[LV] Skip VFs < iterations remaining for epilogue vectorization.

[fhahn]

Summary

If a candidate VF for epilogue vectorization is less than the number of
remaining iterations, the epilogue loop would be dead. Skip such factors.

[Ayal]

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

5679

nit (unrelated): avoid else after return.

5708–5712

nit: can set Type *TCType = Legal->getWidestInductionType(); and use it below.

5721

Worth early-continuing first if (a) !hasPlanWithVF, then if (b) NextVF >= EstimatedRuntimeVF, and last if (c) NextVF >= RemainingIterations?

Note that for IC==1 and non scalable VF's, check (c) subsumes check (b).

Can the checks below be simplified, given that EstimatedRuntimeVF = MainLoopVF if !MainLoopVF.isScalable()?
Instead of checking if Result.Width.isScalar() better check if Result is still uninitialized, i.e., == VectorizationFactor::Disabled()? Or teach isMoreProfitable() to prefer any computed cost over an uncomputed one.

llvm/test/Transforms/LoopVectorize/X86/gather_scatter.ll

866

nit: these UGLY-to-SCEV changes are unrelated.

llvm/test/Transforms/LoopVectorize/X86/limit-vf-by-tripcount.ll

7

This patch addresses this TODO?

llvm/test/Transforms/LoopVectorize/X86/pr42674.ll

8

Note that the change in generated code for this test appears to be indirectly related to fixing the VF selected for the epilog loop.
This loop is vectorized with VF=64, UF=2 both w/ and w/o the patch, yet w/o the patch its epilog loop is also vectorized (with VF=32), which is cleaned up by instcombine and simplifycfg, but the main loop itself remains. Now the epilog loop is not vectorized and the main single-vectorized-iteration loop gets cleaned up.

[fhahn]

Address latest comments, thanks!

[Ayal]

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

5707

While we're here, should EstimatedRuntimeVF (or rather, EstimatedMainLoopVF) be used for both scalable and non-scalable MainLoopVF below?
Are all 4 combinations of main loop and epilog loop VF's being scalable or non-scalable, possible?
Can the filtering of VF's larger than EstimatedMainLoopVF be simplified into a single ElementCount::isKnownGE(NextVF.Width, EstimatedMainLoopVF) check for all combinations?
Is the SCEV RemainingIterations check relevant only for the {non-scalable & non-scalable} combination, or worth leaving a TODO?

5724

nit: reverse the condition to use ICMP_UGE as done above?

[fhahn]

Use UGT instead of ULT as suggeted, thanks!

[fhahn]

Add TODO for scalable vectors.

[fhahn]

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

5679

Will adjust separately.

5707

While we're here, should EstimatedRuntimeVF (or rather, EstimatedMainLoopVF) be used for both scalable and non-scalable MainLoopVF below?
Are all 4 combinations of main loop and epilog loop VF's being scalable or non-scalable, possible?
Can the filtering of VF's larger than EstimatedMainLoopVF be simplified into a single ElementCount::isKnownGE(NextVF.Width, EstimatedMainLoopVF) check for all combinations?

In theory I think yes, all 4 combinations are possible

I tried just checking EstimatedRuntimeVF and not MainVF but there are some SVE tests that are failing.

Is the SCEV RemainingIterations check relevant only for the {non-scalable & non-scalable} combination, or worth leaving a TODO?

I added a TODO. I think for scalable vectors we would also need to estimate the runtime VFs and check those.

5708–5712

Done, thanks!

5721

Applied the first refactoring in b4efc0f070ba, update this patch, thanks! Will do the refactorings mentioned in the last paragraph separately.

5724

updated, thanks!

[Ayal]

This looks good to me, comment still needs to be fixed.

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

5724

nit: comment still needs to be fixed.

llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h

#endif
   static bool
   getDecisionAndClampRange(const std::function<bool(ElementCount)> &Predicate,
                            VFRange &Range);
 
   /// \return The most profitable vectorization factor and the cost of that VF
   /// for vectorizing the epilogue. Returns VectorizationFactor::Disabled if
   /// epilogue vectorization is not supported for the loop.
   VectorizationFactor
-  selectEpilogueVectorizationFactor(const ElementCount MaxVF);
+  selectEpilogueVectorizationFactor(const ElementCount MaxVF, unsigned IC);
 
 protected:
   /// Build VPlans for power-of-2 VF's between \p MinVF and \p MaxVF inclusive,
   /// according to the information gathered by Legal when it checked if it is
   /// legal to vectorize the loop.
   void buildVPlans(ElementCount MinVF, ElementCount MaxVF);
 
 private:

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

bool LoopVectorizationCostModel::isEpilogueVectorizationProfitable(
   if (VF.isScalable())
     Multiplier = getVScaleForTuning(TheLoop, TTI).value_or(1);
   if ((Multiplier * VF.getKnownMinValue()) >= EpilogueVectorizationMinVF)
     return true;
   return false;
 }
 
