[Bug c/30049] New: Variable-length arrays (VLA) should be converted to normal arrays if possible

[burnus at gcc dot gnu dot org]

The following program (split off from PR30032) needs with
"gcc -march=opteron -ffast-math -funroll-loops -ftree-vectorize -msse3
-O3 -g"
  0m6.478s
but if I use   #define NODES 2500  it only needs only
  0m1.390s

(The Intel compiler seems to do this conversion automatically and needs
0m1.879s for the VLA version of the program.)


void eos(const int NODES, const float CGAMMA, float CS[NODES], float
PRES[NODES], const float DENS[NODES])
{
  int j;
  for(j = 0; j < NODES; j ++)
  {
    CS[j] = sqrt(CGAMMA*PRES[j]/DENS[j]);
  }
}

int main() {
  const int NODES = 25000;
  float CGAMMA;
  float DENS[NODES], CS[NODES], PRES[NODES];
  int i,j;
  for(i = 0; i < NODES; i++) {
     DENS[i] = 3.0;
     PRES[i] = 0.25;
  }
  CGAMMA = 2.0;
  for(i = 0; i < 20000; i++) {
    eos(NODES, CGAMMA, &CS, &PRES, &DENS);
    CGAMMA = CGAMMA + CS[1];
  }
  return (int)CGAMMA;
}


-- 
           Summary: Variable-length arrays (VLA) should be converted to
                    normal arrays if possible
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: burnus at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30049