[NFC][LoopVectorize] Autogenerate tests affected by isHighCostExpansionHelper() cost...

diff --git a/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll b/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
index af4a48d..7266d8e 100644 (file)
--- a/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
@@ -1,3 +1,4 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt < %s  -O3 -simplifycfg -keep-loops=false -mcpu=core-avx2 -mtriple=x86_64-unknown-linux-gnu -S | FileCheck --check-prefix AUTO_VEC %s
 
 ; This test checks auto-vectorization with FP induction variable.
@@ -11,12 +12,58 @@
 ;  }
 ;}
 
-
-; AUTO_VEC-LABEL: @fp_iv_loop1(
-; AUTO_VEC: vector.body
-; AUTO_VEC: store <8 x float>
-
 define void @fp_iv_loop1(float* noalias nocapture %A, i32 %N) #0 {
+; AUTO_VEC-LABEL: @fp_iv_loop1(
+; AUTO_VEC-NEXT:  entry:
+; AUTO_VEC-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[N:%.*]], 0
+; AUTO_VEC-NEXT:    br i1 [[CMP4]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
+; AUTO_VEC:       for.body.preheader:
+; AUTO_VEC-NEXT:    [[ZEXT:%.*]] = zext i32 [[N]] to i64
+; AUTO_VEC-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[N]], 32
+; AUTO_VEC-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[FOR_BODY:%.*]], label [[VECTOR_PH:%.*]]
+; AUTO_VEC:       vector.ph:
+; AUTO_VEC-NEXT:    [[N_VEC:%.*]] = and i64 [[ZEXT]], 4294967264
+; AUTO_VEC-NEXT:    [[CAST_CRD:%.*]] = sitofp i64 [[N_VEC]] to float
+; AUTO_VEC-NEXT:    [[TMP0:%.*]] = fmul fast float [[CAST_CRD]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[IND_END:%.*]] = fadd fast float [[TMP0]], 1.000000e+00
+; AUTO_VEC-NEXT:    br label [[VECTOR_BODY:%.*]]
+; AUTO_VEC:       vector.body:
+; AUTO_VEC-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[VEC_IND:%.*]] = phi <8 x float> [ <float 1.000000e+00, float 1.500000e+00, float 2.000000e+00, float 2.500000e+00, float 3.000000e+00, float 3.500000e+00, float 4.000000e+00, float 4.500000e+00>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[STEP_ADD:%.*]] = fadd fast <8 x float> [[VEC_IND]], <float 4.000000e+00, float 4.000000e+00, float 4.000000e+00, float 4.000000e+00, float 4.000000e+00, float 4.000000e+00, float 4.000000e+00, float 4.000000e+00>
+; AUTO_VEC-NEXT:    [[STEP_ADD5:%.*]] = fadd fast <8 x float> [[VEC_IND]], <float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00>
+; AUTO_VEC-NEXT:    [[STEP_ADD6:%.*]] = fadd fast <8 x float> [[VEC_IND]], <float 1.200000e+01, float 1.200000e+01, float 1.200000e+01, float 1.200000e+01, float 1.200000e+01, float 1.200000e+01, float 1.200000e+01, float 1.200000e+01>
+; AUTO_VEC-NEXT:    [[TMP1:%.*]] = getelementptr inbounds float, float* [[A:%.*]], i64 [[INDEX]]
+; AUTO_VEC-NEXT:    [[TMP2:%.*]] = bitcast float* [[TMP1]] to <8 x float>*
+; AUTO_VEC-NEXT:    store <8 x float> [[VEC_IND]], <8 x float>* [[TMP2]], align 4
+; AUTO_VEC-NEXT:    [[TMP3:%.*]] = getelementptr inbounds float, float* [[TMP1]], i64 8
+; AUTO_VEC-NEXT:    [[TMP4:%.*]] = bitcast float* [[TMP3]] to <8 x float>*
+; AUTO_VEC-NEXT:    store <8 x float> [[STEP_ADD]], <8 x float>* [[TMP4]], align 4
+; AUTO_VEC-NEXT:    [[TMP5:%.*]] = getelementptr inbounds float, float* [[TMP1]], i64 16
