From f3dbbac1266c066f1849fb80bb8c900a0556b976 Mon Sep 17 00:00:00 2001
From: Roman Lebedev <lebedev.ri@gmail.com>
Date: Tue, 25 Feb 2020 21:52:46 +0300
Subject: [PATCH] [SCEV] SCEVExpander::isHighCostExpansionHelper(): cost-model
 min/max (PR44668)

Summary:
Previosly we simply always said that `SCEVMinMaxExpr` is too costly to expand.
But this isn't really true, it expands into just a comparison+swap pair.
And again much like with add/mul, there will be one less such pair
than the number of operands. And we need to count the cost of operands themselves.

This does change a number of testcases, and as far as i can tell,
all of these changes are improvements, in the sense that
we fixed up more latches to do the [in]equality comparison.

This concludes cost-modelling changes, no other SCEV expressions exist as of now.

This is a part of addressing [[ https://bugs.llvm.org/show_bug.cgi?id=44668 | PR44668 ]].

Reviewers: reames, mkazantsev, wmi, sanjoy

Reviewed By: mkazantsev

Subscribers: hiraditya, javed.absar, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D73744
---
 lib/Analysis/ScalarEvolutionExpander.cpp      | 35 ++++-----
 test/Transforms/IndVarSimplify/elim-extend.ll | 17 +++--
 .../IndVarSimplify/eliminate-comparison.ll    | 17 +++--
 .../IndVarSimplify/eliminate-trunc.ll         | 73 +++++++++++-------
 .../IndVarSimplify/full_widening.ll           |  8 +-
 test/Transforms/IndVarSimplify/iv-widen.ll    | 16 ++--
 .../IndVarSimplify/lftr-multi-exit.ll         | 12 +--
 test/Transforms/IndVarSimplify/lftr-reuse.ll  |  8 +-
 .../loop-invariant-conditions.ll              | 22 ++++--
 .../IndVarSimplify/widen-loop-comp.ll         | 75 +++++++++++--------
 .../LoopVectorize/X86/float-induction-x86.ll  | 74 +++++++++++++++---
 11 files changed, 236 insertions(+), 121 deletions(-)

diff --git a/lib/Analysis/ScalarEvolutionExpander.cpp b/lib/Analysis/ScalarEvolutionExpander.cpp
index 6d9901dbe1c..65a88ea25c7 100644
--- a/lib/Analysis/ScalarEvolutionExpander.cpp
+++ b/lib/Analysis/ScalarEvolutionExpander.cpp
@@ -2277,27 +2277,31 @@ bool SCEVExpander::isHighCostExpansionHelper(
     return BudgetRemaining < 0;
   }
 
-  if (S->getSCEVType() == scAddExpr || S->getSCEVType() == scMulExpr) {
-    const SCEVNAryExpr *NAry = dyn_cast<SCEVNAryExpr>(S);
+  if (const SCEVNAryExpr *NAry = dyn_cast<SCEVNAryExpr>(S)) {
+    Type *OpType = NAry->getType();
 
-    unsigned Opcode;
+    int PairCost;
     switch (S->getSCEVType()) {
     case scAddExpr:
-      Opcode = Instruction::Add;
+      PairCost = TTI.getOperationCost(Instruction::Add, OpType);
       break;
     case scMulExpr:
-      Opcode = Instruction::Mul;
+      // TODO: this is a very pessimistic cost modelling for Mul,
+      // because of Bin Pow algorithm actually used by the expander,
+      // see SCEVExpander::visitMulExpr(), ExpandOpBinPowN().
+      PairCost = TTI.getOperationCost(Instruction::Mul, OpType);
+      break;
+    case scSMaxExpr:
+    case scUMaxExpr:
+    case scSMinExpr:
+    case scUMinExpr:
+      PairCost = TTI.getOperationCost(Instruction::ICmp, OpType) +
+                 TTI.getOperationCost(Instruction::Select, OpType);
       break;
     default:
       llvm_unreachable("There are no other variants here.");
     }
 
-    Type *OpType = NAry->getType();
-    int PairCost = TTI.getOperationCost(Opcode, OpType);
-    // TODO: this is a very pessimistic cost modelling for Mul,
-    // because of Bin Pow algorithm actually used by the expander,
-    // see SCEVExpander::visitMulExpr(), ExpandOpBinPowN().
-
     assert(NAry->getNumOperands() > 1 &&
            "Nary expr should have more than 1 operand.");
     for (const SCEV *Op : NAry->operands()) {
@@ -2311,14 +2315,7 @@ bool SCEVExpander::isHighCostExpansionHelper(
     return BudgetRemaining < 0;
   }
 
-  // HowManyLessThans uses a Max expression whenever the loop is not guarded by
-  // the exit condition.
-  if (isa<SCEVMinMaxExpr>(S))
-    return true;
-
-  // If we haven't recognized an expensive SCEV pattern, assume it's an
-  // expression produced by program code.