 VectorizationFactor LoopVectorizationPlanner::selectEpilogueVectorizationFactor(
-    const ElementCount MainLoopVF) {
+    const ElementCount MainLoopVF, unsigned IC) {
   VectorizationFactor Result = VectorizationFactor::Disabled();
   if (!EnableEpilogueVectorization) {
     LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization is disabled.\n");
     return Result;
   }
 
   if (!CM.isScalarEpilogueAllowed()) {
     LLVM_DEBUG(dbgs() << "LEV: Unable to vectorize epilogue because no "
if (!isCandidateForEpilogueVectorization(MainLoopVF)) {
     return Result;
   }
 
   if (EpilogueVectorizationForceVF > 1) {
     LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization factor is forced.\n");
     ElementCount ForcedEC = ElementCount::getFixed(EpilogueVectorizationForceVF);
     if (hasPlanWithVF(ForcedEC))
       return {ForcedEC, 0, 0};
     else {
       LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization forced factor is not "
                            "viable.\n");
       return Result;
     }
   }
 
   if (OrigLoop->getHeader()->getParent()->hasOptSize() ||
       OrigLoop->getHeader()->getParent()->hasMinSize()) {
VectorizationFactor LoopVectorizationPlanner::selectEpilogueVectorizationFactor(
   // If MainLoopVF = vscale x 2, and vscale is expected to be 4, then we know
   // the main loop handles 8 lanes per iteration. We could still benefit from
   // vectorizing the epilogue loop with VF=4.
   ElementCount EstimatedRuntimeVF = MainLoopVF;
   if (MainLoopVF.isScalable()) {
     EstimatedRuntimeVF = ElementCount::getFixed(MainLoopVF.getKnownMinValue());
     if (std::optional<unsigned> VScale = getVScaleForTuning(OrigLoop, TTI))
       EstimatedRuntimeVF *= *VScale;
   }
 
+  ScalarEvolution &SE = *PSE.getSE();
+  Type *TCType = Legal->getWidestInductionType();
+  const SCEV *RemainingIterations = nullptr;
   for (auto &NextVF : ProfitableVFs) {
     // Skip candidate VFs without a corresponding VPlan.
     if (!hasPlanWithVF(NextVF.Width))
       continue;
 
     // Skip candidate VFs with widths >= the estimate runtime VF (scalable
     // vectors) or the VF of the main loop (fixed vectors).
     if ((!NextVF.Width.isScalable() && MainLoopVF.isScalable() &&
          ElementCount::isKnownGE(NextVF.Width, EstimatedRuntimeVF)) ||
         ElementCount::isKnownGE(NextVF.Width, MainLoopVF))
       continue;
 
+    // If NextVF is less than the number of remaining iterations, the epilogue
+    // loop would be dead. Skip such factors.
+    if (!MainLoopVF.isScalable() && !NextVF.Width.isScalable()) {
+      if (!RemainingIterations) {
+        const SCEV *TC = createTripCountSCEV(TCType, PSE, OrigLoop);
+        RemainingIterations = SE.getURemExpr(
+            TC, SE.getConstant(TCType, MainLoopVF.getKnownMinValue() * IC));
+      }
+      if (SE.isKnownPredicate(
+              CmpInst::ICMP_ULT, RemainingIterations,
+              SE.getConstant(TCType, NextVF.Width.getKnownMinValue())))
+        continue;
+    }
+
     if (Result.Width.isScalar() || isMoreProfitable(NextVF, Result))
       Result = NextVF;
   }
 
   if (Result != VectorizationFactor::Disabled())
     LLVM_DEBUG(dbgs() << "LEV: Vectorizing epilogue loop with VF = "
                       << Result.Width << "\n");
   return Result;
if (!VectorizeLoop) {
                << "interleaved loop (interleaved count: "
                << NV("InterleaveCount", IC) << ")";
       });
     } else {
       // If we decided that it is *legal* to vectorize the loop, then do it.
 
       // Consider vectorizing the epilogue too if it's profitable.
       VectorizationFactor EpilogueVF =
-          LVP.selectEpilogueVectorizationFactor(VF.Width);
+          LVP.selectEpilogueVectorizationFactor(VF.Width, IC);
       if (EpilogueVF.Width.isVector()) {
 
         // The first pass vectorizes the main loop and creates a scalar epilogue
         // to be vectorized by executing the plan (potentially with a different
         // factor) again shortly afterwards.
         EpilogueLoopVectorizationInfo EPI(VF.Width, IC, EpilogueVF.Width, 1);
         EpilogueVectorizerMainLoop MainILV(L, PSE, LI, DT, TLI, TTI, AC, ORE,
                                            EPI, &LVL, &CM, BFI, PSI, Checks);

