1 /* { dg-require-effective-target vect_int } */
2
3 #include <stdarg.h>
4 #include "tree-vect.h"
5
6 #define N 8
7
8 unsigned int in[N*8] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63};
9 unsigned short in2[N*16] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63};
10
11 int
12 main1 (int n)
13 {
14 int i;
15 unsigned int out[N*8], a0, a1, a2, a3, a4, a5, a6, a7, b1, b0, b2, b3, b4, b5, b6, b7;
16 unsigned short out2[N*16];
17
18 /* Multiple types are now SLPable. */
19 for (i = 0; i < n; i++)
20 {
21 a0 = in[i*8] + 5;
22 a1 = in[i*8 + 1] + 6;
23 a2 = in[i*8 + 2] + 7;
24 a3 = in[i*8 + 3] + 8;
25 a4 = in[i*8 + 4] + 9;
26 a5 = in[i*8 + 5] + 10;
27 a6 = in[i*8 + 6] + 11;
28 a7 = in[i*8 + 7] + 12;
29
30 b0 = a0 * 3;
31 b1 = a1 * 2;
32 b2 = a2 * 12;
33 b3 = a3 * 5;
34 b4 = a4 * 8;
35 b5 = a5 * 4;
36 b6 = a6 * 3;
37 b7 = a7 * 2;
38
39 out[i*8] = b0 - 2;
40 out[i*8 + 1] = b1 - 3;
41 out[i*8 + 2] = b2 - 2;
42 out[i*8 + 3] = b3 - 1;
43 out[i*8 + 4] = b4 - 8;
44 out[i*8 + 5] = b5 - 7;
45 out[i*8 + 6] = b6 - 3;
46 out[i*8 + 7] = b7 - 7;
47
48 out2[i*16] = in2[i*16] + 2;
49 out2[i*16 + 1] = in2[i*16 + 1] + 3;
50 out2[i*16 + 2] = in2[i*16 + 2] + 4;
51 out2[i*16 + 3] = in2[i*16 + 3] + 3;
52 out2[i*16 + 4] = in2[i*16 + 4] + 2;
53 out2[i*16 + 5] = in2[i*16 + 5] + 3;
54 out2[i*16 + 6] = in2[i*16 + 6] + 2;
55 out2[i*16 + 7] = in2[i*16 + 7] + 4;
56 out2[i*16 + 8] = in2[i*16 + 8] + 2;
57 out2[i*16 + 9] = in2[i*16 + 9] + 5;
58 out2[i*16 + 10] = in2[i*16 + 10] + 2;
59 out2[i*16 + 11] = in2[i*16 + 11] + 3;
60 out2[i*16 + 12] = in2[i*16 + 12] + 4;
61 out2[i*16 + 13] = in2[i*16 + 13] + 4;
62 out2[i*16 + 14] = in2[i*16 + 14] + 3;
63 out2[i*16 + 15] = in2[i*16 + 15] + 2;
64 }
65
66 /* check results: */
67 for (i = 0; i < n; i++)
68 {
69 if (out[i*8] != (in[i*8] + 5) * 3 - 2
70 || out[i*8 + 1] != (in[i*8 + 1] + 6) * 2 - 3
71 || out[i*8 + 2] != (in[i*8 + 2] + 7) * 12 - 2
72 || out[i*8 + 3] != (in[i*8 + 3] + 8) * 5 - 1
73 || out[i*8 + 4] != (in[i*8 + 4] + 9) * 8 - 8
74 || out[i*8 + 5] != (in[i*8 + 5] + 10) * 4 - 7
75 || out[i*8 + 6] != (in[i*8 + 6] + 11) * 3 - 3
76 || out[i*8 + 7] != (in[i*8 + 7] + 12) * 2 - 7)
77 abort ();
78
79 if (out2[i*16] != in2[i*16] + 2
80 || out2[i*16 + 1] != in2[i*16 + 1] + 3
81 || out2[i*16 + 2] != in2[i*16 + 2] + 4
82 || out2[i*16 + 3] != in2[i*16 + 3] + 3
83 || out2[i*16 + 4] != in2[i*16 + 4] + 2
84 || out2[i*16 + 5] != in2[i*16 + 5] + 3
85 || out2[i*16 + 6] != in2[i*16 + 6] + 2
86 || out2[i*16 + 7] != in2[i*16 + 7] + 4
87 || out2[i*16 + 8] != in2[i*16 + 8] + 2
88 || out2[i*16 + 9] != in2[i*16 + 9] + 5
89 || out2[i*16 + 10] != in2[i*16 + 10] + 2
90 || out2[i*16 + 11] != in2[i*16 + 11] + 3
91 || out2[i*16 + 12] != in2[i*16 + 12] + 4
92 || out2[i*16 + 13] != in2[i*16 + 13] + 4
93 || out2[i*16 + 14] != in2[i*16 + 14] + 3
94 || out2[i*16 + 15] != in2[i*16 + 15] + 2)
95 abort ();
96
97 }
98
99
100
101 return 0;
102 }
103
104 int main (void)
105 {
106 check_vect ();
107
108 main1 (N);
109
110 return 0;
111 }
112
113 /* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" { target vect_int_mult } } } */
114 /* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 2 "vect" { target vect_int_mult } } } */
115