-  return false;
+  llvm_unreachable("No other scev expressions possible.");
 }
 
 Value *SCEVExpander::expandCodeForPredicate(const SCEVPredicate *Pred,
diff --git a/test/Transforms/IndVarSimplify/elim-extend.ll b/test/Transforms/IndVarSimplify/elim-extend.ll
index fec5ef48f4e..9de9ac2a1e0 100644
--- a/test/Transforms/IndVarSimplify/elim-extend.ll
+++ b/test/Transforms/IndVarSimplify/elim-extend.ll
@@ -8,7 +8,10 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f3
 define void @postincConstIV(i8* %base, i32 %limit) nounwind {
 ; CHECK-LABEL: @postincConstIV(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = sext i32 [[LIMIT:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[LIMIT:%.*]], 0
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[LIMIT]], i32 0
+; CHECK-NEXT:    [[TMP1:%.*]] = add nuw i32 [[SMAX]], 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[TMP1]] to i64
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY:%.*]] ]
@@ -19,8 +22,8 @@ define void @postincConstIV(i8* %base, i32 %limit) nounwind {
 ; CHECK-NEXT:    store i8 0, i8* [[POSTADR]]
 ; CHECK-NEXT:    [[POSTADRNSW:%.*]] = getelementptr inbounds i8, i8* [[BASE]], i64 [[INDVARS_IV_NEXT]]
 ; CHECK-NEXT:    store i8 0, i8* [[POSTADRNSW]]
-; CHECK-NEXT:    [[COND:%.*]] = icmp sgt i64 [[TMP0]], [[INDVARS_IV]]
-; CHECK-NEXT:    br i1 [[COND]], label [[LOOP]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    br label [[RETURN:%.*]]
 ; CHECK:       return:
@@ -113,7 +116,9 @@ define void @nestedIV(i8* %address, i32 %limit) nounwind {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[LIMITDEC:%.*]] = add i32 [[LIMIT:%.*]], -1
 ; CHECK-NEXT:    [[TMP0:%.*]] = sext i32 [[LIMITDEC]] to i64
-; CHECK-NEXT:    [[TMP1:%.*]] = sext i32 [[LIMIT]] to i64
+; CHECK-NEXT:    [[TMP1:%.*]] = icmp sgt i32 [[LIMIT]], 1
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP1]], i32 [[LIMIT]], i32 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[SMAX]] to i64
 ; CHECK-NEXT:    br label [[OUTERLOOP:%.*]]
 ; CHECK:       outerloop:
 ; CHECK-NEXT:    [[INDVARS_IV1:%.*]] = phi i64 [ [[INDVARS_IV_NEXT2:%.*]], [[OUTERMERGE:%.*]] ], [ 0, [[ENTRY:%.*]] ]
@@ -149,8 +154,8 @@ define void @nestedIV(i8* %address, i32 %limit) nounwind {
 ; CHECK-NEXT:    [[ADR5:%.*]] = getelementptr i8, i8* [[ADDRESS]], i64 [[OFS5]]
 ; CHECK-NEXT:    store i8 0, i8* [[ADR5]]
 ; CHECK-NEXT:    [[INDVARS_IV_NEXT2]] = add nuw nsw i64 [[INDVARS_IV1]], 1
-; CHECK-NEXT:    [[TMP47:%.*]] = icmp slt i64 [[INDVARS_IV_NEXT2]], [[TMP1]]
-; CHECK-NEXT:    br i1 [[TMP47]], label [[OUTERLOOP]], label [[RETURN:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT2]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[OUTERLOOP]], label [[RETURN:%.*]]
 ; CHECK:       return:
 ; CHECK-NEXT:    ret void
 ;
diff --git a/test/Transforms/IndVarSimplify/eliminate-comparison.ll b/test/Transforms/IndVarSimplify/eliminate-comparison.ll
index 4b4a05a94f5..3f51f771afb 100644
--- a/test/Transforms/IndVarSimplify/eliminate-comparison.ll
+++ b/test/Transforms/IndVarSimplify/eliminate-comparison.ll
@@ -529,16 +529,19 @@ define void @func_17(i32* %len.ptr) {
 ; CHECK-NEXT:    [[ENTRY_COND:%.*]] = and i1 [[ENTRY_COND_0]], [[ENTRY_COND_1]]
 ; CHECK-NEXT:    br i1 [[ENTRY_COND]], label [[LOOP_PREHEADER:%.*]], label [[LEAVE:%.*]]
 ; CHECK:       loop.preheader:
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[LEN]], 0
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[LEN]], i32 0
+; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[SMAX]], -5
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
-; CHECK-NEXT:    [[IV_2:%.*]] = phi i32 [ [[IV_2_INC:%.*]], [[BE:%.*]] ], [ 0, [[LOOP_PREHEADER]] ]
+; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ [[IV_INC:%.*]], [[BE:%.*]] ], [ -6, [[LOOP_PREHEADER]] ]
 ; CHECK-NEXT:    call void @side_effect()
-; CHECK-NEXT:    [[IV_2_INC]] = add nuw i32 [[IV_2]], 1
+; CHECK-NEXT:    [[IV_INC]] = add nsw i32 [[IV]], 1
 ; CHECK-NEXT:    br i1 true, label [[BE]], label [[LEAVE_LOOPEXIT:%.*]]
 ; CHECK:       be:
 ; CHECK-NEXT:    call void @side_effect()
-; CHECK-NEXT:    [[BE_COND:%.*]] = icmp slt i32 [[IV_2]], [[LEN]]
-; CHECK-NEXT:    br i1 [[BE_COND]], label [[LOOP]], label [[LEAVE_LOOPEXIT]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i32 [[IV_INC]], [[TMP1]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[LEAVE_LOOPEXIT]]
 ; CHECK:       leave.loopexit:
 ; CHECK-NEXT:    br label [[LEAVE]]
 ; CHECK:       leave:
@@ -685,6 +688,8 @@ define void @func_20(i32* %length.ptr) {
 ; CHECK-NEXT:    [[LENGTH_IS_NONZERO:%.*]] = icmp ne i32 [[LENGTH]], 0
 ; CHECK-NEXT:    br i1 [[LENGTH_IS_NONZERO]], label [[LOOP_PREHEADER:%.*]], label [[LEAVE:%.*]]
 ; CHECK:       loop.preheader:
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[LENGTH]], 1
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[LENGTH]], i32 1
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ [[IV_INC:%.*]], [[BE:%.*]] ], [ 0, [[LOOP_PREHEADER]] ]
@@ -693,8 +698,8 @@ define void @func_20(i32* %length.ptr) {
 ; CHECK-NEXT:    br i1 [[EXITCOND]], label [[BE]], label [[LEAVE_LOOPEXIT:%.*]]
 ; CHECK:       be:
 ; CHECK-NEXT:    call void @side_effect()
-; CHECK-NEXT:    [[BE_COND:%.*]] = icmp slt i32 [[IV_INC]], [[LENGTH]]
-; CHECK-NEXT:    br i1 [[BE_COND]], label [[LOOP]], label [[LEAVE_LOOPEXIT]]
+; CHECK-NEXT:    [[EXITCOND1:%.*]] = icmp ne i32 [[IV_INC]], [[SMAX]]
+; CHECK-NEXT:    br i1 [[EXITCOND1]], label [[LOOP]], label [[LEAVE_LOOPEXIT]]
 ; CHECK:       leave.loopexit:
 ; CHECK-NEXT:    br label [[LEAVE]]