llvm/test/Transforms/LoopVectorize/X86/gather_scatter.ll

 ;      out[i] = in[index[i]] + (float) 0.5;
 ;    }
 ;  }
 ;}
 
 ; Function Attrs: nounwind uwtable
 define void @foo1(ptr noalias %in, ptr noalias %out, ptr noalias %trigger, ptr noalias %index) {
 ; AVX512-LABEL: @foo1(
-; AVX512-NEXT:  iter.check:
+; AVX512-NEXT:  entry:
 ; AVX512-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; AVX512:       vector.body:
-; AVX512-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[ITER_CHECK:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; AVX512-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX1]], 0
 ; AVX512-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TRIGGER:%.*]], i64 [[TMP0]]
 ; AVX512-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
 ; AVX512-NEXT:    [[WIDE_LOAD:%.*]] = load <16 x i32>, ptr [[TMP2]], align 4
 ; AVX512-NEXT:    [[TMP3:%.*]] = icmp sgt <16 x i32> [[WIDE_LOAD]], zeroinitializer
 ; AVX512-NEXT:    [[TMP4:%.*]] = getelementptr i32, ptr [[INDEX:%.*]], i64 [[TMP0]]
 ; AVX512-NEXT:    [[TMP5:%.*]] = getelementptr i32, ptr [[TMP4]], i32 0
 ; AVX512-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr [[TMP5]], i32 4, <16 x i1> [[TMP3]], <16 x i32> poison)

 ;    }
 ;  }
 ;}
 
 %struct.In = type { float, float }
 
 define void @foo2(ptr noalias %in, ptr noalias %out, ptr noalias %trigger, ptr noalias %index) #0 {
 ; AVX512-LABEL: @foo2(
-; AVX512-NEXT:  iter.check:
+; AVX512-NEXT:  entry:
 ; AVX512-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; AVX512:       vector.body:
-; AVX512-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[ITER_CHECK:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; AVX512-NEXT:    [[VEC_IND:%.*]] = phi <16 x i64> [ <i64 0, i64 16, i64 32, i64 48, i64 64, i64 80, i64 96, i64 112, i64 128, i64 144, i64 160, i64 176, i64 192, i64 208, i64 224, i64 240>, [[ITER_CHECK]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[VEC_IND:%.*]] = phi <16 x i64> [ <i64 0, i64 16, i64 32, i64 48, i64 64, i64 80, i64 96, i64 112, i64 128, i64 144, i64 160, i64 176, i64 192, i64 208, i64 224, i64 240>, [[ENTRY]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; AVX512-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[TRIGGER:%.*]], <16 x i64> [[VEC_IND]]
 ; AVX512-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> [[TMP0]], i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x i32> poison)
 ; AVX512-NEXT:    [[TMP1:%.*]] = icmp sgt <16 x i32> [[WIDE_MASKED_GATHER]], zeroinitializer
 ; AVX512-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_IN:%.*]], ptr [[IN:%.*]], <16 x i64> [[VEC_IND]], i32 1
 ; AVX512-NEXT:    [[WIDE_MASKED_GATHER2:%.*]] = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> [[TMP2]], i32 4, <16 x i1> [[TMP1]], <16 x float> poison)
 ; AVX512-NEXT:    [[TMP3:%.*]] = fadd <16 x float> [[WIDE_MASKED_GATHER2]], <float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01>
 ; AVX512-NEXT:    [[TMP4:%.*]] = getelementptr inbounds float, ptr [[OUT:%.*]], <16 x i64> [[VEC_IND]]
 ; AVX512-NEXT:    call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> [[TMP3]], <16 x ptr> [[TMP4]], i32 4, <16 x i1> [[TMP1]])

 ;    }
 ;  }
 ;}
 
 %struct.Out = type { float, float }
 
 define void @foo3(ptr noalias %in, ptr noalias %out, ptr noalias %trigger) {
 ; AVX512-LABEL: @foo3(
-; AVX512-NEXT:  iter.check:
+; AVX512-NEXT:  entry:
 ; AVX512-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; AVX512:       vector.body:
-; AVX512-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[ITER_CHECK:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; AVX512-NEXT:    [[VEC_IND:%.*]] = phi <16 x i64> [ <i64 0, i64 16, i64 32, i64 48, i64 64, i64 80, i64 96, i64 112, i64 128, i64 144, i64 160, i64 176, i64 192, i64 208, i64 224, i64 240>, [[ITER_CHECK]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[VEC_IND:%.*]] = phi <16 x i64> [ <i64 0, i64 16, i64 32, i64 48, i64 64, i64 80, i64 96, i64 112, i64 128, i64 144, i64 160, i64 176, i64 192, i64 208, i64 224, i64 240>, [[ENTRY]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; AVX512-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[TRIGGER:%.*]], <16 x i64> [[VEC_IND]]
 ; AVX512-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> [[TMP0]], i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x i32> poison)
 ; AVX512-NEXT:    [[TMP1:%.*]] = icmp sgt <16 x i32> [[WIDE_MASKED_GATHER]], zeroinitializer
 ; AVX512-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_IN:%.*]], ptr [[IN:%.*]], <16 x i64> [[VEC_IND]], i32 1
 ; AVX512-NEXT:    [[WIDE_MASKED_GATHER1:%.*]] = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> [[TMP2]], i32 4, <16 x i1> [[TMP1]], <16 x float> poison)
 ; AVX512-NEXT:    [[TMP3:%.*]] = fadd <16 x float> [[WIDE_MASKED_GATHER1]], <float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01>
 ; AVX512-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_OUT:%.*]], ptr [[OUT:%.*]], <16 x i64> [[VEC_IND]], i32 1
 ; AVX512-NEXT:    call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> [[TMP3]], <16 x ptr> [[TMP4]], i32 4, <16 x i1> [[TMP1]])
for.end:
   ret void
 }
 declare void @llvm.masked.scatter.v16f32.v16p0(<16 x float>, <16 x ptr>, i32, <16 x i1>)
 
