gcc/testsuite/c-c++-common/goacc/classify-kernels.c

   1 /* Check offloaded function's attributes and classification for OpenACC
   2    kernels.  */
   3
   4 /* { dg-additional-options "-O2" }
   5    { dg-additional-options "-fdump-tree-ompexp" }
   6    { dg-additional-options "-fdump-tree-parloops1-all" }
   7    { dg-additional-options "-fdump-tree-oaccdevlow" } */
   8
   9 #define N 1024
  10
  11 extern unsigned int *__restrict a;
  12 extern unsigned int *__restrict b;
  13 extern unsigned int *__restrict c;
  14
  15 void KERNELS ()
  16 {
  17 #pragma acc kernels copyin (a[0:N], b[0:N]) copyout (c[0:N])
  18   for (unsigned int i = 0; i < N; i++)
  19     c[i] = a[i] + b[i];
  20 }
  21
  22 /* Check the offloaded function's attributes.
  23    { dg-final { scan-tree-dump-times "(?n)__attribute__\\(\\(oacc kernels, omp target entrypoint\\)\\)" 1 "ompexp" } } */
  24
  25 /* Check that exactly one OpenACC kernels construct is analyzed, and that it
  26    can be parallelized.
  27    { dg-final { scan-tree-dump-times "SUCCESS: may be parallelized" 1 "parloops1" } }
  28    { dg-final { scan-tree-dump-times "(?n)__attribute__\\(\\(oacc kernels parallelized, oacc function \\(, , \\), oacc kernels, omp target entrypoint\\)\\)" 1 "parloops1" } }
  29    { dg-final { scan-tree-dump-not "FAILED:" "parloops1" } } */
  30
  31 /* Check the offloaded function's classification and compute dimensions (will
  32    always be 1 x 1 x 1 for non-offloading compilation).
  33    { dg-final { scan-tree-dump-times "(?n)Function is parallelized OpenACC kernels offload" 1 "oaccdevlow" } }
  34    { dg-final { scan-tree-dump-times "(?n)Compute dimensions \\\[1, 1, 1\\\]" 1 "oaccdevlow" } }
  35    { dg-final { scan-tree-dump-times "(?n)__attribute__\\(\\(oacc function \\(1, 1, 1\\), oacc kernels parallelized, oacc function \\(, , \\), oacc kernels, omp target entrypoint\\)\\)" 1 "oaccdevlow" } } */