1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: opt < %s -basic-aa -slp-vectorizer -S -mcpu=corei7-avx | FileCheck %s
3
4target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
5target triple = "x86_64-apple-macosx10.8.0"
6
7define i32 @foo(i32* nocapture %A, i32 %n) {
8; CHECK-LABEL: @foo(
9; CHECK-NEXT:  entry:
10; CHECK-NEXT:    [[CMP62:%.*]] = icmp sgt i32 [[N:%.*]], 0
11; CHECK-NEXT:    br i1 [[CMP62]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
12; CHECK:       for.body:
13; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
14; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 [[INDVARS_IV]]
15; CHECK-NEXT:    [[TMP0:%.*]] = or i64 [[INDVARS_IV]], 1
16; CHECK-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP0]]
17; CHECK-NEXT:    [[TMP1:%.*]] = or i64 [[INDVARS_IV]], 2
18; CHECK-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP1]]
19; CHECK-NEXT:    [[TMP2:%.*]] = or i64 [[INDVARS_IV]], 3
20; CHECK-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP2]]
21; CHECK-NEXT:    [[TMP3:%.*]] = or i64 [[INDVARS_IV]], 4
22; CHECK-NEXT:    [[ARRAYIDX16:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP3]]
23; CHECK-NEXT:    [[TMP4:%.*]] = or i64 [[INDVARS_IV]], 5
24; CHECK-NEXT:    [[ARRAYIDX20:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP4]]
25; CHECK-NEXT:    [[TMP5:%.*]] = or i64 [[INDVARS_IV]], 6
26; CHECK-NEXT:    [[ARRAYIDX24:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP5]]
27; CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[INDVARS_IV]], 7
28; CHECK-NEXT:    [[ARRAYIDX28:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP6]]
29; CHECK-NEXT:    [[TMP7:%.*]] = bitcast i32* [[ARRAYIDX]] to <8 x i32>*
30; CHECK-NEXT:    [[TMP8:%.*]] = load <8 x i32>, <8 x i32>* [[TMP7]], align 4
31; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <8 x i32> poison, i32 [[N]], i32 0
32; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <8 x i32> [[TMP9]], <8 x i32> poison, <8 x i32> zeroinitializer
33; CHECK-NEXT:    [[TMP10:%.*]] = add nsw <8 x i32> [[TMP8]], [[SHUFFLE]]
34; CHECK-NEXT:    [[TMP11:%.*]] = bitcast i32* [[ARRAYIDX]] to <8 x i32>*
35; CHECK-NEXT:    store <8 x i32> [[TMP10]], <8 x i32>* [[TMP11]], align 4
36; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 8
37; CHECK-NEXT:    [[TMP12:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
38; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP12]], [[N]]
39; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END]]
40; CHECK:       for.end:
41; CHECK-NEXT:    ret i32 undef
42;
43entry:
44  %cmp62 = icmp sgt i32 %n, 0
45  br i1 %cmp62, label %for.body, label %for.end
46
47for.body:
48  %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
49  %arrayidx = getelementptr inbounds i32, i32* %A, i64 %indvars.iv
50  %0 = load i32, i32* %arrayidx, align 4
51  %add1 = add nsw i32 %0, %n
52  store i32 %add1, i32* %arrayidx, align 4
53  %1 = or i64 %indvars.iv, 1
54  %arrayidx4 = getelementptr inbounds i32, i32* %A, i64 %1
55  %2 = load i32, i32* %arrayidx4, align 4
56  %add5 = add nsw i32 %2, %n
57  store i32 %add5, i32* %arrayidx4, align 4
58  %3 = or i64 %indvars.iv, 2
59  %arrayidx8 = getelementptr inbounds i32, i32* %A, i64 %3
60  %4 = load i32, i32* %arrayidx8, align 4
61  %add9 = add nsw i32 %4, %n
62  store i32 %add9, i32* %arrayidx8, align 4
63  %5 = or i64 %indvars.iv, 3
64  %arrayidx12 = getelementptr inbounds i32, i32* %A, i64 %5
65  %6 = load i32, i32* %arrayidx12, align 4
66  %add13 = add nsw i32 %6, %n
67  store i32 %add13, i32* %arrayidx12, align 4
68  %7 = or i64 %indvars.iv, 4
69  %arrayidx16 = getelementptr inbounds i32, i32* %A, i64 %7
70  %8 = load i32, i32* %arrayidx16, align 4
71  %add17 = add nsw i32 %8, %n
72  store i32 %add17, i32* %arrayidx16, align 4
73  %9 = or i64 %indvars.iv, 5
74  %arrayidx20 = getelementptr inbounds i32, i32* %A, i64 %9
75  %10 = load i32, i32* %arrayidx20, align 4
76  %add21 = add nsw i32 %10, %n
77  store i32 %add21, i32* %arrayidx20, align 4
78  %11 = or i64 %indvars.iv, 6
79  %arrayidx24 = getelementptr inbounds i32, i32* %A, i64 %11
80  %12 = load i32, i32* %arrayidx24, align 4
81  %add25 = add nsw i32 %12, %n
82  store i32 %add25, i32* %arrayidx24, align 4
83  %13 = or i64 %indvars.iv, 7
84  %arrayidx28 = getelementptr inbounds i32, i32* %A, i64 %13
85  %14 = load i32, i32* %arrayidx28, align 4
86  %add29 = add nsw i32 %14, %n
87  store i32 %add29, i32* %arrayidx28, align 4
88  %indvars.iv.next = add i64 %indvars.iv, 8
89  %15 = trunc i64 %indvars.iv.next to i32
90  %cmp = icmp slt i32 %15, %n
91  br i1 %cmp, label %for.body, label %for.end
92
93for.end:
94  ret i32 undef
95}
96
97