 ; The same as @foo2 but scatter/gather argument is a vecotr of ptrs with addresspace 1
 
 define void @foo2_addrspace(ptr addrspace(1) noalias %in, ptr addrspace(1) noalias %out, ptr noalias %trigger, ptr noalias %index) #0 {
 ; AVX512-LABEL: @foo2_addrspace(
-; AVX512-NEXT:  iter.check:
+; AVX512-NEXT:  entry:
 ; AVX512-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; AVX512:       vector.body:
-; AVX512-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[ITER_CHECK:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; AVX512-NEXT:    [[VEC_IND:%.*]] = phi <16 x i64> [ <i64 0, i64 16, i64 32, i64 48, i64 64, i64 80, i64 96, i64 112, i64 128, i64 144, i64 160, i64 176, i64 192, i64 208, i64 224, i64 240>, [[ITER_CHECK]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[VEC_IND:%.*]] = phi <16 x i64> [ <i64 0, i64 16, i64 32, i64 48, i64 64, i64 80, i64 96, i64 112, i64 128, i64 144, i64 160, i64 176, i64 192, i64 208, i64 224, i64 240>, [[ENTRY]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; AVX512-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[TRIGGER:%.*]], <16 x i64> [[VEC_IND]]
 ; AVX512-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> [[TMP0]], i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x i32> poison)
 ; AVX512-NEXT:    [[TMP1:%.*]] = icmp sgt <16 x i32> [[WIDE_MASKED_GATHER]], zeroinitializer
 ; AVX512-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_IN:%.*]], ptr addrspace(1) [[IN:%.*]], <16 x i64> [[VEC_IND]], i32 1
 ; AVX512-NEXT:    [[WIDE_MASKED_GATHER2:%.*]] = call <16 x float> @llvm.masked.gather.v16f32.v16p1(<16 x ptr addrspace(1)> [[TMP2]], i32 4, <16 x i1> [[TMP1]], <16 x float> poison)
 ; AVX512-NEXT:    [[TMP3:%.*]] = fadd <16 x float> [[WIDE_MASKED_GATHER2]], <float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01>
 ; AVX512-NEXT:    [[TMP4:%.*]] = getelementptr inbounds float, ptr addrspace(1) [[OUT:%.*]], <16 x i64> [[VEC_IND]]
 ; AVX512-NEXT:    call void @llvm.masked.scatter.v16f32.v16p1(<16 x float> [[TMP3]], <16 x ptr addrspace(1)> [[TMP4]], i32 4, <16 x i1> [[TMP1]])

 for.end:
   ret void
 }
 
 ; Same as foo2_addrspace but here only the input has the non-default address space.
 
 define void @foo2_addrspace2(ptr addrspace(1) noalias %in, ptr addrspace(0) noalias %out, ptr noalias %trigger, ptr noalias %index) {
 ; AVX512-LABEL: @foo2_addrspace2(
-; AVX512-NEXT:  iter.check:
+; AVX512-NEXT:  entry:
 ; AVX512-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; AVX512:       vector.body:
-; AVX512-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[ITER_CHECK:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; AVX512-NEXT:    [[VEC_IND:%.*]] = phi <16 x i64> [ <i64 0, i64 16, i64 32, i64 48, i64 64, i64 80, i64 96, i64 112, i64 128, i64 144, i64 160, i64 176, i64 192, i64 208, i64 224, i64 240>, [[ITER_CHECK]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[VEC_IND:%.*]] = phi <16 x i64> [ <i64 0, i64 16, i64 32, i64 48, i64 64, i64 80, i64 96, i64 112, i64 128, i64 144, i64 160, i64 176, i64 192, i64 208, i64 224, i64 240>, [[ENTRY]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; AVX512-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[TRIGGER:%.*]], <16 x i64> [[VEC_IND]]
 ; AVX512-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> [[TMP0]], i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x i32> poison)
 ; AVX512-NEXT:    [[TMP1:%.*]] = icmp sgt <16 x i32> [[WIDE_MASKED_GATHER]], zeroinitializer
 ; AVX512-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_IN:%.*]], ptr addrspace(1) [[IN:%.*]], <16 x i64> [[VEC_IND]], i32 1
 ; AVX512-NEXT:    [[WIDE_MASKED_GATHER2:%.*]] = call <16 x float> @llvm.masked.gather.v16f32.v16p1(<16 x ptr addrspace(1)> [[TMP2]], i32 4, <16 x i1> [[TMP1]], <16 x float> poison)
 ; AVX512-NEXT:    [[TMP3:%.*]] = fadd <16 x float> [[WIDE_MASKED_GATHER2]], <float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01>
 ; AVX512-NEXT:    [[TMP4:%.*]] = getelementptr inbounds float, ptr [[OUT:%.*]], <16 x i64> [[VEC_IND]]
 ; AVX512-NEXT:    call void @llvm.masked.scatter.v16f32.v16p0(<16 x float> [[TMP3]], <16 x ptr> [[TMP4]], i32 4, <16 x i1> [[TMP1]])

