|
111
|
1 /* { dg-additional-options "-O2" } */
|
|
|
2 /* { dg-additional-options "-fdump-tree-parloops1-all" } */
|
|
|
3 /* { dg-additional-options "-fdump-tree-optimized" } */
|
|
|
4
|
|
|
5 #include <stdlib.h>
|
|
|
6
|
|
|
7 #define N ((1024 * 512) + 1)
|
|
|
8 #define COUNTERTYPE unsigned int
|
|
|
9
|
|
|
10 int
|
|
|
11 main (void)
|
|
|
12 {
|
|
|
13 unsigned int *__restrict a;
|
|
|
14 unsigned int *__restrict b;
|
|
|
15 unsigned int *__restrict c;
|
|
|
16
|
|
|
17 a = (unsigned int *__restrict)malloc (N * sizeof (unsigned int));
|
|
|
18 b = (unsigned int *__restrict)malloc (N * sizeof (unsigned int));
|
|
|
19 c = (unsigned int *__restrict)malloc (N * sizeof (unsigned int));
|
|
|
20
|
|
|
21 for (COUNTERTYPE i = 0; i < N; i++)
|
|
|
22 a[i] = i * 2;
|
|
|
23
|
|
|
24 for (COUNTERTYPE i = 0; i < N; i++)
|
|
|
25 b[i] = i * 4;
|
|
|
26
|
|
|
27 #pragma acc kernels copyin (a[0:N], b[0:N]) copyout (c[0:N])
|
|
|
28 {
|
|
|
29 for (COUNTERTYPE ii = 0; ii < N; ii++)
|
|
|
30 c[ii] = a[ii] + b[ii];
|
|
|
31 }
|
|
|
32
|
|
|
33 for (COUNTERTYPE i = 0; i < N; i++)
|
|
|
34 if (c[i] != a[i] + b[i])
|
|
|
35 abort ();
|
|
|
36
|
|
|
37 free (a);
|
|
|
38 free (b);
|
|
|
39 free (c);
|
|
|
40
|
|
|
41 return 0;
|
|
|
42 }
|
|
|
43
|
|
|
44 /* Check that only one loop is analyzed, and that it can be parallelized. */
|
|
|
45 /* { dg-final { scan-tree-dump-times "SUCCESS: may be parallelized" 1 "parloops1" } } */
|
|
|
46 /* { dg-final { scan-tree-dump-times "(?n)__attribute__\\(\\(oacc kernels parallelized, oacc function \\(, , \\), oacc kernels, omp target entrypoint\\)\\)" 1 "parloops1" } } */
|
|
|
47 /* { dg-final { scan-tree-dump-not "FAILED:" "parloops1" } } */
|
|
|
48
|
|
|
49 /* Check that the loop has been split off into a function. */
|
|
|
50 /* { dg-final { scan-tree-dump-times "(?n);; Function .*main._omp_fn.0" 1 "optimized" } } */
|
