Parallel Numerical Simulations of Three-Dimensional Electromagnetic Radiation with MPI-CUDA Paradigms

Algorithm 1

 #pragma omp parallel for schedule(static)

 for (; _tot; ++)

  for (; _tot; ++)

   for (; _tot; ++)

    ⋯

 In order to increase OpenMP load balance and get better scalability, we merge

the nested loops and reduce from 3 nest layers to 2 layers.

 _end = _tot;

 _end = (_tot) * (_tot);

 _start = _end + 1;

 #pragma omp parallel for schedule(static)

 for (_start; _end; ++)

 {

  /_end;

  ;

  if (!) continue;

  for (; _tot; ++)

   ⋯

 }