+; AUTO_VEC-NEXT:    [[TMP6:%.*]] = bitcast float* [[TMP5]] to <8 x float>*
+; AUTO_VEC-NEXT:    store <8 x float> [[STEP_ADD5]], <8 x float>* [[TMP6]], align 4
+; AUTO_VEC-NEXT:    [[TMP7:%.*]] = getelementptr inbounds float, float* [[TMP1]], i64 24
+; AUTO_VEC-NEXT:    [[TMP8:%.*]] = bitcast float* [[TMP7]] to <8 x float>*
+; AUTO_VEC-NEXT:    store <8 x float> [[STEP_ADD6]], <8 x float>* [[TMP8]], align 4
+; AUTO_VEC-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 32
+; AUTO_VEC-NEXT:    [[VEC_IND_NEXT]] = fadd fast <8 x float> [[VEC_IND]], <float 1.600000e+01, float 1.600000e+01, float 1.600000e+01, float 1.600000e+01, float 1.600000e+01, float 1.600000e+01, float 1.600000e+01, float 1.600000e+01>
+; AUTO_VEC-NEXT:    [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; AUTO_VEC-NEXT:    br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop !0
+; AUTO_VEC:       middle.block:
+; AUTO_VEC-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[ZEXT]]
+; AUTO_VEC-NEXT:    br i1 [[CMP_N]], label [[FOR_END]], label [[FOR_BODY]]
+; AUTO_VEC:       for.body:
+; AUTO_VEC-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
+; AUTO_VEC-NEXT:    [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ 1.000000e+00, [[FOR_BODY_PREHEADER]] ], [ [[IND_END]], [[MIDDLE_BLOCK]] ]
+; AUTO_VEC-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[INDVARS_IV]]
+; AUTO_VEC-NEXT:    store float [[X_06]], float* [[ARRAYIDX]], align 4
+; AUTO_VEC-NEXT:    [[CONV1]] = fadd float [[X_06]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
+; AUTO_VEC-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[ZEXT]]
+; AUTO_VEC-NEXT:    br i1 [[TMP10]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop !2
+; AUTO_VEC:       for.end:
+; AUTO_VEC-NEXT:    ret void
+;
 entry:
   %cmp4 = icmp sgt i32 %N, 0
   br i1 %cmp4, label %for.body.preheader, label %for.end
@@ -52,11 +99,78 @@ for.end:                                          ; preds = %for.end.loopexit, %
 ;  }
 ;}
 
-; AUTO_VEC-LABEL: @fp_iv_loop2(
-; AUTO_VEC-NOT: vector.body
-; AUTO_VEC-NOT: store <{{.*}} x float>
-
 define void @fp_iv_loop2(float* noalias nocapture %A, i32 %N) #1 {
+; AUTO_VEC-LABEL: @fp_iv_loop2(
+; AUTO_VEC-NEXT:  entry:
+; AUTO_VEC-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[N:%.*]], 0
+; AUTO_VEC-NEXT:    br i1 [[CMP4]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
+; AUTO_VEC:       for.body.preheader:
+; AUTO_VEC-NEXT:    [[ZEXT:%.*]] = zext i32 [[N]] to i64
+; AUTO_VEC-NEXT:    [[TMP0:%.*]] = add nsw i64 [[ZEXT]], -1
+; AUTO_VEC-NEXT:    [[XTRAITER:%.*]] = and i64 [[ZEXT]], 7
+; AUTO_VEC-NEXT:    [[TMP1:%.*]] = icmp ult i64 [[TMP0]], 7
+; AUTO_VEC-NEXT:    br i1 [[TMP1]], label [[FOR_END_LOOPEXIT_UNR_LCSSA:%.*]], label [[FOR_BODY_PREHEADER_NEW:%.*]]
+; AUTO_VEC:       for.body.preheader.new:
+; AUTO_VEC-NEXT:    [[UNROLL_ITER:%.*]] = sub nsw i64 [[ZEXT]], [[XTRAITER]]
+; AUTO_VEC-NEXT:    br label [[FOR_BODY:%.*]]
+; AUTO_VEC:       for.body:
+; AUTO_VEC-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, [[FOR_BODY_PREHEADER_NEW]] ], [ [[INDVARS_IV_NEXT_7:%.*]], [[FOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[X_06:%.*]] = phi float [ 1.000000e+00, [[FOR_BODY_PREHEADER_NEW]] ], [ [[CONV1_7:%.*]], [[FOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[NITER:%.*]] = phi i64 [ [[UNROLL_ITER]], [[FOR_BODY_PREHEADER_NEW]] ], [ [[NITER_NSUB_7:%.*]], [[FOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[A:%.*]], i64 [[INDVARS_IV]]
+; AUTO_VEC-NEXT:    store float [[X_06]], float* [[ARRAYIDX]], align 4
+; AUTO_VEC-NEXT:    [[CONV1:%.*]] = fadd float [[X_06]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[INDVARS_IV_NEXT:%.*]] = or i64 [[INDVARS_IV]], 1
+; AUTO_VEC-NEXT:    [[ARRAYIDX_1:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[INDVARS_IV_NEXT]]
+; AUTO_VEC-NEXT:    store float [[CONV1]], float* [[ARRAYIDX_1]], align 4
+; AUTO_VEC-NEXT:    [[CONV1_1:%.*]] = fadd float [[CONV1]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[INDVARS_IV_NEXT_1:%.*]] = or i64 [[INDVARS_IV]], 2
+; AUTO_VEC-NEXT:    [[ARRAYIDX_2:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[INDVARS_IV_NEXT_1]]
+; AUTO_VEC-NEXT:    store float [[CONV1_1]], float* [[ARRAYIDX_2]], align 4
+; AUTO_VEC-NEXT:    [[CONV1_2:%.*]] = fadd float [[CONV1_1]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[INDVARS_IV_NEXT_2:%.*]] = or i64 [[INDVARS_IV]], 3
+; AUTO_VEC-NEXT:    [[ARRAYIDX_3:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[INDVARS_IV_NEXT_2]]
+; AUTO_VEC-NEXT:    store float [[CONV1_2]], float* [[ARRAYIDX_3]], align 4
+; AUTO_VEC-NEXT:    [[CONV1_3:%.*]] = fadd float [[CONV1_2]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[INDVARS_IV_NEXT_3:%.*]] = or i64 [[INDVARS_IV]], 4
+; AUTO_VEC-NEXT:    [[ARRAYIDX_4:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[INDVARS_IV_NEXT_3]]
+; AUTO_VEC-NEXT:    store float [[CONV1_3]], float* [[ARRAYIDX_4]], align 4
+; AUTO_VEC-NEXT:    [[CONV1_4:%.*]] = fadd float [[CONV1_3]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[INDVARS_IV_NEXT_4:%.*]] = or i64 [[INDVARS_IV]], 5
+; AUTO_VEC-NEXT:    [[ARRAYIDX_5:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[INDVARS_IV_NEXT_4]]
+; AUTO_VEC-NEXT:    store float [[CONV1_4]], float* [[ARRAYIDX_5]], align 4
+; AUTO_VEC-NEXT:    [[CONV1_5:%.*]] = fadd float [[CONV1_4]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[INDVARS_IV_NEXT_5:%.*]] = or i64 [[INDVARS_IV]], 6
+; AUTO_VEC-NEXT:    [[ARRAYIDX_6:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[INDVARS_IV_NEXT_5]]
+; AUTO_VEC-NEXT:    store float [[CONV1_5]], float* [[ARRAYIDX_6]], align 4
+; AUTO_VEC-NEXT:    [[CONV1_6:%.*]] = fadd float [[CONV1_5]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[INDVARS_IV_NEXT_6:%.*]] = or i64 [[INDVARS_IV]], 7
+; AUTO_VEC-NEXT:    [[ARRAYIDX_7:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[INDVARS_IV_NEXT_6]]
+; AUTO_VEC-NEXT:    store float [[CONV1_6]], float* [[ARRAYIDX_7]], align 4
+; AUTO_VEC-NEXT:    [[CONV1_7]] = fadd float [[CONV1_6]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[INDVARS_IV_NEXT_7]] = add nuw nsw i64 [[INDVARS_IV]], 8
+; 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_LOOPEXIT_UNR_LCSSA]], label [[FOR_BODY]]
+; AUTO_VEC:       for.end.loopexit.unr-lcssa:
+; AUTO_VEC-NEXT:    [[INDVARS_IV_UNR:%.*]] = phi i64 [ 0, [[FOR_BODY_PREHEADER]] ], [ [[INDVARS_IV_NEXT_7]], [[FOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[X_06_UNR:%.*]] = phi float [ 1.000000e+00, [[FOR_BODY_PREHEADER]] ], [ [[CONV1_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:    [[INDVARS_IV_EPIL:%.*]] = phi i64 [ [[INDVARS_IV_NEXT_EPIL:%.*]], [[FOR_BODY_EPIL]] ], [ [[INDVARS_IV_UNR]], [[FOR_END_LOOPEXIT_UNR_LCSSA]] ]
+; AUTO_VEC-NEXT:    [[X_06_EPIL:%.*]] = phi float [ [[CONV1_EPIL:%.*]], [[FOR_BODY_EPIL]] ], [ [[X_06_UNR]], [[FOR_END_LOOPEXIT_UNR_LCSSA]] ]
+; AUTO_VEC-NEXT:    [[EPIL_ITER:%.*]] = phi i64 [ [[EPIL_ITER_SUB:%.*]], [[FOR_BODY_EPIL]] ], [ [[XTRAITER]], [[FOR_END_LOOPEXIT_UNR_LCSSA]] ]
+; AUTO_VEC-NEXT:    [[ARRAYIDX_EPIL:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[INDVARS_IV_EPIL]]
+; AUTO_VEC-NEXT:    store float [[X_06_EPIL]], float* [[ARRAYIDX_EPIL]], align 4
+; AUTO_VEC-NEXT:    [[CONV1_EPIL]] = fadd float [[X_06_EPIL]], 5.000000e-01
+; AUTO_VEC-NEXT:    [[INDVARS_IV_NEXT_EPIL]] = add nuw nsw i64 [[INDVARS_IV_EPIL]], 1
+; 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 !4
+; AUTO_VEC:       for.end:
+; AUTO_VEC-NEXT:    ret void
+;
 entry:
   %cmp4 = icmp sgt i32 %N, 0
   br i1 %cmp4, label %for.body.preheader, label %for.end
@@ -82,24 +196,59 @@ for.end:                                          ; preds = %for.end.loopexit, %
   ret void
 }
 
+define double @external_use_with_fast_math(double* %a, i64 %n) {
 ; AUTO_VEC-LABEL: @external_use_with_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:         br i1 {{.*}}, label %for.body, label %vector.ph
+; AUTO_VEC-NEXT:    [[TMP0:%.*]] = icmp sgt i64 [[N:%.*]], 1
+; AUTO_VEC-NEXT:    [[SMAX:%.*]] = select i1 [[TMP0]], i64 [[N]], i64 1
+; AUTO_VEC-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[SMAX]], 16
+; AUTO_VEC-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[FOR_BODY:%.*]], label [[VECTOR_PH:%.*]]
 ; AUTO_VEC:       vector.ph:
 ; AUTO_VEC-NEXT:    [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775792
-; AUTO_VEC:         br label %vector.body
+; AUTO_VEC-NEXT:    [[CAST_CRD:%.*]] = sitofp i64 [[N_VEC]] to double
+; AUTO_VEC-NEXT:    [[TMP1:%.*]] = fmul fast double [[CAST_CRD]], 3.000000e+00
+; AUTO_VEC-NEXT:    br label [[VECTOR_BODY:%.*]]
+; AUTO_VEC:       vector.body:
+; AUTO_VEC-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[VEC_IND:%.*]] = phi <4 x double> [ <double 0.000000e+00, double 3.000000e+00, double 6.000000e+00, double 9.000000e+00>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AUTO_VEC-NEXT:    [[STEP_ADD:%.*]] = fadd fast <4 x double> [[VEC_IND]], <double 1.200000e+01, double 1.200000e+01, double 1.200000e+01, double 1.200000e+01>
+; AUTO_VEC-NEXT:    [[STEP_ADD5:%.*]] = fadd fast <4 x double> [[VEC_IND]], <double 2.400000e+01, double 2.400000e+01, double 2.400000e+01, double 2.400000e+01>
+; AUTO_VEC-NEXT:    [[STEP_ADD6:%.*]] = fadd fast <4 x double> [[VEC_IND]], <double 3.600000e+01, double 3.600000e+01, double 3.600000e+01, double 3.600000e+01>
+; AUTO_VEC-NEXT:    [[TMP2:%.*]] = getelementptr double, double* [[A:%.*]], i64 [[INDEX]]
+; AUTO_VEC-NEXT:    [[TMP3:%.*]] = bitcast double* [[TMP2]] to <4 x double>*