 for.end:
   ret void
 }
 
 ; Same as foo2_addrspace but here only the output has the non-default address space.
 
 define void @foo2_addrspace3(ptr addrspace(0) noalias %in, ptr addrspace(1) noalias %out, ptr noalias %trigger, ptr noalias %index) {
 ; AVX512-LABEL: @foo2_addrspace3(
-; AVX512-NEXT:  iter.check:
+; AVX512-NEXT:  entry:
 ; AVX512-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; AVX512:       vector.body:
-; AVX512-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[ITER_CHECK:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; AVX512-NEXT:    [[VEC_IND:%.*]] = phi <16 x i64> [ <i64 0, i64 16, i64 32, i64 48, i64 64, i64 80, i64 96, i64 112, i64 128, i64 144, i64 160, i64 176, i64 192, i64 208, i64 224, i64 240>, [[ITER_CHECK]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[INDEX1:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX512-NEXT:    [[VEC_IND:%.*]] = phi <16 x i64> [ <i64 0, i64 16, i64 32, i64 48, i64 64, i64 80, i64 96, i64 112, i64 128, i64 144, i64 160, i64 176, i64 192, i64 208, i64 224, i64 240>, [[ENTRY]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; AVX512-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[TRIGGER:%.*]], <16 x i64> [[VEC_IND]]
 ; AVX512-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <16 x i32> @llvm.masked.gather.v16i32.v16p0(<16 x ptr> [[TMP0]], i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x i32> poison)
 ; AVX512-NEXT:    [[TMP1:%.*]] = icmp sgt <16 x i32> [[WIDE_MASKED_GATHER]], zeroinitializer
 ; AVX512-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_IN:%.*]], ptr [[IN:%.*]], <16 x i64> [[VEC_IND]], i32 1
 ; AVX512-NEXT:    [[WIDE_MASKED_GATHER2:%.*]] = call <16 x float> @llvm.masked.gather.v16f32.v16p0(<16 x ptr> [[TMP2]], i32 4, <16 x i1> [[TMP1]], <16 x float> poison)
 ; AVX512-NEXT:    [[TMP3:%.*]] = fadd <16 x float> [[WIDE_MASKED_GATHER2]], <float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01, float 5.000000e-01>
 ; AVX512-NEXT:    [[TMP4:%.*]] = getelementptr inbounds float, ptr addrspace(1) [[OUT:%.*]], <16 x i64> [[VEC_IND]]
 ; AVX512-NEXT:    call void @llvm.masked.scatter.v16f32.v16p1(<16 x float> [[TMP3]], <16 x ptr addrspace(1)> [[TMP4]], i32 4, <16 x i1> [[TMP1]])

 ; AVX512-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP3]], 8
 ; AVX512-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH:%.*]], label [[VECTOR_MEMCHECK:%.*]]
 ; AVX512:       vector.memcheck:
 ; AVX512-NEXT:    [[TMP4:%.*]] = shl nsw i64 [[IDX_EXT]], 2
 ; AVX512-NEXT:    [[TMP5:%.*]] = add nsw i64 [[TMP4]], -4
 ; AVX512-NEXT:    [[TMP6:%.*]] = lshr i64 [[TMP5]], 2
 ; AVX512-NEXT:    [[TMP7:%.*]] = shl i64 [[TMP6]], 6
 ; AVX512-NEXT:    [[TMP8:%.*]] = add nuw nsw i64 [[TMP7]], 8
 ; AVX512-NEXT:    [[SCEVGEP:%.*]] = getelementptr i8, ptr [[DEST:%.*]], i64 [[TMP8]]
 ; AVX512-NEXT:    [[TMP9:%.*]] = shl nuw i64 [[TMP6]], 2
 ; AVX512-NEXT:    [[TMP10:%.*]] = add i64 [[TMP9]], 4
 ; AVX512-NEXT:    [[SCEVGEP1:%.*]] = getelementptr i8, ptr [[PTR]], i64 [[TMP10]]
 ; AVX512-NEXT:    [[TMP11:%.*]] = mul nsw i64 [[IDX_EXT]], -4
 ; AVX512-NEXT:    [[SCEVGEP2:%.*]] = getelementptr i8, ptr [[PTR]], i64 [[TMP11]]
 ; AVX512-NEXT:    [[TMP12:%.*]] = sub i64 [[TMP10]], [[TMP4]]
 ; AVX512-NEXT:    [[SCEVGEP3:%.*]] = getelementptr i8, ptr [[PTR]], i64 [[TMP12]]
 ; AVX512-NEXT:    [[BOUND0:%.*]] = icmp ult ptr [[DEST]], [[SCEVGEP1]]

llvm/test/Transforms/LoopVectorize/X86/limit-vf-by-tripcount.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt -passes=loop-vectorize -mcpu=skylake-avx512 -S %s | FileCheck %s
 
 target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128-ni:1-p2:32:8:8:32-ni:2"
 target triple = "x86_64-unknown-linux-gnu"
 
