I'm trying to produce a performance model for a piece of software, and I'd
like to know the expected performance behavior of MPI_Scatter and
MPI_Gather as the number of processors increases. I've searched to no
avail for a publication on this topic.
Can you point me in the direction of something like that?