+; AUTO_VEC-NEXT:    store <4 x double> [[VEC_IND]], <4 x double>* [[TMP3]], align 8
+; AUTO_VEC-NEXT:    [[TMP4:%.*]] = getelementptr double, double* [[TMP2]], i64 4
+; AUTO_VEC-NEXT:    [[TMP5:%.*]] = bitcast double* [[TMP4]] to <4 x double>*
+; AUTO_VEC-NEXT:    store <4 x double> [[STEP_ADD]], <4 x double>* [[TMP5]], align 8
+; AUTO_VEC-NEXT:    [[TMP6:%.*]] = getelementptr double, double* [[TMP2]], i64 8
+; AUTO_VEC-NEXT:    [[TMP7:%.*]] = bitcast double* [[TMP6]] to <4 x double>*
+; AUTO_VEC-NEXT:    store <4 x double> [[STEP_ADD5]], <4 x double>* [[TMP7]], align 8
+; AUTO_VEC-NEXT:    [[TMP8:%.*]] = getelementptr double, double* [[TMP2]], i64 12
+; AUTO_VEC-NEXT:    [[TMP9:%.*]] = bitcast double* [[TMP8]] to <4 x double>*
+; AUTO_VEC-NEXT:    store <4 x double> [[STEP_ADD6]], <4 x double>* [[TMP9]], align 8
+; AUTO_VEC-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
+; AUTO_VEC-NEXT:    [[VEC_IND_NEXT]] = fadd fast <4 x double> [[VEC_IND]], <double 4.800000e+01, double 4.800000e+01, double 4.800000e+01, double 4.800000e+01>
+; AUTO_VEC-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; AUTO_VEC-NEXT:    br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop !6
 ; AUTO_VEC:       middle.block:
-; AUTO_VEC:         [[TMP11:%.*]] = add nsw i64 [[N_VEC]], -1
+; AUTO_VEC-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[SMAX]], [[N_VEC]]
+; AUTO_VEC-NEXT:    [[TMP11:%.*]] = add nsw i64 [[N_VEC]], -1
 ; AUTO_VEC-NEXT:    [[CAST_CMO:%.*]] = sitofp i64 [[TMP11]] to double
 ; AUTO_VEC-NEXT:    [[TMP12:%.*]] = fmul fast double [[CAST_CMO]], 3.000000e+00
-; AUTO_VEC-NEXT:    br i1 {{.*}}, label %for.end, label %for.body
+; AUTO_VEC-NEXT:    br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[FOR_BODY]]
+; AUTO_VEC:       for.body:
+; AUTO_VEC-NEXT:    [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
+; AUTO_VEC-NEXT:    [[J:%.*]] = phi double [ [[J_NEXT:%.*]], [[FOR_BODY]] ], [ 0.000000e+00, [[ENTRY]] ], [ [[TMP1]], [[MIDDLE_BLOCK]] ]
+; 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 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 !7
 ; AUTO_VEC:       for.end:
-; AUTO_VEC-NEXT:    [[J_LCSSA:%.*]] = phi double [ [[TMP12]], %middle.block ], [ %j, %for.body ]
+; AUTO_VEC-NEXT:    [[J_LCSSA:%.*]] = phi double [ [[TMP12]], [[MIDDLE_BLOCK]] ], [ [[J]], [[FOR_BODY]] ]
 ; AUTO_VEC-NEXT:    ret double [[J_LCSSA]]
 ;
-define double @external_use_with_fast_math(double* %a, i64 %n) {
 entry:
   br label %for.body
 
@@ -118,15 +267,22 @@ for.end:
   ret double %tmp1
 }
 
+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:    br label [[FOR_BODY:%.*]]
 ; AUTO_VEC:       for.body:
-; AUTO_VEC:         [[J:%.*]] = phi double [ 0.000000e+00, %entry ], [ [[J_NEXT:%.*]], %for.body ]
-; AUTO_VEC:         [[J_NEXT]] = fadd double [[J]], 3.000000e+00
-; AUTO_VEC:         br i1 {{.*}}, label %for.body, label %for.end
+; 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:    [[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:       for.end:
 ; AUTO_VEC-NEXT:    ret double [[J]]
 ;
-define double @external_use_without_fast_math(double* %a, i64 %n) {
 entry:
   br label %for.body