On May 4, 2011, at 1:51 PM, Tony Lam wrote: > Hi, > > I understand a single orted is shared by all MPI processes from the same > communicator on each execution host, does anyone see any problem that > MPI/OMPI may have problem with each process has its owner orted? My guess it > is less efficient in terms of MPI communication and memory foot print, but > for simplification of our integration with OMPI, launching one orted for each > MPI process is much easier to do.
The orteds won't care, but the mapper may get confused, and the MPI apps definitely will. Locality is based on being connected to the same orted, so the MPI apps will all declare themselves to be on different nodes - which means shared memory will be disabled. If you don't plan to use shared memory, then this won't matter. You'll also see some inefficiencies in out-of-band messaging, for example when infiniband connections are being opened, but that's pretty minor. > > I will appreciate if someone can confirm this setup will or will not work. > > Thanks. > > Tony > > _______________________________________________ > devel mailing list > de...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/devel
