In our second MPI lesson we met two new functions:
MPI_Reduce
performs the reduction operation on
data sent to a designated process; explained in section
4.5
MPI_Wtime
times the process with respect to wall clock;
explained in section 4.6.4
We also demonstrated how to distribute a computational task between processors, and how we may end up producing a code which does not scale with the size of the problem if the task is partitioned equally, regardless of the differences in the speed of the nodes participating in the cluster.