1; RUN: opt %loadPolly -basic-aa -polly-codegen -polly-vectorizer=polly  -dce -S < %s | FileCheck %s
2
3;#define N 1024
4;float A[N];
5;float B[N];
6;
7;void simple_vec_stride_x(void) {
8;  int i;
9;
10;  for (i = 0; i < 4; i++)
11;    B[2 * i] = A[2 * i];
12;}
13;int main()
14;{
15;  simple_vec_stride_x();
16;  return A[42];
17;}
18
19target 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"
20
21@A = common global [1024 x float] zeroinitializer, align 16
22@B = common global [1024 x float] zeroinitializer, align 16
23
24define void @simple_vec_stride_x() nounwind {
25bb:
26  br label %bb2
27
28bb2:                                              ; preds = %bb5, %bb
29  %indvar = phi i64 [ %indvar.next, %bb5 ], [ 0, %bb ]
30  %tmp = mul i64 %indvar, 2
31  %scevgep = getelementptr [1024 x float], [1024 x float]* @B, i64 0, i64 %tmp
32  %scevgep1 = getelementptr [1024 x float], [1024 x float]* @A, i64 0, i64 %tmp
33  %exitcond = icmp ne i64 %indvar, 4
34  br i1 %exitcond, label %bb3, label %bb6
35
36bb3:                                              ; preds = %bb2
37  %tmp4 = load float, float* %scevgep1, align 8
38  store float %tmp4, float* %scevgep, align 8
39  br label %bb5
40
41bb5:                                              ; preds = %bb3
42  %indvar.next = add i64 %indvar, 1
43  br label %bb2
44
45bb6:                                              ; preds = %bb2
46  ret void
47}
48
49define i32 @main() nounwind {
50bb:
51  call void @simple_vec_stride_x()
52  %tmp = load float, float* getelementptr inbounds ([1024 x float], [1024 x float]* @A, i64 0, i64 42), align 8
53  %tmp1 = fptosi float %tmp to i32
54  ret i32 %tmp1
55}
56
57; CHECK: [[LOAD1:%[a-zA-Z0-9_]+_scalar_]] = load float, float*
58; CHECK: [[VEC1:%[a-zA-Z0-9_]+]] = insertelement <4 x float> undef, float [[LOAD1]], i32 0
59; CHECK: [[LOAD2:%[a-zA-Z0-9_]+]] = load float, float*
60; CHECK: [[VEC2:%[a-zA-Z0-9_]+]] = insertelement <4 x float> [[VEC1]], float [[LOAD2]], i32 1
61; CHECK: [[LOAD3:%[a-zA-Z0-9_]+]] = load float, float*
62; CHECK: [[VEC3:%[a-zA-Z0-9_]+]] = insertelement <4 x float> [[VEC2]], float [[LOAD3]], i32 2
63; CHECK: [[LOAD4:%[a-zA-Z0-9_]+]] = load float, float*
64; CHECK: [[VEC4:%[a-zA-Z0-9_]+]] = insertelement <4 x float> [[VEC3]], float [[LOAD4]], i32 3
65; CHECK: [[EL1:%[a-zA-Z0-9_]+]] = extractelement <4 x float> [[VEC4]], i32 0
66; CHECK: store float [[EL1]]
67; CHECK: [[EL2:%[a-zA-Z0-9_]+]] = extractelement <4 x float> [[VEC4]], i32 1
68; CHECK: store float [[EL2]]
69; CHECK: [[EL3:%[a-zA-Z0-9_]+]] = extractelement <4 x float> [[VEC4]], i32 2
70; CHECK: store float [[EL3]]
71; CHECK: [[EL4:%[a-zA-Z0-9_]+]] = extractelement <4 x float> [[VEC4]], i32 3
72; CHECK: store float [[EL4]]
73