1  /* { dg-do compile } */
       2  /* { dg-require-effective-target vect_shift } */
       3  /* { dg-require-effective-target vect_int } */
       4  
       5  typedef unsigned int uint32_t;
       6  typedef short unsigned int uint16_t;
       7  
       8  int foo (uint32_t arr[4][4])
       9  {
      10    int sum = 0;
      11    for(int i = 0; i < 4; i++)
      12      {
      13        sum += ((arr[0][i] >> 10) * 20) + ((arr[1][i] >> 11) & 53)
      14  	     + ((arr[2][i] >> 12) * 7)  + ((arr[3][i] >> 13) ^ 43);
      15      }
      16      return (((uint16_t)sum) + ((uint32_t)sum >> 16)) >> 1;
      17  }
      18  
      19  /* For a target that has a vector/scalar shift/rotate optab, check
      20     that we are not adding the cost of creating a vector from the scalar
      21     in the prologue.  */
      22  /* { dg-final { scan-tree-dump {vectorizable_shift ===[\n\r][^\n]*prologue_cost = 0} "vect" { target { aarch64*-*-* x86_64-*-* } } } } */