Add FAQ entries about non-MPI performance issues

ompiteam commented 10 years ago

From an email thread with Eugene and Terry, it might be good to have some additional info in our FAQ about optimization that may not be germane to MPI issues:

...one possibility is to have a section that discusses how different MPI implementations may differ in performance in "non-MPI" kinds of ways. E.g.,

Some MPI implementations add optimization flags to their mpicc/mpif90 wrappers by default. So, default compilation with one MPI gives you optimization, while another does not.
Different MPI implementations have different default process binding behaviors.
The whole MALLOC issue.

Arguably, these issues aren't inherent to MPI. They are general HPC performance tuning considerations. It's a slippery slope to get into these issues (our users care about performance and these things are important to performance, therefore we should tell our users about these things, but where do we stop?). On the other hand, given that these are knobs that various MPI implementations touch (and touch in different ways), they may make sense for us to discuss.

ompiteam commented 10 years ago

Imported from trac issue 2198. Created by jsquyres on 2010-01-25T23:29:51, last modified: 2013-12-13T13:50:38

ompiteam commented 10 years ago

Trac comment by bbenton on 2012-02-21 13:40:41:

Milestone Open MPI 1.4.5 deleted

ompiteam commented 10 years ago

Trac comment by bbenton on 2012-10-10 17:50:24:

Moving to 1.6 series for consideration there.

open-mpi / ompi

Add FAQ entries about non-MPI performance issues #97