 ; CHECK:       leave:
diff --git a/test/Transforms/IndVarSimplify/eliminate-trunc.ll b/test/Transforms/IndVarSimplify/eliminate-trunc.ll
index 8258445d4f3..8042268391d 100644
--- a/test/Transforms/IndVarSimplify/eliminate-trunc.ll
+++ b/test/Transforms/IndVarSimplify/eliminate-trunc.ll
@@ -36,13 +36,16 @@ define void @test_01(i32 %n) {
 ;
 ; CHECK-LABEL: @test_01(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[SEXT:%.*]] = sext i32 [[N:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[N:%.*]], 0
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[N]], i32 0
+; CHECK-NEXT:    [[TMP1:%.*]] = add nuw i32 [[SMAX]], 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[TMP1]] to i64
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[TMP0:%.*]] = icmp slt i64 [[IV]], [[SEXT]]
-; CHECK-NEXT:    br i1 [[TMP0]], label [[LOOP]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[IV_NEXT]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
@@ -63,13 +66,16 @@ define void @test_02(i32 %n) {
 ;
 ; CHECK-LABEL: @test_02(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[SEXT:%.*]] = sext i32 [[N:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[N:%.*]], 2147483646
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[N]], i32 2147483646
+; CHECK-NEXT:    [[TMP1:%.*]] = add nuw i32 [[SMAX]], 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[TMP1]] to i64
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 2147483646, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[TMP0:%.*]] = icmp slt i64 [[IV]], [[SEXT]]
-; CHECK-NEXT:    br i1 [[TMP0]], label [[LOOP]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[IV_NEXT]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
@@ -113,13 +119,16 @@ define void @test_04(i32 %n) {
 ;
 ; CHECK-LABEL: @test_04(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[SEXT:%.*]] = sext i32 [[N:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[N:%.*]], -2147483647
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[N]], i32 -2147483647
+; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[SMAX]], 1
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ -2147483647, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[TMP0:%.*]] = icmp slt i64 [[IV]], [[SEXT]]
-; CHECK-NEXT:    br i1 [[TMP0]], label [[LOOP]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[LFTR_WIDEIV:%.*]] = trunc i64 [[IV_NEXT]] to i32
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i32 [[LFTR_WIDEIV]], [[TMP1]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
@@ -243,13 +252,16 @@ exit:
 define void @test_02_unsigned(i32 %n) {
 ; CHECK-LABEL: @test_02_unsigned(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[ZEXT:%.*]] = zext i32 [[N:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp ugt i32 [[N:%.*]], -2
+; CHECK-NEXT:    [[UMAX:%.*]] = select i1 [[TMP0]], i32 [[N]], i32 -2
+; CHECK-NEXT:    [[TMP1:%.*]] = add nsw i32 [[UMAX]], 1
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 4294967294, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[TMP0:%.*]] = icmp ult i64 [[IV]], [[ZEXT]]
-; CHECK-NEXT:    br i1 [[TMP0]], label [[LOOP]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[LFTR_WIDEIV:%.*]] = trunc i64 [[IV_NEXT]] to i32
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i32 [[LFTR_WIDEIV]], [[TMP1]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
@@ -318,13 +330,16 @@ exit:
 define void @test_05_unsigned(i32 %n) {
 ; CHECK-LABEL: @test_05_unsigned(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[ZEXT:%.*]] = zext i32 [[N:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp ugt i32 [[N:%.*]], 1
+; CHECK-NEXT:    [[UMAX:%.*]] = select i1 [[TMP0]], i32 [[N]], i32 1
+; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[UMAX]], 1
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 1, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[TMP0:%.*]] = icmp ult i64 [[IV]], [[ZEXT]]
-; CHECK-NEXT:    br i1 [[TMP0]], label [[LOOP]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[LFTR_WIDEIV:%.*]] = trunc i64 [[IV_NEXT]] to i32
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i32 [[LFTR_WIDEIV]], [[TMP1]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
@@ -366,14 +381,18 @@ exit:
 define void @test_07(i32* %p, i32 %n) {
 ; CHECK-LABEL: @test_07(
 ; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[N:%.*]], 0
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[N]], i32 0
+; CHECK-NEXT:    [[TMP1:%.*]] = add nuw i32 [[SMAX]], 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[TMP1]] to i64
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
 ; CHECK-NEXT:    [[NARROW_IV:%.*]] = trunc i64 [[IV]] to i32
 ; CHECK-NEXT:    store i32 [[NARROW_IV]], i32* [[P:%.*]]
-; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i32 [[NARROW_IV]], [[N:%.*]]
-; CHECK-NEXT:    br i1 [[CMP]], label [[LOOP]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[IV_NEXT]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
@@ -451,15 +470,17 @@ exit:
 define void @test_10(i32 %n) {
 ; CHECK-LABEL: @test_10(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[SEXT:%.*]] = sext i32 [[N:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[N:%.*]], 100
+; CHECK-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ult i64 [[TMP1]], 90
+; CHECK-NEXT:    [[UMIN:%.*]] = select i1 [[TMP2]], i64 [[TMP1]], i64 90
+; CHECK-NEXT:    [[TMP3:%.*]] = add i64 [[UMIN]], -99
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ -100, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[TMP0:%.*]] = icmp ne i64 [[IV]], [[SEXT]]
