gcc/testsuite/gcc.dg/gomp/pr68128-1.c

   1 /* PR tree-optimization/68128 */
   2 /* { dg-do compile } */
   3 /* { dg-options "-Ofast -fopenmp -fdump-tree-vect-details" } */
   4 /* { dg-additional-options "-mavx" { target i?86-*-* x86_64-*-* } } */
   5
   6 /* Make sure the following loop is vectorized even when not using
   7    firstprivate variables for scalar vars that are not modified
   8    in the parallel region.  */
   9
  10 void
  11 foo (float *u, float v, float w, float x, float y, float z, float t)
  12 {
  13   int i, j, k, l;
  14   float a, *b, c, s, e;
  15 #pragma omp parallel for private (i, j, k, l, a, b, c, s, e)
  16   for (j = 0; j < 1024; j++)
  17     {
  18       k = j * 64;
  19       l = j * 64 + 63;
  20       a = v + j * w;
  21       b = u + j * 64;
  22       for (i = k; i <= l; i++, b++, a += w)
  23         {
  24           c = a * a + y;
  25           s = (1.f - c * x) * (1.f - c * x);
  26           e = t * (1 / __builtin_sqrtf (c)) * s;
  27           *b += (c < z ? e : 0);
  28         }
  29     }
  30 }
  31
  32 /* { dg-final { scan-tree-dump "vectorized 1 loops in function" "vect" { target i?86-*-* x86_64-*-* } } } */