ; INTERLEAVE:       for.cond.cleanup:
 ; INTERLEAVE-NEXT:    ret void
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-icmpcost.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-icmpcost.ll
index bdf832f32964f..f92cbc029ffe4 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/mve-icmpcost.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-icmpcost.ll
@@ -22,7 +22,6 @@ target triple = "thumbv8.1m.main-arm-none-eabi"
 ; CHECK: Cost of 1 for VF 2: induction instruction   %inc = add nuw nsw i32 %i.016, 1
 ; CHECK: Cost of 0 for VF 2: induction instruction   %i.016 = phi i32 [ 0, %for.body.lr.ph ], [ %inc, %for.inc ]
 ; CHECK: Cost of 1 for VF 2: exit condition instruction   %exitcond.not = icmp eq i32 %inc, %n
-; CHECK: Cost of 0 for VF 2: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK: Cost of 0 for VF 2: vp<{{.+}}> = SCALAR-STEPS vp<{{.+}}>, ir<1>
 ; CHECK: Cost of 0 for VF 2: CLONE ir<%arrayidx> = getelementptr inbounds ir<%s>, vp<{{.+}}>
 ; CHECK: Cost of 0 for VF 2: vp<{{.+}}> = vector-pointer ir<%arrayidx>
@@ -39,7 +38,6 @@ target triple = "thumbv8.1m.main-arm-none-eabi"
 ; CHECK: Cost of 1 for VF 4: induction instruction   %inc = add nuw nsw i32 %i.016, 1
 ; CHECK: Cost of 0 for VF 4: induction instruction   %i.016 = phi i32 [ 0, %for.body.lr.ph ], [ %inc, %for.inc ]
 ; CHECK: Cost of 1 for VF 4: exit condition instruction   %exitcond.not = icmp eq i32 %inc, %n
-; CHECK: Cost of 0 for VF 4: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK: Cost of 0 for VF 4: vp<{{.+}}> = SCALAR-STEPS vp<{{.+}}>, ir<1>
 ; CHECK: Cost of 0 for VF 4: CLONE ir<%arrayidx> = getelementptr inbounds ir<%s>, vp<{{.+}}>
 ; CHECK: Cost of 0 for VF 4: vp<{{.+}}> = vector-pointer ir<%arrayidx>
@@ -56,7 +54,6 @@ target triple = "thumbv8.1m.main-arm-none-eabi"
 ; CHECK: Cost of 1 for VF 8: induction instruction   %inc = add nuw nsw i32 %i.016, 1
 ; CHECK: Cost of 0 for VF 8: induction instruction   %i.016 = phi i32 [ 0, %for.body.lr.ph ], [ %inc, %for.inc ]
 ; CHECK: Cost of 1 for VF 8: exit condition instruction   %exitcond.not = icmp eq i32 %inc, %n
-; CHECK: Cost of 0 for VF 8: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK: Cost of 0 for VF 8: vp<{{.+}}> = SCALAR-STEPS vp<{{.+}}>, ir<1>
 ; CHECK: Cost of 0 for VF 8: CLONE ir<%arrayidx> = getelementptr inbounds ir<%s>, vp<{{.+}}>
 ; CHECK: Cost of 0 for VF 8: vp<{{.+}}> = vector-pointer ir<%arrayidx>
@@ -135,8 +132,7 @@ for.inc:                                          ; preds = %for.body, %if.then
 ; CHECK: Cost of 0 for VF 2: induction instruction   %incdec.ptr2 = getelementptr inbounds i8, ptr %pSrcB.addr.09, i32 1
 ; CHECK: Cost of 0 for VF 2: induction instruction   %pSrcB.addr.09 = phi ptr [ %incdec.ptr2, %while.body ], [ %pSrcB, %while.body.preheader ]
 ; CHECK: Cost of 1 for VF 2: exit condition instruction   %cmp.not = icmp eq i32 %dec, 0
-; CHECK: Cost of 0 for VF 2: EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
-; CHECK: Cost of 0 for VF 2: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
+; CHECK: Cost of 0 for VF 2: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV:%.+]]>, ir<1>
 ; CHECK: Cost of 0 for VF 2: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<[[STEPS1]]>
 ; CHECK: Cost of 0 for VF 2: vp<[[STEPS2:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
 ; CHECK: Cost of 0 for VF 2: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<[[STEPS2]]>
@@ -167,8 +163,7 @@ for.inc:                                          ; preds = %for.body, %if.then
 ; CHECK: Cost of 0 for VF 4: induction instruction   %incdec.ptr2 = getelementptr inbounds i8, ptr %pSrcB.addr.09, i32 1
 ; CHECK: Cost of 0 for VF 4: induction instruction   %pSrcB.addr.09 = phi ptr [ %incdec.ptr2, %while.body ], [ %pSrcB, %while.body.preheader ]
 ; CHECK: Cost of 1 for VF 4: exit condition instruction   %cmp.not = icmp eq i32 %dec, 0
-; CHECK: Cost of 0 for VF 4: EMIT vp<[[CAN_IV:%.]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
-; CHECK: Cost of 0 for VF 4: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
+; CHECK: Cost of 0 for VF 4: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV:%.+]]>, ir<1>
 ; CHECK: Cost of 0 for VF 4: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<[[STEPS1]]>
 ; CHECK: Cost of 0 for VF 4: vp<[[STEPS2:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
 ; CHECK: Cost of 0 for VF 4: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<[[STEPS2]]>
@@ -199,8 +194,7 @@ for.inc:                                          ; preds = %for.body, %if.then
 ; CHECK: Cost of 0 for VF 8: induction instruction   %incdec.ptr2 = getelementptr inbounds i8, ptr %pSrcB.addr.09, i32 1
 ; CHECK: Cost of 0 for VF 8: induction instruction   %pSrcB.addr.09 = phi ptr [ %incdec.ptr2, %while.body ], [ %pSrcB, %while.body.preheader ]
 ; CHECK: Cost of 1 for VF 8: exit condition instruction   %cmp.not = icmp eq i32 %dec, 0
-; CHECK: Cost of 0 for VF 8: EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
-; CHECK: Cost of 0 for VF 8: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
+; CHECK: Cost of 0 for VF 8: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV:%.+]]>, ir<1>
 ; CHECK: Cost of 0 for VF 8: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<[[STEPS1]]>
 ; CHECK: Cost of 0 for VF 8: vp<[[STEPS2:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
 ; CHECK: Cost of 0 for VF 8: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<[[STEPS2]]>
@@ -231,8 +225,7 @@ for.inc:                                          ; preds = %for.body, %if.then
 ; CHECK: Cost of 0 for VF 16: induction instruction   %incdec.ptr2 = getelementptr inbounds i8, ptr %pSrcB.addr.09, i32 1
 ; CHECK: Cost of 0 for VF 16: induction instruction   %pSrcB.addr.09 = phi ptr [ %incdec.ptr2, %while.body ], [ %pSrcB, %while.body.preheader ]
 ; CHECK: Cost of 1 for VF 16: exit condition instruction   %cmp.not = icmp eq i32 %dec, 0
-; CHECK: Cost of 0 for VF 16: EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
-; CHECK: Cost of 0 for VF 16: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
+; CHECK: Cost of 0 for VF 16: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV:%.+]]>, ir<1>
 ; CHECK: Cost of 0 for VF 16: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<[[STEPS1]]>
 ; CHECK: Cost of 0 for VF 16: vp<[[STEPS2:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
 ; CHECK: Cost of 0 for VF 16: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<[[STEPS2]]>
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/optsize_minsize.ll b/llvm/test/Transforms/LoopVectorize/ARM/optsize_minsize.ll
index dcf4bee728b29..e379420c5cdfa 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/optsize_minsize.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/optsize_minsize.ll
@@ -380,9 +380,9 @@ define void @tail_predicate_without_optsize(ptr %p, i8 %a, i8 %b, i8 %c, i32 %n)