-; CHECK-NEXT:    [[NEGCMP:%.*]] = icmp slt i64 [[IV]], -10
-; CHECK-NEXT:    [[CMP:%.*]] = and i1 [[TMP0]], [[NEGCMP]]
-; CHECK-NEXT:    br i1 [[CMP]], label [[LOOP]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[IV_NEXT]], [[TMP3]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
@@ -530,13 +551,15 @@ define void @test_12(i32* %p) {
 ; CHECK-LABEL: @test_12(
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[N:%.*]] = load i32, i32* [[P:%.*]], !range !0
-; CHECK-NEXT:    [[ZEXT:%.*]] = zext i32 [[N]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[N]], 1
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[N]], i32 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[SMAX]] to i64
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[TMP0:%.*]] = icmp ult i64 [[IV_NEXT]], [[ZEXT]]
-; CHECK-NEXT:    br i1 [[TMP0]], label [[LOOP]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[IV_NEXT]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
diff --git a/test/Transforms/IndVarSimplify/full_widening.ll b/test/Transforms/IndVarSimplify/full_widening.ll
index bdcde6a296f..00cf07ff7d2 100644
--- a/test/Transforms/IndVarSimplify/full_widening.ll
+++ b/test/Transforms/IndVarSimplify/full_widening.ll
@@ -7,7 +7,9 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f3
 define i32 @test_01(double* %p, double %x, i32* %np, i32* %mp, i32 %k) {
 ; CHECK-LABEL: @test_01(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = sext i32 [[K:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[K:%.*]], 1
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[K]], i32 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[SMAX]] to i64
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV_WIDE:%.*]] = phi i64 [ [[CANONICAL_IV_NEXT_I:%.*]], [[LOOP]] ], [ 0, [[ENTRY:%.*]] ]
@@ -17,8 +19,8 @@ define i32 @test_01(double* %p, double %x, i32* %np, i32* %mp, i32 %k) {
 ; CHECK-NEXT:    [[MUL:%.*]] = fmul double [[X:%.*]], [[LOAD]]
 ; CHECK-NEXT:    [[GEP2:%.*]] = getelementptr inbounds double, double* [[P]], i64 [[IV_WIDE]]
 ; CHECK-NEXT:    store atomic double [[MUL]], double* [[GEP2]] unordered, align 8
-; CHECK-NEXT:    [[LOOP_COND:%.*]] = icmp slt i64 [[CANONICAL_IV_NEXT_I]], [[TMP0]]
-; CHECK-NEXT:    br i1 [[LOOP_COND]], label [[LOOP]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[CANONICAL_IV_NEXT_I]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret i32 0
 ;
diff --git a/test/Transforms/IndVarSimplify/iv-widen.ll b/test/Transforms/IndVarSimplify/iv-widen.ll
index 9370bcd539a..de48c632f24 100644
--- a/test/Transforms/IndVarSimplify/iv-widen.ll
+++ b/test/Transforms/IndVarSimplify/iv-widen.ll
@@ -122,15 +122,17 @@ define void @loop_1(i32 %lim) {
 ; CHECK-NEXT:    [[ENTRY_COND:%.*]] = icmp ne i32 [[LIM:%.*]], 0
 ; CHECK-NEXT:    br i1 [[ENTRY_COND]], label [[LOOP_PREHEADER:%.*]], label [[LEAVE:%.*]]
 ; CHECK:       loop.preheader:
-; CHECK-NEXT:    [[TMP0:%.*]] = zext i32 [[LIM]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp ugt i32 [[LIM]], 2
+; CHECK-NEXT:    [[UMAX:%.*]] = select i1 [[TMP0]], i32 [[LIM]], i32 2
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[UMAX]] to i64
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 1, [[LOOP_PREHEADER]] ], [ [[INDVARS_IV_NEXT:%.*]], [[LOOP]] ]
 ; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
 ; CHECK-NEXT:    [[TMP1:%.*]] = add nsw i64 [[INDVARS_IV]], -1
 ; CHECK-NEXT:    call void @dummy.i64(i64 [[TMP1]])
-; CHECK-NEXT:    [[BE_COND:%.*]] = icmp ult i64 [[INDVARS_IV_NEXT]], [[TMP0]]
-; CHECK-NEXT:    br i1 [[BE_COND]], label [[LOOP]], label [[LEAVE_LOOPEXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[LEAVE_LOOPEXIT:%.*]]
 ; CHECK:       leave.loopexit:
 ; CHECK-NEXT:    br label [[LEAVE]]
 ; CHECK:       leave:
@@ -165,7 +167,9 @@ define void @loop_2(i32 %size, i32 %nsteps, i32 %hsize, i32* %lined, i8 %tmp1) {
 ; CHECK-NEXT:    [[BC0:%.*]] = bitcast i32* [[LINED:%.*]] to i8*
 ; CHECK-NEXT:    [[TMP0:%.*]] = sext i32 [[SIZE]] to i64
 ; CHECK-NEXT:    [[TMP1:%.*]] = sext i32 [[HSIZE:%.*]] to i64
-; CHECK-NEXT:    [[TMP2:%.*]] = sext i32 [[NSTEPS:%.*]] to i64
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp sgt i32 [[NSTEPS:%.*]], 1
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP2]], i32 [[NSTEPS]], i32 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT11:%.*]] = zext i32 [[SMAX]] to i64
 ; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
 ; CHECK:       for.body:
 ; CHECK-NEXT:    [[INDVARS_IV7:%.*]] = phi i64 [ [[INDVARS_IV_NEXT8:%.*]], [[FOR_INC:%.*]] ], [ 0, [[ENTRY:%.*]] ]
@@ -200,8 +204,8 @@ define void @loop_2(i32 %size, i32 %nsteps, i32 %hsize, i32* %lined, i8 %tmp1) {
 ; CHECK-NEXT:    br label [[FOR_INC]]
 ; CHECK:       for.inc:
 ; CHECK-NEXT:    [[INDVARS_IV_NEXT8]] = add nuw nsw i64 [[INDVARS_IV7]], 1
-; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i64 [[INDVARS_IV_NEXT8]], [[TMP2]]
-; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END_LOOPEXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND12:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT8]], [[WIDE_TRIP_COUNT11]]
+; CHECK-NEXT:    br i1 [[EXITCOND12]], label [[FOR_BODY]], label [[FOR_END_LOOPEXIT:%.*]]
 ; CHECK:       for.end.loopexit:
 ; CHECK-NEXT:    ret void
 ;
diff --git a/test/Transforms/IndVarSimplify/lftr-multi-exit.ll b/test/Transforms/IndVarSimplify/lftr-multi-exit.ll
index 9b720b0758f..66951eda7a5 100644
--- a/test/Transforms/IndVarSimplify/lftr-multi-exit.ll
