From a performance point of view, which one is better:
... /* 10 recv handlers */
I understand MPI_Brecv will require an extra message copy, from the
attached buffer to the MPI_Brecv() buffer. I'd like to know if there
other differences between the two methods.