 ; DEFAULT-NEXT:    store i8 [[TMP71]], ptr [[TMP70]], align 1
 ; DEFAULT-NEXT:    br label %[[PRED_STORE_CONTINUE35]]
 ; DEFAULT:       [[PRED_STORE_CONTINUE35]]:
-; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
 ; DEFAULT-NEXT:    [[VEC_IND_NEXT]] = add <16 x i8> [[VEC_IND]], splat (i8 16)
 ; DEFAULT-NEXT:    [[VEC_IND_NEXT2]] = add <16 x i8> [[VEC_IND1]], splat (i8 16)
+; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
 ; DEFAULT-NEXT:    br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    br label %[[FOR_COND_CLEANUP:.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/X86/CostModel/vpinstruction-cost.ll b/llvm/test/Transforms/LoopVectorize/X86/CostModel/vpinstruction-cost.ll
index bb85b88f181f7..d72e33a0355a2 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/CostModel/vpinstruction-cost.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/CostModel/vpinstruction-cost.ll
@@ -10,7 +10,6 @@ define void @wide_or_replaced_with_add_vpinstruction(ptr %src, ptr noalias %dst)
 ; CHECK:  Cost of 1 for VF 2: induction instruction %iv.next = add nuw nsw i64 %iv, 1
 ; CHECK:  Cost of 0 for VF 2: induction instruction %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
 ; CHECK:  Cost of 1 for VF 2: exit condition instruction %exitcond = icmp eq i64 %iv.next, 32
-; CHECK:  Cost of 0 for VF 2: EMIT vp<%3> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK:  Cost of 0 for VF 2: ir<%iv> = WIDEN-INDUCTION ir<0>, ir<1>, vp<%0>
 ; CHECK:  Cost of 0 for VF 2: vp<%4> = SCALAR-STEPS vp<%3>, ir<1>
 ; CHECK:  Cost of 0 for VF 2: CLONE ir<%g.src> = getelementptr inbounds ir<%src>, vp<%4>
@@ -28,7 +27,6 @@ define void @wide_or_replaced_with_add_vpinstruction(ptr %src, ptr noalias %dst)
 ; CHECK:  Cost of 1 for VF 4: induction instruction %iv.next = add nuw nsw i64 %iv, 1
 ; CHECK:  Cost of 0 for VF 4: induction instruction %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
 ; CHECK:  Cost of 1 for VF 4: exit condition instruction %exitcond = icmp eq i64 %iv.next, 32
-; CHECK:  Cost of 0 for VF 4: EMIT vp<%3> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK:  Cost of 0 for VF 4: ir<%iv> = WIDEN-INDUCTION ir<0>, ir<1>, vp<%0>
 ; CHECK:  Cost of 0 for VF 4: vp<%4> = SCALAR-STEPS vp<%3>, ir<1>
 ; CHECK:  Cost of 0 for VF 4: CLONE ir<%g.src> = getelementptr inbounds ir<%src>, vp<%4>
diff --git a/llvm/test/Transforms/LoopVectorize/iv-select-cmp-decreasing.ll b/llvm/test/Transforms/LoopVectorize/iv-select-cmp-decreasing.ll
index 615f50124b41d..85f0e326b9663 100644
--- a/llvm/test/Transforms/LoopVectorize/iv-select-cmp-decreasing.ll
+++ b/llvm/test/Transforms/LoopVectorize/iv-select-cmp-decreasing.ll
@@ -449,8 +449,8 @@ define i16 @select_decreasing_induction_icmp_table_i16(i16 noundef %val) {
 ; IC4VF4-NEXT:    [[TMP113:%.*]] = select <4 x i1> [[TMP1]], <4 x i16> [[TMP109]], <4 x i16> [[VEC_PHI1]]
 ; IC4VF4-NEXT:    [[TMP114:%.*]] = select <4 x i1> [[TMP2]], <4 x i16> [[TMP110]], <4 x i16> [[VEC_PHI2]]
 ; IC4VF4-NEXT:    [[TMP115:%.*]] = select <4 x i1> [[TMP3]], <4 x i16> [[TMP111]], <4 x i16> [[VEC_PHI3]]
-; IC4VF4-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 16
 ; IC4VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i16> [[STEP_ADD_3]], splat (i16 -4)
+; IC4VF4-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 16
 ; IC4VF4-NEXT:    br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; IC4VF4:       [[MIDDLE_BLOCK]]:
 ; IC4VF4-NEXT:    [[RDX_MINMAX:%.*]] = call <4 x i16> @llvm.smin.v4i16(<4 x i16> [[TMP112]], <4 x i16> [[TMP113]])
@@ -814,8 +814,8 @@ define i16 @select_decreasing_induction_icmp_table_half(half noundef %val) {
 ; IC4VF4-NEXT:    [[TMP113:%.*]] = select <4 x i1> [[TMP1]], <4 x i16> [[TMP109]], <4 x i16> [[VEC_PHI1]]
 ; IC4VF4-NEXT:    [[TMP114:%.*]] = select <4 x i1> [[TMP2]], <4 x i16> [[TMP110]], <4 x i16> [[VEC_PHI2]]
 ; IC4VF4-NEXT:    [[TMP115:%.*]] = select <4 x i1> [[TMP3]], <4 x i16> [[TMP111]], <4 x i16> [[VEC_PHI3]]
-; IC4VF4-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 16
 ; IC4VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i16> [[STEP_ADD_3]], splat (i16 -4)
+; IC4VF4-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 16
 ; IC4VF4-NEXT:    br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; IC4VF4:       [[MIDDLE_BLOCK]]:
 ; IC4VF4-NEXT:    [[RDX_MINMAX:%.*]] = call <4 x i16> @llvm.smin.v4i16(<4 x i16> [[TMP112]], <4 x i16> [[TMP113]])
diff --git a/llvm/test/Transforms/LoopVectorize/load-deref-pred-poison-ub-ops-feeding-pointer.ll b/llvm/test/Transforms/LoopVectorize/load-deref-pred-poison-ub-ops-feeding-pointer.ll
index 70e730f0284c0..2f580956731ed 100644
--- a/llvm/test/Transforms/LoopVectorize/load-deref-pred-poison-ub-ops-feeding-pointer.ll
+++ b/llvm/test/Transforms/LoopVectorize/load-deref-pred-poison-ub-ops-feeding-pointer.ll
@@ -63,8 +63,8 @@ define void @ptr_depends_on_sdiv(ptr noalias %dst, i16 noundef %off) {
 ; CHECK-NEXT:    store i64 [[TMP15]], ptr [[TMP17]], align 1
 ; CHECK-NEXT:    br label %[[PRED_STORE_CONTINUE4]]
 ; CHECK:       [[PRED_STORE_CONTINUE4]]:
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i16> [[VEC_IND]], splat (i16 2)
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       [[MIDDLE_BLOCK]]:
 ; CHECK-NEXT:    br [[EXIT:label %.*]]
@@ -139,8 +139,8 @@ define void @ptr_depends_on_possibly_poison_value(ptr noalias %dst, i16 %off) {
 ; CHECK-NEXT:    store i64 [[TMP13]], ptr [[TMP12]], align 1
 ; CHECK-NEXT:    br label %[[PRED_STORE_CONTINUE2]]
 ; CHECK:       [[PRED_STORE_CONTINUE2]]:
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i16> [[VEC_IND]], splat (i16 2)
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       [[MIDDLE_BLOCK]]:
 ; CHECK-NEXT:    br [[EXIT:label %.*]]
@@ -209,8 +209,8 @@ define void @ptr_doesnt_depend_on_poison_or_ub(ptr noalias %dst, i16 noundef %of
 ; CHECK-NEXT:    store i64 [[TMP13]], ptr [[TMP12]], align 1
 ; CHECK-NEXT:    br label %[[PRED_STORE_CONTINUE2]]
 ; CHECK:       [[PRED_STORE_CONTINUE2]]:
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i16> [[VEC_IND]], splat (i16 2)
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; CHECK:       [[MIDDLE_BLOCK]]:
 ; CHECK-NEXT:    br [[EXIT:label %.*]]
@@ -285,8 +285,8 @@ define void @ptr_depends_on_possibly_poison_value_from_load(ptr noalias %dst) {
 ; CHECK-NEXT:    store i64 [[TMP14]], ptr [[TMP13]], align 1
 ; CHECK-NEXT:    br label %[[PRED_STORE_CONTINUE2]]
 ; CHECK:       [[PRED_STORE_CONTINUE2]]:
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i16> [[VEC_IND]], splat (i16 2)
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
 ; CHECK:       [[MIDDLE_BLOCK]]:
 ; CHECK-NEXT:    br [[EXIT:label %.*]]
@@ -358,8 +358,8 @@ define void @ptr_depends_on_noundef_load(ptr noalias %dst) {
 ; CHECK-NEXT:    store i64 [[TMP14]], ptr [[TMP13]], align 1