+++ b/test/Transforms/IndVarSimplify/lftr-multi-exit.ll
@@ -128,18 +128,18 @@ exit:
 define void @compound_early_exit(i32 %n, i32 %m) {
 ; CHECK-LABEL: @compound_early_exit(
 ; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp ult i32 [[M:%.*]], [[N:%.*]]
+; CHECK-NEXT:    [[UMIN:%.*]] = select i1 [[TMP0]], i32 [[M]], i32 [[N]]
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
-; CHECK-NEXT:    [[EARLYCND:%.*]] = icmp ult i32 [[IV]], [[N:%.*]]
-; CHECK-NEXT:    [[EARLYCND2:%.*]] = icmp ult i32 [[IV]], [[M:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and i1 [[EARLYCND]], [[EARLYCND2]]
-; CHECK-NEXT:    br i1 [[AND]], label [[LATCH]], label [[EXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i32 [[IV]], [[UMIN]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LATCH]], label [[EXIT:%.*]]
 ; CHECK:       latch:
 ; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i32 [[IV]], 1
 ; CHECK-NEXT:    store volatile i32 [[IV]], i32* @A
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i32 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT]]
+; CHECK-NEXT:    [[EXITCOND1:%.*]] = icmp ne i32 [[IV_NEXT]], 1000
+; CHECK-NEXT:    br i1 [[EXITCOND1]], label [[LOOP]], label [[EXIT]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
diff --git a/test/Transforms/IndVarSimplify/lftr-reuse.ll b/test/Transforms/IndVarSimplify/lftr-reuse.ll
index ba1fba9187e..8a55ec09ee6 100644
--- a/test/Transforms/IndVarSimplify/lftr-reuse.ll
+++ b/test/Transforms/IndVarSimplify/lftr-reuse.ll
@@ -187,13 +187,15 @@ define void @unguardedloop([0 x double]* %matrix, [0 x double]* %vector,
 ;
 ; CHECK-LABEL: @unguardedloop(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = sext i32 [[IROW:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[IROW:%.*]], 1
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[IROW]], i32 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[SMAX]] to i64
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
 ; CHECK-NEXT:    [[INDVARS_IV2:%.*]] = phi i64 [ [[INDVARS_IV_NEXT3:%.*]], [[LOOP]] ], [ 0, [[ENTRY:%.*]] ]
 ; CHECK-NEXT:    [[INDVARS_IV_NEXT3]] = add nuw nsw i64 [[INDVARS_IV2]], 1
-; CHECK-NEXT:    [[CMP196:%.*]] = icmp slt i64 [[INDVARS_IV_NEXT3]], [[TMP0]]
-; CHECK-NEXT:    br i1 [[CMP196]], label [[LOOP]], label [[RETURN:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT3]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[RETURN:%.*]]
 ; CHECK:       return:
 ; CHECK-NEXT:    ret void
 ;
diff --git a/test/Transforms/IndVarSimplify/loop-invariant-conditions.ll b/test/Transforms/IndVarSimplify/loop-invariant-conditions.ll
index e2229609798..ad11bc015b6 100644
--- a/test/Transforms/IndVarSimplify/loop-invariant-conditions.ll
+++ b/test/Transforms/IndVarSimplify/loop-invariant-conditions.ll
@@ -311,12 +311,15 @@ for.end:                                          ; preds = %if.end, %entry
 define void @test3_neg(i64 %start) {
 ; CHECK-LABEL: @test3_neg(
 ; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i64 [[START:%.*]], -1
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i64 [[START]], i64 -1
+; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[SMAX]], 1
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[START:%.*]], [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[CMP1:%.*]] = icmp slt i64 [[INDVARS_IV]], -1
-; CHECK-NEXT:    br i1 [[CMP1]], label [[LOOP]], label [[FOR_END:%.*]]
+; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[START]], [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[LOOP]] ]
+; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT]], [[TMP1]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[LOOP]], label [[FOR_END:%.*]]
 ; CHECK:       for.end:
 ; CHECK-NEXT:    ret void
 ;
@@ -336,16 +339,19 @@ for.end:                                          ; preds = %if.end, %entry
 define void @test4_neg(i64 %start) {
 ; CHECK-LABEL: @test4_neg(
 ; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i64 [[START:%.*]], 0
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i64 [[START]], i64 0
+; CHECK-NEXT:    [[TMP1:%.*]] = add nuw i64 [[SMAX]], 1
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[START:%.*]], [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[BACKEDGE:%.*]] ]
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nsw i64 [[INDVARS_IV]], 1
+; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[START]], [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[BACKEDGE:%.*]] ]
+; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 25
 ; CHECK-NEXT:    br i1 [[CMP]], label [[BACKEDGE]], label [[FOR_END:%.*]]
 ; CHECK:       backedge:
 ; CHECK-NEXT:    call void @foo()
-; CHECK-NEXT:    [[CMP1:%.*]] = icmp sgt i64 [[INDVARS_IV]], -1
-; CHECK-NEXT:    br i1 [[CMP1]], label [[FOR_END]], label [[LOOP]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[TMP1]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_END]], label [[LOOP]]
 ; CHECK:       for.end:
 ; CHECK-NEXT:    ret void
 ;
diff --git a/test/Transforms/IndVarSimplify/widen-loop-comp.ll b/test/Transforms/IndVarSimplify/widen-loop-comp.ll
index bc06d3ff743..d741514fa4f 100644
--- a/test/Transforms/IndVarSimplify/widen-loop-comp.ll
+++ b/test/Transforms/IndVarSimplify/widen-loop-comp.ll
@@ -24,30 +24,33 @@ define i32 @test1() {
 ; CHECK:       for.body.lr.ph:
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i32*, i32** @ptr, align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load i32, i32* @e, align 4
-; CHECK-NEXT:    [[TMP3:%.*]] = sext i32 [[TMP2]] to i64