-; TODO: Make sure selected VF for the epilog loop doesn't exceed remaining TC.
+; Make sure selected VF for the epilog loop doesn't exceed remaining TC.
 define void @test_tc_17_no_epilogue_vectorization(ptr noalias %src, ptr noalias %dst) {
 ; CHECK-LABEL: @test_tc_17_no_epilogue_vectorization(
-; CHECK-NEXT:  iter.check:
-; CHECK-NEXT:    br i1 false, label [[VEC_EPILOG_SCALAR_PH:%.*]], label [[VECTOR_MAIN_LOOP_ITER_CHECK:%.*]]
-; CHECK:       vector.main.loop.iter.check:
-; CHECK-NEXT:    br i1 false, label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
 ; CHECK:       vector.ph:
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[SRC:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP2]], align 64
 ; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[DST:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP3]], i32 0
 ; CHECK-NEXT:    store <16 x i8> [[WIDE_LOAD]], ptr [[TMP4]], align 64
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 16
 ; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 17, 16
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
-; CHECK:       vec.epilog.iter.check:
-; CHECK-NEXT:    br i1 true, label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
-; CHECK:       vec.epilog.ph:
-; CHECK-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ 16, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
-; CHECK-NEXT:    br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
-; CHECK:       vec.epilog.vector.body:
-; CHECK-NEXT:    [[INDEX2:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT4:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP6:%.*]] = add i64 [[INDEX2]], 0
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP7]], i32 0
-; CHECK-NEXT:    [[WIDE_LOAD3:%.*]] = load <8 x i8>, ptr [[TMP8]], align 64
-; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP9]], i32 0
-; CHECK-NEXT:    store <8 x i8> [[WIDE_LOAD3]], ptr [[TMP10]], align 64
-; CHECK-NEXT:    [[INDEX_NEXT4]] = add nuw i64 [[INDEX2]], 8
-; CHECK-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT4]], 16
-; CHECK-NEXT:    br i1 [[TMP11]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
-; CHECK:       vec.epilog.middle.block:
-; CHECK-NEXT:    [[CMP_N1:%.*]] = icmp eq i64 17, 16
-; CHECK-NEXT:    br i1 [[CMP_N1]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
-; CHECK:       vec.epilog.scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 16, [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ 16, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
+; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; CHECK:       scalar.ph:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 16, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
-; CHECK-NEXT:    [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[LOOP]] ]
+; CHECK-NEXT:    [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[LDADDR:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[I]]
 ; CHECK-NEXT:    [[VAL:%.*]] = load i8, ptr [[LDADDR]], align 64
 ; CHECK-NEXT:    [[STADDR:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[I]]
 ; CHECK-NEXT:    store i8 [[VAL]], ptr [[STADDR]], align 64
 ; CHECK-NEXT:    [[I_NEXT]] = add i64 [[I]], 1
 ; CHECK-NEXT:    [[IS_NEXT:%.*]] = icmp ult i64 [[I_NEXT]], 17
-; CHECK-NEXT:    br i1 [[IS_NEXT]], label [[LOOP]], label [[EXIT]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK-NEXT:    br i1 [[IS_NEXT]], label [[LOOP]], label [[EXIT]], !llvm.loop [[LOOP3:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
 entry:
   br label %loop
 
 loop:
   %i = phi i64 [ 0, %entry ], [ %i.next, %loop ]