 ; CHECK-NEXT:    br label %[[PRED_STORE_CONTINUE2]]
 ; CHECK:       [[PRED_STORE_CONTINUE2]]:
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i16> [[VEC_IND]], splat (i16 2)
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
 ; CHECK:       [[MIDDLE_BLOCK]]:
 ; CHECK-NEXT:    br [[EXIT:label %.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/vplan-printing.ll b/llvm/test/Transforms/LoopVectorize/vplan-printing.ll
index 87b11dc77d417..c176d69d4ae47 100644
--- a/llvm/test/Transforms/LoopVectorize/vplan-printing.ll
+++ b/llvm/test/Transforms/LoopVectorize/vplan-printing.ll
@@ -21,8 +21,9 @@ define void @print_call_and_memory(i64 %n, ptr noalias %y, ptr noalias %x) nounw
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT: vector.body:
-; CHECK-NEXT:   EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:   vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:   CLONE ir<%arrayidx> = getelementptr inbounds ir<%y>, vp<[[STEPS]]>
 ; CHECK-NEXT:   vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%arrayidx>
@@ -31,7 +32,7 @@ define void @print_call_and_memory(i64 %n, ptr noalias %y, ptr noalias %x) nounw
 ; CHECK-NEXT:   CLONE ir<%arrayidx2> = getelementptr inbounds ir<%x>, vp<[[STEPS]]>
 ; CHECK-NEXT:   vp<[[VEC_PTR2:%.+]]> = vector-pointer ir<%arrayidx2>
 ; CHECK-NEXT:   WIDEN store vp<[[VEC_PTR2]]>, ir<%call>
-; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:   EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT: No successors
 ; CHECK-NEXT: }
@@ -89,8 +90,9 @@ define void @print_widen_gep_and_select(i64 %n, ptr noalias %y, ptr noalias %x,
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT: vector.body:
-; CHECK-NEXT:   EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:   ir<%iv> = WIDEN-INDUCTION ir<0>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:   vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
 ; CHECK-NEXT:   WIDEN-GEP Inv[Var] ir<%arrayidx> = getelementptr inbounds ir<%y>, ir<%iv>
@@ -102,7 +104,7 @@ define void @print_widen_gep_and_select(i64 %n, ptr noalias %y, ptr noalias %x,
 ; CHECK-NEXT:   CLONE ir<%arrayidx2> = getelementptr inbounds ir<%x>, vp<[[STEPS]]>
 ; CHECK-NEXT:   vp<[[VEC_PTR2:%.+]]> = vector-pointer ir<%arrayidx2>
 ; CHECK-NEXT:   WIDEN store vp<[[VEC_PTR2]]>, ir<%add>
-; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:   EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT: No successors
 ; CHECK-NEXT: }
@@ -163,8 +165,9 @@ define void @print_replicate_predicated_phi(i64 %n, ptr %x) {
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT: vector.body:
-; CHECK-NEXT:   EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:   ir<%i> = WIDEN-INDUCTION ir<0>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:   vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
 ; CHECK-NEXT:   WIDEN ir<%cmp> = icmp ult ir<%i>, ir<5>
@@ -191,7 +194,7 @@ define void @print_replicate_predicated_phi(i64 %n, ptr %x) {
 ; CHECK-NEXT:   CLONE ir<%idx> = getelementptr ir<%x>, vp<[[STEPS]]>
 ; CHECK-NEXT:   vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%idx>
 ; CHECK-NEXT:   WIDEN store vp<[[VEC_PTR]]>, ir<%d>
-; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:   EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT: No successors
 ; CHECK-NEXT: }
@@ -258,8 +261,9 @@ define void @print_interleave_groups(i32 %C, i32 %D) {
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT:  vector.body:
-; CHECK-NEXT:   EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:   vp<[[DERIVED_IV:%.+]]> = DERIVED-IV ir<0> + vp<[[CAN_IV]]> * ir<4>
 ; CHECK-NEXT:   vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[DERIVED_IV]]>, ir<4>, vp<[[VF]]>
 ; CHECK-NEXT:   CLONE ir<%gep.AB.0> = getelementptr inbounds ir<@AB>, ir<0>, vp<[[STEPS]]>
@@ -274,7 +278,7 @@ define void @print_interleave_groups(i32 %C, i32 %D) {
 ; CHECK-NEXT:     store ir<1> to index 1
 ; CHECK-NEXT:     store ir<2> to index 2
 ; CHECK-NEXT:     store ir<%AB.3> to index 3
-; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:   EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT: No successors
 ; CHECK-NEXT: }
@@ -344,8 +348,9 @@ define void @debug_loc_vpinstruction(ptr nocapture %asd, ptr nocapture %bsd) !db
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT:  vector.body:
-; CHECK-NEXT:    EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:    vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:    CLONE ir<%isd> = getelementptr inbounds ir<%asd>, vp<[[STEPS]]>
 ; CHECK-NEXT:    vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%isd>
@@ -379,7 +384,7 @@ define void @debug_loc_vpinstruction(ptr nocapture %asd, ptr nocapture %bsd) !db
 ; CHECK-NEXT:    BLEND ir<%ysd.0> = vp<[[PHI]]> ir<%psd>/vp<[[SEL2]]>
 ; CHECK-NEXT:    vp<[[VEC_PTR2:%.+]]> = vector-pointer ir<%isd>
 ; CHECK-NEXT:    WIDEN store vp<[[VEC_PTR2]]>, ir<%ysd.0>
-; CHECK-NEXT:    EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:    EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:    EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT:  No successors
 ; CHECK-NEXT:}
@@ -456,15 +461,16 @@ define void @print_expand_scev(i64 %y, ptr %ptr) {
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT:   vector.body:
-; CHECK-NEXT:     EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:     ir<%iv> = WIDEN-INDUCTION ir<0>, vp<[[EXP_SCEV]]>, vp<[[VF]]> (truncated to i8)
 ; CHECK-NEXT:     vp<[[DERIVED_IV:%.+]]> = DERIVED-IV ir<0> + vp<[[CAN_IV]]> * vp<[[EXP_SCEV]]>
 ; CHECK-NEXT:     vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[DERIVED_IV]]>, vp<[[EXP_SCEV]]>
 ; CHECK-NEXT:     WIDEN ir<%v3> = add nuw ir<%iv>, ir<1>
 ; CHECK-NEXT:     REPLICATE ir<%gep> = getelementptr inbounds ir<%ptr>, vp<[[STEPS]]>
 ; CHECK-NEXT:     REPLICATE store ir<%v3>, ir<%gep>
-; CHECK-NEXT:     EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:     EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:     EMIT branch-on-count  vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT:   No successors
 ; CHECK-NEXT: }
@@ -523,15 +529,16 @@ define i32 @print_exit_value(ptr %ptr, i32 %off) {
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT:   vector.body:
-; CHECK-NEXT:    EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:     ir<%iv> = WIDEN-INDUCTION ir<0>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:     vp<[[STEPS:%.+]]>    = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
 ; CHECK-NEXT:     CLONE ir<%gep> = getelementptr inbounds ir<%ptr>, vp<[[STEPS]]>
 ; CHECK-NEXT:     WIDEN ir<%add> = add ir<%iv>, ir<%off>
 ; CHECK-NEXT:     vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%gep>
 ; CHECK-NEXT:     WIDEN store vp<[[VEC_PTR]]>, ir<0>
-; CHECK-NEXT:     EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:     EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:     EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT:   No successors
