This is the baseline version of the pils program. The program iterates on
different phases, splitted by MPI_barrier()
’s. Each phase execute a given
number of parallel regions (determined by the grain
parameter), which main
body is a loop executing iterations of task-duration
.