 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[SRC:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP2]], align 64
 ; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[DST:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP3]], i32 0
 ; CHECK-NEXT:    store <16 x i8> [[WIDE_LOAD]], ptr [[TMP4]], align 64
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 16
-; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 18, 16
 ; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
 ; CHECK:       vec.epilog.iter.check:
-; CHECK-NEXT:    br i1 true, label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
+; CHECK-NEXT:    br i1 false, label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
 ; CHECK:       vec.epilog.ph:
 ; CHECK-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ 16, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
 ; CHECK-NEXT:    br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
 ; CHECK:       vec.epilog.vector.body:
 ; CHECK-NEXT:    [[INDEX2:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT4:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP6:%.*]] = add i64 [[INDEX2]], 0
 ; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[TMP6]]
 ; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP7]], i32 0
-; CHECK-NEXT:    [[WIDE_LOAD3:%.*]] = load <8 x i8>, ptr [[TMP8]], align 64
+; CHECK-NEXT:    [[WIDE_LOAD3:%.*]] = load <2 x i8>, ptr [[TMP8]], align 64
 ; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP6]]
 ; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP9]], i32 0
-; CHECK-NEXT:    store <8 x i8> [[WIDE_LOAD3]], ptr [[TMP10]], align 64
-; CHECK-NEXT:    [[INDEX_NEXT4]] = add nuw i64 [[INDEX2]], 8
-; CHECK-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT4]], 16
-; CHECK-NEXT:    br i1 [[TMP11]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; CHECK-NEXT:    store <2 x i8> [[WIDE_LOAD3]], ptr [[TMP10]], align 64
+; CHECK-NEXT:    [[INDEX_NEXT4]] = add nuw i64 [[INDEX2]], 2
+; CHECK-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT4]], 18
+; CHECK-NEXT:    br i1 [[TMP11]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
 ; CHECK:       vec.epilog.middle.block:
-; CHECK-NEXT:    [[CMP_N1:%.*]] = icmp eq i64 18, 16
+; CHECK-NEXT:    [[CMP_N1:%.*]] = icmp eq i64 18, 18
 ; CHECK-NEXT:    br i1 [[CMP_N1]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
 ; CHECK:       vec.epilog.scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 16, [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ 16, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 18, [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ 16, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[LDADDR:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[I]]
 ; CHECK-NEXT:    [[VAL:%.*]] = load i8, ptr [[LDADDR]], align 64
 ; CHECK-NEXT:    [[STADDR:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[I]]
 ; CHECK-NEXT:    store i8 [[VAL]], ptr [[STADDR]], align 64
 ; CHECK-NEXT:    [[I_NEXT]] = add i64 [[I]], 1
 ; CHECK-NEXT:    [[IS_NEXT:%.*]] = icmp ult i64 [[I_NEXT]], 18
-; CHECK-NEXT:    br i1 [[IS_NEXT]], label [[LOOP]], label [[EXIT]], !llvm.loop [[LOOP7:![0-9]+]]
+; CHECK-NEXT:    br i1 [[IS_NEXT]], label [[LOOP]], label [[EXIT]], !llvm.loop [[LOOP6:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
 entry:
   br label %loop
 
 loop:
   %i = phi i64 [ 0, %entry ], [ %i.next, %loop ]

 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[SRC:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP2]], align 64
 ; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[DST:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP3]], i32 0
 ; CHECK-NEXT:    store <16 x i8> [[WIDE_LOAD]], ptr [[TMP4]], align 64
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 16
-; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 19, 16
 ; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
 ; CHECK:       vec.epilog.iter.check:
-; CHECK-NEXT:    br i1 true, label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
+; CHECK-NEXT:    br i1 false, label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
 ; CHECK:       vec.epilog.ph:
 ; CHECK-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ 16, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
 ; CHECK-NEXT:    br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
 ; CHECK:       vec.epilog.vector.body:
 ; CHECK-NEXT:    [[INDEX2:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT4:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[TMP6:%.*]] = add i64 [[INDEX2]], 0
 ; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[TMP6]]
 ; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP7]], i32 0
-; CHECK-NEXT:    [[WIDE_LOAD3:%.*]] = load <8 x i8>, ptr [[TMP8]], align 64
+; CHECK-NEXT:    [[WIDE_LOAD3:%.*]] = load <2 x i8>, ptr [[TMP8]], align 64
 ; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP6]]
 ; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP9]], i32 0
-; CHECK-NEXT:    store <8 x i8> [[WIDE_LOAD3]], ptr [[TMP10]], align 64
-; CHECK-NEXT:    [[INDEX_NEXT4]] = add nuw i64 [[INDEX2]], 8
-; CHECK-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT4]], 16
-; CHECK-NEXT:    br i1 [[TMP11]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
+; CHECK-NEXT:    store <2 x i8> [[WIDE_LOAD3]], ptr [[TMP10]], align 64
+; CHECK-NEXT:    [[INDEX_NEXT4]] = add nuw i64 [[INDEX2]], 2
+; CHECK-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT4]], 18
+; CHECK-NEXT:    br i1 [[TMP11]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
 ; CHECK:       vec.epilog.middle.block:
-; CHECK-NEXT:    [[CMP_N1:%.*]] = icmp eq i64 19, 16
+; CHECK-NEXT:    [[CMP_N1:%.*]] = icmp eq i64 19, 18
 ; CHECK-NEXT:    br i1 [[CMP_N1]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
 ; CHECK:       vec.epilog.scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 16, [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ 16, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 18, [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ 16, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[LDADDR:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[I]]
 ; CHECK-NEXT:    [[VAL:%.*]] = load i8, ptr [[LDADDR]], align 64
 ; CHECK-NEXT:    [[STADDR:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[I]]
 ; CHECK-NEXT:    store i8 [[VAL]], ptr [[STADDR]], align 64
 ; CHECK-NEXT:    [[I_NEXT]] = add i64 [[I]], 1
 ; CHECK-NEXT:    [[IS_NEXT:%.*]] = icmp ult i64 [[I_NEXT]], 19
-; CHECK-NEXT:    br i1 [[IS_NEXT]], label [[LOOP]], label [[EXIT]], !llvm.loop [[LOOP10:![0-9]+]]
+; CHECK-NEXT:    br i1 [[IS_NEXT]], label [[LOOP]], label [[EXIT]], !llvm.loop [[LOOP9:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
 entry:
   br label %loop
 
 loop:
   %i = phi i64 [ 0, %entry ], [ %i.next, %loop ]