 ; CHECK-NEXT: }
@@ -589,8 +596,9 @@ define void @print_fast_math_flags(i64 %n, ptr noalias %y, ptr noalias %x, ptr %
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT: vector.body:
-; CHECK-NEXT:   EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:   vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:   CLONE ir<%gep.y> = getelementptr inbounds ir<%y>, vp<[[STEPS]]>
 ; CHECK-NEXT:   vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%gep.y>
@@ -601,7 +609,7 @@ define void @print_fast_math_flags(i64 %n, ptr noalias %y, ptr noalias %x, ptr %
 ; CHECK-NEXT:   CLONE ir<%gep.x> = getelementptr inbounds ir<%x>, vp<[[STEPS]]>
 ; CHECK-NEXT:   vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%gep.x>
 ; CHECK-NEXT:   WIDEN store vp<[[VEC_PTR]]>, ir<%div>
-; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:   EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT: No successors
 ; CHECK-NEXT: }
@@ -660,8 +668,9 @@ define void @print_exact_flags(i64 %n, ptr noalias %x) {
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT: vector.body:
-; CHECK-NEXT:   EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:   vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:   CLONE ir<%gep.x> = getelementptr inbounds ir<%x>, vp<[[STEPS]]>
 ; CHECK-NEXT:   vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%gep.x>
@@ -671,7 +680,7 @@ define void @print_exact_flags(i64 %n, ptr noalias %x) {
 ; CHECK-NEXT:   WIDEN ir<%add> = add nuw nsw ir<%div.1>, ir<%div.2>
 ; CHECK-NEXT:   vp<[[VEC_PTR2:%.+]]> = vector-pointer ir<%gep.x>
 ; CHECK-NEXT:   WIDEN store vp<[[VEC_PTR2]]>, ir<%add>
-; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:   EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT: No successors
 ; CHECK-NEXT: }
@@ -729,8 +738,9 @@ define void @print_call_flags(ptr readonly %src, ptr noalias %dest, i64 %n) {
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT: vector.body:
-; CHECK-NEXT:   EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:   vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:   CLONE ir<%ld.addr> = getelementptr inbounds ir<%src>, vp<[[STEPS]]>
 ; CHECK-NEXT:   vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%ld.addr>
@@ -761,7 +771,7 @@ define void @print_call_flags(ptr readonly %src, ptr noalias %dest, i64 %n) {
 ; CHECK-NEXT:    CLONE ir<%st.addr> = getelementptr inbounds ir<%dest>, vp<[[STEPS]]>
 ; CHECK-NEXT:    vp<[[VEC_PTR2:%.+]]> = vector-pointer ir<%st.addr>
 ; CHECK-NEXT:    WIDEN store vp<[[VEC_PTR2]]>, ir<%st.value>
-; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:   EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT: No successors
 ; CHECK-NEXT: }
@@ -829,8 +839,9 @@ define void @print_disjoint_flags(i64 %n, ptr noalias %x) {
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT: vector.body:
-; CHECK-NEXT:   EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:   vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:   CLONE ir<%gep.x> = getelementptr inbounds ir<%x>, vp<[[STEPS]]>
 ; CHECK-NEXT:   vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%gep.x>
@@ -840,7 +851,7 @@ define void @print_disjoint_flags(i64 %n, ptr noalias %x) {
 ; CHECK-NEXT:   WIDEN ir<%add> = add nuw nsw ir<%or.1>, ir<%or.2>
 ; CHECK-NEXT:   vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%gep.x>
 ; CHECK-NEXT:   WIDEN store vp<[[VEC_PTR]]>, ir<%add>
-; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
+; CHECK-NEXT:   EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>, vp<[[VFxUF]]>
 ; CHECK-NEXT:   EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT: No successors
 ; CHECK-NEXT: }
@@ -898,15 +909,16 @@ define void @zext_nneg(ptr noalias %p, ptr noalias %p1) {
 ; CHECK-NEXT:  Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT:  <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT:  vector.body:
-; CHECK-NEXT:    EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<[[CAN_IV_NEXT:%.+]]>
 ; CHECK-NEXT:    vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:    CLONE ir<%idx> = getelementptr ir<%p>, vp<[[STEPS]]>
 ; CHECK-NEXT:    vp<[[VEC_PTR:%.+]]> = vector-pointer ir<%idx>
 ; CHECK-NEXT:    WIDEN ir<%l> = load vp<[[VEC_PTR]]>
 ; CHECK-NEXT:    WIDEN-CAST ir<%zext> = zext nneg ir<%l>
 ; CHECK-NEXT:    REPLICATE store ir<%zext>, ir<%p1>
-; CHECK-NEXT:    EMIT vp<[[CAN_IV_NEXT]]> = add nuw vp<[[CAN_IV]]>
+; CHECK-NEXT:    EMIT vp<[[CAN_IV_NEXT:%.+]]> = add nuw vp<[[CAN_IV]]>
 ; CHECK-NEXT:    EMIT branch-on-count vp<[[CAN_IV_NEXT]]>, vp<[[VTC]]>
 ; CHECK-NEXT:  No successors
 ; CHECK-NEXT: }
@@ -943,8 +955,9 @@ define i16 @print_first_order_recurrence_and_result(ptr %ptr) {
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK-NEXT: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[CAN_IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT:   vector.body:
-; CHECK-NEXT:     EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION
 ; CHECK-NEXT:     FIRST-ORDER-RECURRENCE-PHI ir<%for.1> = phi ir<22>, ir<%for.1.next>
 ; CHECK-NEXT:     vp<[[STEPS:%.+]]>    = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:     CLONE ir<%gep.ptr> = getelementptr inbounds ir<%ptr>, vp<[[STEPS]]>
@@ -1016,8 +1029,9 @@ define void @print_select_with_fastmath_flags(ptr noalias %a, ptr noalias %b, pt
 ; CHECK-NEXT: Successor(s): vector loop
 ; CHECK-EMPTY:
 ; CHECK: <x1> vector loop: {
+; CHECK-NEXT:   vp<[[IV:%.+]]> = CANONICAL-IV ir<0>
+; CHECK-EMPTY:
 ; CHECK-NEXT:   vector.body:
-; CHECK-NEXT:     EMIT vp<[[IV:%.+]]> =  CANONICAL-INDUCTION ir<0>, vp<[[IV_NEXT_EXIT:%.+]]>
 ; CHECK-NEXT:     vp<[[ST:%.+]]> = SCALAR-STEPS vp<[[IV]]>, ir<1>, vp<[[VF]]>
 ; CHECK-NEXT:     CLONE ir<[[GEP1:%.+]]> = getelementptr inbounds nuw ir<%b>, vp<[[ST]]>
 ; CHECK-NEXT:     vp<[[PTR1:%.+]]> = vector-pointer ir<[[GEP1]]>
@@ -1031,7 +1045,7 @@ define void @print_select_with_fastmath_flags(ptr noalias %a, ptr noalias %b, pt
 ; CHECK-NEXT:     CLONE ir<[[GEP3:%.+]]> = getelementptr inbounds nuw ir<%a>, vp<[[ST]]>
 ; CHECK-NEXT:     vp<[[PTR3:%.+]]> = vector-pointer ir<[[GEP3]]>
 ; CHECK-NEXT:     WIDEN store vp<[[PTR3]]>, ir<[[SELECT]]>
-; CHECK-NEXT:     EMIT vp<[[IV_NEXT_EXIT]]> = add nuw vp<[[IV]]>, vp<[[VFUF]]>
+; CHECK-NEXT:     EMIT vp<[[IV_NEXT_EXIT:%.+]]> = add nuw vp<[[IV]]>, vp<[[VFUF]]>
 ; CHECK-NEXT:     EMIT branch-on-count vp<[[IV_NEXT_EXIT]]>, vp<[[VTC]]>
 ; CHECK-NEXT:   No successors
 ; CHECK-NEXT: }
diff --git a/llvm/unittests/Transforms/Vectorize/VPlanHCFGTest.cpp b/llvm/unittests/Transforms/Vectorize/VPlanHCFGTest.cpp
index a943e7ac12b1b..91168d6de21fe 100644
--- a/llvm/unittests/Transforms/Vectorize/VPlanHCFGTest.cpp
+++ b/llvm/unittests/Transforms/Vectorize/VPlanHCFGTest.cpp
@@ -57,9 +57,7 @@ TEST_F(VPlanHCFGTest, testBuildHCFGInnerLoop) {
   EXPECT_EQ(&*Plan, VecBB->getPlan());
 
