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

   1 /* Check offloaded function's attributes and classification for unparallelized
   2    OpenACC 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 /* An "extern"al mapping of loop iterations/array indices makes the loop
  16    unparallelizable.  */
  17 extern unsigned int f (unsigned int);
  18
  19 void KERNELS ()
  20 {
  21 #pragma acc kernels copyin (a[0:N], b[0:N]) copyout (c[0:N])
  22   for (unsigned int i = 0; i < N; i++)
  23     c[i] = a[f (i)] + b[f (i)];
  24 }
  25
  26 /* Check the offloaded function's attributes.
  27    { dg-final { scan-tree-dump-times "(?n)__attribute__\\(\\(oacc kernels, omp target entrypoint\\)\\)" 1 "ompexp" } } */
  28
  29 /* Check that exactly one OpenACC kernels construct is analyzed, and that it
  30    can't be parallelized.
  31    { dg-final { scan-tree-dump-times "FAILED:" 1 "parloops1" } }
  32    { dg-final { scan-tree-dump-times "(?n)__attribute__\\(\\(oacc function \\(, , \\), oacc kernels, omp target entrypoint\\)\\)" 1 "parloops1" } }
  33    { dg-final { scan-tree-dump-not "SUCCESS: may be parallelized" "parloops1" } } */
  34
  35 /* Check the offloaded function's classification and compute dimensions (will
  36    always be 1 x 1 x 1 for non-offloading compilation).
  37    { dg-final { scan-tree-dump-times "(?n)Function is OpenACC kernels offload" 1 "oaccdevlow" } }
  38    { dg-final { scan-tree-dump-times "(?n)Compute dimensions \\\[1, 1, 1\\\]" 1 "oaccdevlow" } }
  39    { dg-final { scan-tree-dump-times "(?n)__attribute__\\(\\(oacc function \\(1, 1, 1\\), oacc kernels, omp target entrypoint\\)\\)" 1 "oaccdevlow" } } */