+; CHECK-NEXT:    [[TMP3:%.*]] = icmp sgt i32 [[TMP2]], 0
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP3]], i32 [[TMP2]], i32 0
+; CHECK-NEXT:    [[TMP4:%.*]] = add nuw i32 [[SMAX]], 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[TMP4]] to i64
 ; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
 ; CHECK:       for.cond:
 ; CHECK-NEXT:    [[INDVARS_IV_NEXT:%.*]] = add nuw nsw i64 [[INDVARS_IV:%.*]], 1
-; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i64 [[INDVARS_IV]], [[TMP3]]
-; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_COND_FOR_END_LOOPEXIT_CRIT_EDGE:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_BODY]], label [[FOR_COND_FOR_END_LOOPEXIT_CRIT_EDGE:%.*]]
 ; CHECK:       for.body:
 ; CHECK-NEXT:    [[INDVARS_IV]] = phi i64 [ [[INDVARS_IV_NEXT]], [[FOR_COND:%.*]] ], [ 0, [[FOR_BODY_LR_PH]] ]
 ; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP1]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[TMP4:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
-; CHECK-NEXT:    [[TOBOOL:%.*]] = icmp eq i32 [[TMP4]], 0
+; CHECK-NEXT:    [[TMP5:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
+; CHECK-NEXT:    [[TOBOOL:%.*]] = icmp eq i32 [[TMP5]], 0
 ; CHECK-NEXT:    br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_COND]]
 ; CHECK:       if.then:
 ; CHECK-NEXT:    [[I_05_LCSSA_WIDE:%.*]] = phi i64 [ [[INDVARS_IV]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[TMP5:%.*]] = trunc i64 [[I_05_LCSSA_WIDE]] to i32
-; CHECK-NEXT:    store i32 [[TMP5]], i32* @idx, align 4
+; CHECK-NEXT:    [[TMP6:%.*]] = trunc i64 [[I_05_LCSSA_WIDE]] to i32
+; CHECK-NEXT:    store i32 [[TMP6]], i32* @idx, align 4
 ; CHECK-NEXT:    br label [[FOR_END:%.*]]
 ; CHECK:       for.cond.for.end.loopexit_crit_edge:
 ; CHECK-NEXT:    br label [[FOR_END_LOOPEXIT]]
 ; CHECK:       for.end.loopexit:
 ; CHECK-NEXT:    br label [[FOR_END]]
 ; CHECK:       for.end:
-; CHECK-NEXT:    [[TMP6:%.*]] = load i32, i32* @idx, align 4
-; CHECK-NEXT:    ret i32 [[TMP6]]
+; CHECK-NEXT:    [[TMP7:%.*]] = load i32, i32* @idx, align 4
+; CHECK-NEXT:    ret i32 [[TMP7]]
 ;
 entry:
   store i32 -1, i32* @idx, align 4
@@ -96,7 +99,8 @@ define void @test2([8 x i8]* %a, i8* %b, i8 %limit) {
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i8 [[LIMIT:%.*]] to i32
 ; CHECK-NEXT:    br i1 undef, label [[FOR_COND1_PREHEADER_PREHEADER:%.*]], label [[FOR_COND1_PREHEADER_US_PREHEADER:%.*]]
 ; CHECK:       for.cond1.preheader.us.preheader:
-; CHECK-NEXT:    [[TMP0:%.*]] = zext i32 [[CONV]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[CONV]], 1
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[CONV]], i32 1
 ; CHECK-NEXT:    br label [[FOR_COND1_PREHEADER_US:%.*]]
 ; CHECK:       for.cond1.preheader.preheader:
 ; CHECK-NEXT:    br label [[FOR_COND1_PREHEADER:%.*]]
@@ -107,8 +111,8 @@ define void @test2([8 x i8]* %a, i8* %b, i8 %limit) {
 ; CHECK-NEXT:    br label [[FOR_INC13_US]]
 ; CHECK:       for.inc13.us:
 ; CHECK-NEXT:    [[INDVARS_IV_NEXT3]] = add nuw nsw i64 [[INDVARS_IV2]], 1
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT3]], 4
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND1_PREHEADER_US]], label [[FOR_END_LOOPEXIT1:%.*]]
+; CHECK-NEXT:    [[EXITCOND4:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT3]], 4
+; CHECK-NEXT:    br i1 [[EXITCOND4]], label [[FOR_COND1_PREHEADER_US]], label [[FOR_END_LOOPEXIT1:%.*]]
 ; CHECK:       for.body4.us:
 ; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, [[FOR_BODY4_LR_PH_US]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY4_US:%.*]] ]
 ; CHECK-NEXT:    [[ARRAYIDX6_US:%.*]] = getelementptr inbounds [8 x i8], [8 x i8]* [[A:%.*]], i64 [[INDVARS_IV2]], i64 [[INDVARS_IV]]
@@ -118,9 +122,10 @@ define void @test2([8 x i8]* %a, i8* %b, i8 %limit) {
 ; CHECK-NEXT:    [[TMP2:%.*]] = load i8, i8* [[ARRAYIDX8_US]], align 1
 ; CHECK-NEXT:    store i8 [[TMP2]], i8* [[ARRAYIDX6_US]], align 1
 ; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[CMP2_US:%.*]] = icmp ult i64 [[INDVARS_IV_NEXT]], [[TMP0]]
-; CHECK-NEXT:    br i1 [[CMP2_US]], label [[FOR_BODY4_US]], label [[FOR_INC13_US_LOOPEXIT:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT:%.*]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_BODY4_US]], label [[FOR_INC13_US_LOOPEXIT:%.*]]
 ; CHECK:       for.body4.lr.ph.us:
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT]] = zext i32 [[SMAX]] to i64
 ; CHECK-NEXT:    br label [[FOR_BODY4_US]]
 ; CHECK:       for.cond1.preheader:
 ; CHECK-NEXT:    br i1 false, label [[FOR_INC13:%.*]], label [[FOR_INC13]]
@@ -180,13 +185,15 @@ for.end:
 define i32 @test3(i32* %a, i32 %b) {
 ; CHECK-LABEL: @test3(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = sext i32 [[B:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[B:%.*]], 0
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[B]], i32 0
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[SMAX]] to i64
 ; CHECK-NEXT:    br label [[FOR_COND:%.*]]
 ; CHECK:       for.cond:
 ; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY:%.*]] ], [ 0, [[ENTRY:%.*]] ]
 ; CHECK-NEXT:    [[SUM_0:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i64 [[INDVARS_IV]], [[TMP0]]
-; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_BODY]], label [[FOR_END:%.*]]
 ; CHECK:       for.body:
 ; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 [[INDVARS_IV]]
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
@@ -300,17 +307,20 @@ for.end:
 define i32 @test6(i32* %a, i32 %b) {