   auto Iter = VecBB->begin();
-  auto *CanIV = dyn_cast<VPCanonicalIVPHIRecipe>(&*Iter++);
-  EXPECT_NE(nullptr, CanIV);
-  auto *Phi = dyn_cast<VPPhi>(&*Iter++);
+  VPWidenPHIRecipe *Phi = dyn_cast<VPWidenPHIRecipe>(&*Iter++);
   EXPECT_NE(nullptr, Phi);
 
   VPInstruction *Idx = dyn_cast<VPInstruction>(&*Iter++);
@@ -218,7 +216,6 @@ TEST_F(VPlanHCFGTest, testVPInstructionToVPRecipesInner) {
   EXPECT_EQ(VecBB->getParent()->getEntryBasicBlock(), VecBB);
 
   auto Iter = VecBB->begin();
-  EXPECT_NE(nullptr, dyn_cast<VPCanonicalIVPHIRecipe>(&*Iter++));
   EXPECT_NE(nullptr, dyn_cast<VPWidenPHIRecipe>(&*Iter++));
   EXPECT_NE(nullptr, dyn_cast<VPWidenGEPRecipe>(&*Iter++));
   EXPECT_NE(nullptr, dyn_cast<VPWidenMemoryRecipe>(&*Iter++));
diff --git a/llvm/unittests/Transforms/Vectorize/VPlanPatternMatchTest.cpp b/llvm/unittests/Transforms/Vectorize/VPlanPatternMatchTest.cpp
index e38b4fad80b0e..7fdeb7328bee7 100644
--- a/llvm/unittests/Transforms/Vectorize/VPlanPatternMatchTest.cpp
+++ b/llvm/unittests/Transforms/Vectorize/VPlanPatternMatchTest.cpp
@@ -23,32 +23,31 @@ using VPPatternMatchTest = VPlanTestBase;
 
