Take the previous example but replace the
MPI_Send
calls with
MPI_Isend
. Perform an
MPI_Waitall
on all eight requests. Observe the time that communication takes, and compare it with the
MPI_Send
version.