Parallel Implementation

Hi everybody,

First of all, this forum is doing a wonderful job in helping the newbies. I am planning to make my code parallel. Can anyone suggest which one to use (OpenMP, MPI etc…). and why.

Thanks a lot,
Narender

Hi,

if you need quick and probably the simplest possible parallelization, that will split the load between multiple cores on the same computer, use OpenMP. As long as your code will be local. I suppose, that the main loop of the program is iteration through the lattice points (probably in for loop). Only thing you need, to parallelize it to multiple cores, is to add


#pragma omp parallel for

directive right before the for-loop.

Nothing too sophisticated, but has probably the best effort/effect ratio :wink:

Hi AirGuru,

Thanks for the information. :slight_smile: . I am working on OpenMP now.

Regards,
Narender