 TEST_F(VPPatternMatchTest, ScalarIVSteps) {
   VPlan &Plan = getPlan();
+  IntegerType *I64Ty = IntegerType::get(C, 64);
+  VPRegionBlock *VPR =
+      Plan.createVPRegionBlock(DebugLoc::getCompilerGenerated(), "");
+  VPValue *CanIV = VPR->getCanonicalIV();
   VPBasicBlock *VPBB = Plan.createVPBasicBlock("");
   VPBuilder Builder(VPBB);
 
-  IntegerType *I64Ty = IntegerType::get(C, 64);
-  VPValue *StartV = Plan.getOrAddLiveIn(ConstantInt::get(I64Ty, 0));
-  auto *CanonicalIVPHI = new VPCanonicalIVPHIRecipe(StartV, DebugLoc());
-  Builder.insert(CanonicalIVPHI);
-
   VPValue *Inc = Plan.getOrAddLiveIn(ConstantInt::get(I64Ty, 1));
   VPValue *VF = &Plan.getVF();
-  VPValue *Steps = Builder.createScalarIVSteps(
-      Instruction::Add, nullptr, CanonicalIVPHI, Inc, VF, DebugLoc());
+  VPValue *Steps = Builder.createScalarIVSteps(Instruction::Add, nullptr, CanIV,
+                                               Inc, VF, DebugLoc());
 