 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[SRC:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP2]], align 64
 ; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[DST:%.*]], i64 [[TMP0]]
 ; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP3]], i32 0
 ; CHECK-NEXT:    store <4 x i8> [[WIDE_LOAD]], ptr [[TMP4]], align 64
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 20
-; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 20, 20
 ; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
 ; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 20, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[LDADDR:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[I]]
 ; CHECK-NEXT:    [[VAL:%.*]] = load i8, ptr [[LDADDR]], align 64
 ; CHECK-NEXT:    [[STADDR:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[I]]
 ; CHECK-NEXT:    store i8 [[VAL]], ptr [[STADDR]], align 64
 ; CHECK-NEXT:    [[I_NEXT]] = add i64 [[I]], 1
 ; CHECK-NEXT:    [[IS_NEXT:%.*]] = icmp ult i64 [[I_NEXT]], 20
-; CHECK-NEXT:    br i1 [[IS_NEXT]], label [[LOOP]], label [[EXIT]], !llvm.loop [[LOOP12:![0-9]+]]
+; CHECK-NEXT:    br i1 [[IS_NEXT]], label [[LOOP]], label [[EXIT]], !llvm.loop [[LOOP11:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
 entry:
   br label %loop
 
 loop:
   %i = phi i64 [ 0, %entry ], [ %i.next, %loop ]

llvm/test/Transforms/LoopVectorize/X86/pr42674.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt %s -passes=loop-vectorize,instcombine,simplifycfg -simplifycfg-require-and-preserve-domtree=1 -mtriple=x86_64-unknown-linux-gnu -mattr=avx512vl,avx512dq,avx512bw -S | FileCheck %s
 
 @bytes = global [128 x i8] zeroinitializer, align 16
 
 ; Make sure we end up with vector code for this loop. We used to try to create
 ; a VF=64,UF=4 loop, but the scalar trip count is only 128 so
 ; the vector loop was dead code leaving only a scalar remainder.
 define zeroext i8 @sum() {
 ; CHECK-LABEL: @sum(
-; CHECK-NEXT:  iter.check:
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[ITER_CHECK:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <64 x i8> [ zeroinitializer, [[ITER_CHECK]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI1:%.*]] = phi <64 x i8> [ zeroinitializer, [[ITER_CHECK]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [128 x i8], ptr @bytes, i64 0, i64 [[INDEX]]
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [128 x i8], ptr @bytes, i64 0, i64 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <64 x i8>, ptr [[TMP0]], align 16
-; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 64
-; CHECK-NEXT:    [[WIDE_LOAD2:%.*]] = load <64 x i8>, ptr [[TMP2]], align 16
-; CHECK-NEXT:    [[TMP4]] = add <64 x i8> [[WIDE_LOAD]], [[VEC_PHI]]
-; CHECK-NEXT:    [[TMP5]] = add <64 x i8> [[WIDE_LOAD2]], [[VEC_PHI1]]
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 128
-; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX]], 0
-; CHECK-NEXT:    br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[BIN_RDX:%.*]] = add <64 x i8> [[TMP5]], [[TMP4]]
-; CHECK-NEXT:    [[TMP7:%.*]] = call i8 @llvm.vector.reduce.add.v64i8(<64 x i8> [[BIN_RDX]])
-; CHECK-NEXT:    ret i8 [[TMP7]]
+; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 64
+; CHECK-NEXT:    [[WIDE_LOAD2:%.*]] = load <64 x i8>, ptr [[TMP1]], align 16
+; CHECK-NEXT:    [[TMP2:%.*]] = add <64 x i8> [[WIDE_LOAD]], zeroinitializer
+; CHECK-NEXT:    [[TMP3:%.*]] = add <64 x i8> [[WIDE_LOAD2]], zeroinitializer
+; CHECK-NEXT:    [[INDEX_NEXT:%.*]] = add nuw i64 0, 128
+; CHECK-NEXT:    [[BIN_RDX:%.*]] = add <64 x i8> [[TMP3]], [[TMP2]]
+; CHECK-NEXT:    [[TMP4:%.*]] = call i8 @llvm.vector.reduce.add.v64i8(<64 x i8> [[BIN_RDX]])
+; CHECK-NEXT:    ret i8 [[TMP4]]
 ;
 entry:
   br label %for.body
 
 for.body:                                         ; preds = %for.body, %entry
   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
   %r.010 = phi i8 [ 0, %entry ], [ %add, %for.body ]
   %arrayidx = getelementptr inbounds [128 x i8], ptr @bytes, i64 0, i64 %indvars.iv