 ; CHECK-LABEL: @test6(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = sext i32 [[B:%.*]] to i64
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp sgt i32 [[B:%.*]], -1
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i32 [[B]], i32 -1
+; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[SMAX]], 1
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[TMP1]] to i64
 ; CHECK-NEXT:    br label [[FOR_COND:%.*]]
 ; CHECK:       for.cond:
 ; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY:%.*]] ], [ 0, [[ENTRY:%.*]] ]
 ; CHECK-NEXT:    [[SUM_0:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[CMP:%.*]] = icmp sle i64 [[INDVARS_IV]], [[TMP0]]
-; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_BODY]], label [[FOR_END:%.*]]
 ; CHECK:       for.body:
 ; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
-; CHECK-NEXT:    [[ADD]] = add nsw i32 [[SUM_0]], [[TMP1]]
+; CHECK-NEXT:    [[TMP2:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
+; CHECK-NEXT:    [[ADD]] = add nsw i32 [[SUM_0]], [[TMP2]]
 ; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
 ; CHECK-NEXT:    br label [[FOR_COND]]
 ; CHECK:       for.end:
@@ -342,7 +352,10 @@ define i32 @test7(i32* %a, i32 %b) {
 ; CHECK-LABEL: @test7(
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = zext i32 [[B:%.*]] to i64
-; CHECK-NEXT:    [[TMP1:%.*]] = sext i32 [[B]] to i64
+; CHECK-NEXT:    [[TMP1:%.*]] = icmp sgt i32 [[B]], -1
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP1]], i32 [[B]], i32 -1
+; CHECK-NEXT:    [[TMP2:%.*]] = add i32 [[SMAX]], 2
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[TMP2]] to i64
 ; CHECK-NEXT:    br label [[FOR_COND:%.*]]
 ; CHECK:       for.cond:
 ; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY:%.*]] ], [ 0, [[ENTRY:%.*]] ]
@@ -351,11 +364,11 @@ define i32 @test7(i32* %a, i32 %b) {
 ; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
 ; CHECK:       for.body:
 ; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[TMP2:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
-; CHECK-NEXT:    [[ADD]] = add nsw i32 [[SUM_0]], [[TMP2]]
+; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
+; CHECK-NEXT:    [[ADD]] = add nsw i32 [[SUM_0]], [[TMP3]]
 ; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[CMP2:%.*]] = icmp sle i64 [[INDVARS_IV]], [[TMP1]]
-; CHECK-NEXT:    br i1 [[CMP2]], label [[FOR_COND]], label [[FOR_END]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND]], label [[FOR_END]]
 ; CHECK:       for.end:
 ; CHECK-NEXT:    [[SUM_0_LCSSA:%.*]] = phi i32 [ [[SUM_0]], [[FOR_BODY]] ], [ [[SUM_0]], [[FOR_COND]] ]
 ; CHECK-NEXT:    ret i32 [[SUM_0_LCSSA]]
@@ -444,13 +457,15 @@ define i32 @test9(i32* %a, i32 %b, i32 %init) {
 ; CHECK-NEXT:    br i1 [[E]], label [[FOR_COND_PREHEADER:%.*]], label [[LEAVE:%.*]]
 ; CHECK:       for.cond.preheader:
 ; CHECK-NEXT:    [[TMP0:%.*]] = zext i32 [[INIT]] to i64
-; CHECK-NEXT:    [[TMP1:%.*]] = sext i32 [[B:%.*]] to i64
+; CHECK-NEXT:    [[TMP1:%.*]] = icmp sgt i32 [[INIT]], [[B:%.*]]
+; CHECK-NEXT:    [[SMAX:%.*]] = select i1 [[TMP1]], i32 [[INIT]], i32 [[B]]
+; CHECK-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[SMAX]] to i64
 ; CHECK-NEXT:    br label [[FOR_COND:%.*]]
 ; CHECK:       for.cond:
 ; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[TMP0]], [[FOR_COND_PREHEADER]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY:%.*]] ]
 ; CHECK-NEXT:    [[SUM_0:%.*]] = phi i32 [ [[ADD:%.*]], [[FOR_BODY]] ], [ 0, [[FOR_COND_PREHEADER]] ]
-; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i64 [[INDVARS_IV]], [[TMP1]]
-; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
+; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_BODY]], label [[FOR_END:%.*]]
 ; CHECK:       for.body:
 ; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 [[INDVARS_IV]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
diff --git a/test/Transforms/LoopVectorize/X86/float-induction-x86.ll b/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
index bfa9c727950..3d579915566 100644
--- a/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
+++ b/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
@@ -425,8 +425,8 @@ define double @external_use_with_fast_math(double* %a, i64 %n) {
 ; AUTO_VEC-NEXT:    store double [[J]], double* [[TMP0]], align 8
 ; AUTO_VEC-NEXT:    [[I_NEXT]] = add nuw nsw i64 [[I]], 1
 ; AUTO_VEC-NEXT:    [[J_NEXT]] = fadd fast double [[J]], 3.000000e+00
-; AUTO_VEC-NEXT:    [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; AUTO_VEC-NEXT:    br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop !9
+; AUTO_VEC-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[I_NEXT]], [[SMAX]]
+; AUTO_VEC-NEXT:    br i1 [[EXITCOND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop !9
 ; AUTO_VEC:       for.end:
 ; AUTO_VEC-NEXT:    [[J_LCSSA:%.*]] = phi double [ [[TMP47]], [[MIDDLE_BLOCK]] ], [ [[J]], [[FOR_BODY]] ]
 ; AUTO_VEC-NEXT:    ret double [[J_LCSSA]]
@@ -452,18 +452,74 @@ for.end:
 define double @external_use_without_fast_math(double* %a, i64 %n) {
 ; AUTO_VEC-LABEL: @external_use_without_fast_math(
 ; AUTO_VEC-NEXT:  entry:
+; AUTO_VEC-NEXT:    [[TMP0:%.*]] = icmp sgt i64 [[N:%.*]], 1
+; AUTO_VEC-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i64 [[N]], i64 1
+; AUTO_VEC-NEXT:    [[TMP1:%.*]] = add nsw i64 [[SMAX]], -1
+; AUTO_VEC-NEXT:    [[XTRAITER:%.*]] = and i64 [[SMAX]], 7
+; AUTO_VEC-NEXT:    [[TMP2:%.*]] = icmp ult i64 [[TMP1]], 7
+; AUTO_VEC-NEXT:    br i1 [[TMP2]], label [[FOR_END_UNR_LCSSA:%.*]], label [[ENTRY_NEW:%.*]]
+; AUTO_VEC:       entry.new:
+; AUTO_VEC-NEXT:    [[UNROLL_ITER:%.*]] = sub nsw i64 [[SMAX]], [[XTRAITER]]
 ; AUTO_VEC-NEXT:    br label [[FOR_BODY:%.*]]
 ; AUTO_VEC:       for.body:
-; AUTO_VEC-NEXT:    [[I:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[I_NEXT:%.*]], [[FOR_BODY]] ]
-; AUTO_VEC-NEXT:    [[J:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[J_NEXT:%.*]], [[FOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[I:%.*]] = phi i64 [ 0, [[ENTRY_NEW]] ], [ [[I_NEXT_7:%.*]], [[FOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[J:%.*]] = phi double [ 0.000000e+00, [[ENTRY_NEW]] ], [ [[J_NEXT_7:%.*]], [[FOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[NITER:%.*]] = phi i64 [ [[UNROLL_ITER]], [[ENTRY_NEW]] ], [ [[NITER_NSUB_7:%.*]], [[FOR_BODY]] ]
 ; AUTO_VEC-NEXT:    [[TMP0:%.*]] = getelementptr double, double* [[A:%.*]], i64 [[I]]
 ; AUTO_VEC-NEXT:    store double [[J]], double* [[TMP0]], align 8
-; AUTO_VEC-NEXT:    [[I_NEXT]] = add nuw nsw i64 [[I]], 1
-; AUTO_VEC-NEXT:    [[J_NEXT]] = fadd double [[J]], 3.000000e+00
-; AUTO_VEC-NEXT:    [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N:%.*]]
-; AUTO_VEC-NEXT:    br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END:%.*]]
+; AUTO_VEC-NEXT:    [[I_NEXT:%.*]] = or i64 [[I]], 1
+; AUTO_VEC-NEXT:    [[J_NEXT:%.*]] = fadd double [[J]], 3.000000e+00
+; AUTO_VEC-NEXT:    [[TMP0_1:%.*]] = getelementptr double, double* [[A]], i64 [[I_NEXT]]
+; AUTO_VEC-NEXT:    store double [[J_NEXT]], double* [[TMP0_1]], align 8
+; AUTO_VEC-NEXT:    [[I_NEXT_1:%.*]] = or i64 [[I]], 2
+; AUTO_VEC-NEXT:    [[J_NEXT_1:%.*]] = fadd double [[J_NEXT]], 3.000000e+00
+; AUTO_VEC-NEXT:    [[TMP0_2:%.*]] = getelementptr double, double* [[A]], i64 [[I_NEXT_1]]
+; AUTO_VEC-NEXT:    store double [[J_NEXT_1]], double* [[TMP0_2]], align 8
+; AUTO_VEC-NEXT:    [[I_NEXT_2:%.*]] = or i64 [[I]], 3
+; AUTO_VEC-NEXT:    [[J_NEXT_2:%.*]] = fadd double [[J_NEXT_1]], 3.000000e+00
+; AUTO_VEC-NEXT:    [[TMP0_3:%.*]] = getelementptr double, double* [[A]], i64 [[I_NEXT_2]]
+; AUTO_VEC-NEXT:    store double [[J_NEXT_2]], double* [[TMP0_3]], align 8
+; AUTO_VEC-NEXT:    [[I_NEXT_3:%.*]] = or i64 [[I]], 4
+; AUTO_VEC-NEXT:    [[J_NEXT_3:%.*]] = fadd double [[J_NEXT_2]], 3.000000e+00
+; AUTO_VEC-NEXT:    [[TMP0_4:%.*]] = getelementptr double, double* [[A]], i64 [[I_NEXT_3]]
+; AUTO_VEC-NEXT:    store double [[J_NEXT_3]], double* [[TMP0_4]], align 8
+; AUTO_VEC-NEXT:    [[I_NEXT_4:%.*]] = or i64 [[I]], 5
+; AUTO_VEC-NEXT:    [[J_NEXT_4:%.*]] = fadd double [[J_NEXT_3]], 3.000000e+00
+; AUTO_VEC-NEXT:    [[TMP0_5:%.*]] = getelementptr double, double* [[A]], i64 [[I_NEXT_4]]
+; AUTO_VEC-NEXT:    store double [[J_NEXT_4]], double* [[TMP0_5]], align 8
+; AUTO_VEC-NEXT:    [[I_NEXT_5:%.*]] = or i64 [[I]], 6
+; AUTO_VEC-NEXT:    [[J_NEXT_5:%.*]] = fadd double [[J_NEXT_4]], 3.000000e+00
+; AUTO_VEC-NEXT:    [[TMP0_6:%.*]] = getelementptr double, double* [[A]], i64 [[I_NEXT_5]]
+; AUTO_VEC-NEXT:    store double [[J_NEXT_5]], double* [[TMP0_6]], align 8
+; AUTO_VEC-NEXT:    [[I_NEXT_6:%.*]] = or i64 [[I]], 7
+; AUTO_VEC-NEXT:    [[J_NEXT_6:%.*]] = fadd double [[J_NEXT_5]], 3.000000e+00
+; AUTO_VEC-NEXT:    [[TMP0_7:%.*]] = getelementptr double, double* [[A]], i64 [[I_NEXT_6]]
+; AUTO_VEC-NEXT:    store double [[J_NEXT_6]], double* [[TMP0_7]], align 8
+; AUTO_VEC-NEXT:    [[I_NEXT_7]] = add nuw nsw i64 [[I]], 8
+; AUTO_VEC-NEXT:    [[J_NEXT_7]] = fadd double [[J_NEXT_6]], 3.000000e+00
+; AUTO_VEC-NEXT:    [[NITER_NSUB_7]] = add i64 [[NITER]], -8
+; AUTO_VEC-NEXT:    [[NITER_NCMP_7:%.*]] = icmp eq i64 [[NITER_NSUB_7]], 0
+; AUTO_VEC-NEXT:    br i1 [[NITER_NCMP_7]], label [[FOR_END_UNR_LCSSA]], label [[FOR_BODY]]
+; AUTO_VEC:       for.end.unr-lcssa:
+; AUTO_VEC-NEXT:    [[J_LCSSA_PH:%.*]] = phi double [ undef, [[ENTRY:%.*]] ], [ [[J_NEXT_6]], [[FOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[I_UNR:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[I_NEXT_7]], [[FOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[J_UNR:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[J_NEXT_7]], [[FOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[LCMP_MOD:%.*]] = icmp eq i64 [[XTRAITER]], 0
+; AUTO_VEC-NEXT:    br i1 [[LCMP_MOD]], label [[FOR_END:%.*]], label [[FOR_BODY_EPIL:%.*]]
+; AUTO_VEC:       for.body.epil:
+; AUTO_VEC-NEXT:    [[I_EPIL:%.*]] = phi i64 [ [[I_NEXT_EPIL:%.*]], [[FOR_BODY_EPIL]] ], [ [[I_UNR]], [[FOR_END_UNR_LCSSA]] ]
+; AUTO_VEC-NEXT:    [[J_EPIL:%.*]] = phi double [ [[J_NEXT_EPIL:%.*]], [[FOR_BODY_EPIL]] ], [ [[J_UNR]], [[FOR_END_UNR_LCSSA]] ]
+; AUTO_VEC-NEXT:    [[EPIL_ITER:%.*]] = phi i64 [ [[EPIL_ITER_SUB:%.*]], [[FOR_BODY_EPIL]] ], [ [[XTRAITER]], [[FOR_END_UNR_LCSSA]] ]
+; AUTO_VEC-NEXT:    [[TMP0_EPIL:%.*]] = getelementptr double, double* [[A]], i64 [[I_EPIL]]
+; AUTO_VEC-NEXT:    store double [[J_EPIL]], double* [[TMP0_EPIL]], align 8
+; AUTO_VEC-NEXT:    [[I_NEXT_EPIL]] = add nuw nsw i64 [[I_EPIL]], 1
+; AUTO_VEC-NEXT:    [[J_NEXT_EPIL]] = fadd double [[J_EPIL]], 3.000000e+00
+; AUTO_VEC-NEXT:    [[EPIL_ITER_SUB]] = add i64 [[EPIL_ITER]], -1
+; AUTO_VEC-NEXT:    [[EPIL_ITER_CMP:%.*]] = icmp eq i64 [[EPIL_ITER_SUB]], 0
+; AUTO_VEC-NEXT:    br i1 [[EPIL_ITER_CMP]], label [[FOR_END]], label [[FOR_BODY_EPIL]], !llvm.loop !10
 ; AUTO_VEC:       for.end:
-; AUTO_VEC-NEXT:    ret double [[J]]
+; AUTO_VEC-NEXT:    [[J_LCSSA:%.*]] = phi double [ [[J_LCSSA_PH]], [[FOR_END_UNR_LCSSA]] ], [ [[J_EPIL]], [[FOR_BODY_EPIL]] ]
+; AUTO_VEC-NEXT:    ret double [[J_LCSSA]]
 ;
 entry:
   br label %for.body