   VPValue *Inc2 = Plan.getOrAddLiveIn(ConstantInt::get(I64Ty, 2));
-  VPValue *Steps2 = Builder.createScalarIVSteps(
-      Instruction::Add, nullptr, CanonicalIVPHI, Inc2, VF, DebugLoc());
+  VPValue *Steps2 = Builder.createScalarIVSteps(Instruction::Add, nullptr,
+                                                CanIV, Inc2, VF, DebugLoc());
 
   using namespace VPlanPatternMatch;
 
-  ASSERT_TRUE(match(Steps, m_ScalarIVSteps(m_Specific(CanonicalIVPHI),
-                                           m_SpecificInt(1), m_Specific(VF))));
+  ASSERT_TRUE(match(Steps, m_ScalarIVSteps(m_Specific(CanIV), m_SpecificInt(1),
+                                           m_Specific(VF))));
   ASSERT_FALSE(
-      match(Steps2, m_ScalarIVSteps(m_Specific(CanonicalIVPHI),
-                                    m_SpecificInt(1), m_Specific(VF))));
-  ASSERT_TRUE(match(Steps2, m_ScalarIVSteps(m_Specific(CanonicalIVPHI),
-                                            m_SpecificInt(2), m_Specific(VF))));
+      match(Steps2, m_ScalarIVSteps(m_Specific(CanIV), m_SpecificInt(1),
+                                    m_Specific(VF))));
+  ASSERT_TRUE(match(Steps2, m_ScalarIVSteps(m_Specific(CanIV), m_SpecificInt(2),
+                                            m_Specific(VF))));
 }
 
 } // namespace
diff --git a/llvm/unittests/Transforms/Vectorize/VPlanVerifierTest.cpp b/llvm/unittests/Transforms/Vectorize/VPlanVerifierTest.cpp
index c2f045bf524e9..b4de7b8780e3b 100644
--- a/llvm/unittests/Transforms/Vectorize/VPlanVerifierTest.cpp
+++ b/llvm/unittests/Transforms/Vectorize/VPlanVerifierTest.cpp
@@ -24,14 +24,12 @@ TEST_F(VPVerifierTest, VPInstructionUseBeforeDefSameBB) {
   VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0));
   VPInstruction *DefI = new VPInstruction(Instruction::Add, {Zero});
   VPInstruction *UseI = new VPInstruction(Instruction::Sub, {DefI});
-  auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {});
 
   VPBasicBlock *VPBB1 = Plan.getEntry();
   VPBB1->appendRecipe(UseI);
   VPBB1->appendRecipe(DefI);
 
   VPBasicBlock *VPBB2 = Plan.createVPBasicBlock("");
-  VPBB2->appendRecipe(CanIV);
   VPRegionBlock *R1 = Plan.createVPRegionBlock(VPBB2, VPBB2, "R1");
   VPBlockUtils::connectBlocks(VPBB1, R1);
   VPBlockUtils::connectBlocks(R1, Plan.getScalarHeader());
@@ -59,15 +57,13 @@ TEST_F(VPVerifierTest, VPInstructionUseBeforeDefDifferentBB) {
   VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0));
   VPInstruction *DefI = new VPInstruction(Instruction::Add, {Zero});
   VPInstruction *UseI = new VPInstruction(Instruction::Sub, {DefI});
-  auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {});
   VPInstruction *BranchOnCond =
-      new VPInstruction(VPInstruction::BranchOnCond, {CanIV});
+      new VPInstruction(VPInstruction::BranchOnCond, {UseI});
 
   VPBasicBlock *VPBB1 = Plan.getEntry();
   VPBasicBlock *VPBB2 = Plan.createVPBasicBlock("");
 
   VPBB1->appendRecipe(UseI);
-  VPBB2->appendRecipe(CanIV);
   VPBB2->appendRecipe(DefI);
   VPBB2->appendRecipe(BranchOnCond);
 
@@ -100,9 +96,8 @@ TEST_F(VPVerifierTest, VPBlendUseBeforeDefDifferentBB) {
   VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Int32, 0));
 
   VPInstruction *DefI = new VPInstruction(Instruction::Add, {Zero});
-  auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {});
   VPInstruction *BranchOnCond =
-      new VPInstruction(VPInstruction::BranchOnCond, {CanIV});
+      new VPInstruction(VPInstruction::BranchOnCond, {DefI});
   auto *Blend = new VPBlendRecipe(Phi, {DefI}, {});
 
   VPBasicBlock *VPBB1 = Plan.getEntry();
@@ -110,7 +105,6 @@ TEST_F(VPVerifierTest, VPBlendUseBeforeDefDifferentBB) {
   VPBasicBlock *VPBB3 = Plan.createVPBasicBlock("");
   VPBasicBlock *VPBB4 = Plan.createVPBasicBlock("");
 
-  VPBB2->appendRecipe(CanIV);
   VPBB3->appendRecipe(Blend);
   VPBB4->appendRecipe(DefI);
   VPBB4->appendRecipe(BranchOnCond);
@@ -157,8 +151,6 @@ TEST_F(VPVerifierTest, VPPhiIncomingValueDoesntDominateIncomingBlock) {
   VPPhi *Phi = new VPPhi({DefI}, {});
   VPBB2->appendRecipe(Phi);
   VPBB2->appendRecipe(DefI);
-  auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {});
-  VPBB3->appendRecipe(CanIV);
 
   VPRegionBlock *R1 = Plan.createVPRegionBlock(VPBB3, VPBB3, "R1");
   VPBlockUtils::connectBlocks(VPBB1, VPBB2);
@@ -186,9 +178,8 @@ TEST_F(VPVerifierTest, DuplicateSuccessorsOutsideRegion) {
   VPlan &Plan = getPlan();
   VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0));
   VPInstruction *I1 = new VPInstruction(Instruction::Add, {Zero});
-  auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {});
   VPInstruction *BranchOnCond =
-      new VPInstruction(VPInstruction::BranchOnCond, {CanIV});
+      new VPInstruction(VPInstruction::BranchOnCond, {I1});
   VPInstruction *BranchOnCond2 =
       new VPInstruction(VPInstruction::BranchOnCond, {I1});
 
@@ -197,7 +188,6 @@ TEST_F(VPVerifierTest, DuplicateSuccessorsOutsideRegion) {
 
   VPBB1->appendRecipe(I1);
   VPBB1->appendRecipe(BranchOnCond2);
-  VPBB2->appendRecipe(CanIV);
   VPBB2->appendRecipe(BranchOnCond);
 
   VPRegionBlock *R1 = Plan.createVPRegionBlock(VPBB2, VPBB2, "R1");
@@ -220,9 +210,8 @@ TEST_F(VPVerifierTest, DuplicateSuccessorsInsideRegion) {
   VPlan &Plan = getPlan();
   VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0));
   VPInstruction *I1 = new VPInstruction(Instruction::Add, {Zero});
-  auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {});
   VPInstruction *BranchOnCond =
-      new VPInstruction(VPInstruction::BranchOnCond, {CanIV});
+      new VPInstruction(VPInstruction::BranchOnCond, {I1});
   VPInstruction *BranchOnCond2 =
       new VPInstruction(VPInstruction::BranchOnCond, {I1});
 
@@ -231,7 +220,6 @@ TEST_F(VPVerifierTest, DuplicateSuccessorsInsideRegion) {
   VPBasicBlock *VPBB3 = Plan.createVPBasicBlock("");
 
   VPBB1->appendRecipe(I1);
-  VPBB2->appendRecipe(CanIV);
   VPBB2->appendRecipe(BranchOnCond2);
   VPBB3->appendRecipe(BranchOnCond);
 
@@ -260,8 +248,6 @@ TEST_F(VPVerifierTest, BlockOutsideRegionWithParent) {
   VPBasicBlock *VPBB2 = Plan.createVPBasicBlock("");
 
   VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0));
-  auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {});
-  VPBB2->appendRecipe(CanIV);
 
   VPInstruction *DefI = new VPInstruction(Instruction::Add, {Zero});
   VPInstruction *BranchOnCond =
@@ -289,14 +275,11 @@ TEST_F(VPVerifierTest, BlockOutsideRegionWithParent) {
 TEST_F(VPVerifierTest, NonHeaderPHIInHeader) {
   VPlan &Plan = getPlan();
   VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0));
-  auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {});
-  auto *BranchOnCond = new VPInstruction(VPInstruction::BranchOnCond, {CanIV});
+  auto *BranchOnCond = new VPInstruction(VPInstruction::BranchOnCond, {Zero});
 
   VPBasicBlock *VPBB1 = Plan.getEntry();
   VPBasicBlock *VPBB2 = Plan.createVPBasicBlock("header");
 
-  VPBB2->appendRecipe(CanIV);
-
   PHINode *PHINode = PHINode::Create(Type::getInt32Ty(C), 2);
   auto *IRPhi = new VPIRPhi(*PHINode);
   VPBB2->appendRecipe(IRPhi);



More information about the llvm-commits mailing list