On Feb 27, 2012, at 4:42 AM, Paul Kapinos wrote:
> Dear Open MPI developer,
> are enlisted four envvars Open MPI set for every process. We use they for some scripting and thank you for providing they.
> But simple "mpiexec -np 1 env | grep OMPI" brings lotz more enviers.
Yes, we set quite a few more, but those are intended solely for internal use and are not guaranteed. The list on the web site only identifies a set that are guaranteed to be provided.
> These are interesting for us:
> 1) OMPI_COMM_WORLD_LOCAL_SIZE - seem to contain the number of processes which are running on the specific node, see also
> Is this envvar also "stable" as OMPI_COMM_WORLD_LOCAL_RANK is? (This would make sense as it looks like the OMPI_COMM_WORLD_SIZE, OMPI_COMM_WORLD_RANK pair.)
Yes, and I'll add it to the page
> If yes, maybe it also should be documented in the Wiki page.
> 2) OMPI_COMM_WORLD_NODE_RANK - is that just a double of OMPI_COMM_WORLD_LOCAL_RANK ?
No - the "local rank" is your rank on the node within your own job. The "node rank" is your rank on the node overall. The two differ when you do a comm_spawn. For example, suppose you have two ranks from your initial job on a node, and then comm_spawn three additional ranks. Their values would look like this:
job/rank local_rank node_rank
0/0 0 0
0/1 1 1
1/0 0 2
1/1 1 3
1/2 2 4
Again, I'll add it to the page
> Best wishes,
> Paul Kapinos
> Dipl.-Inform. Paul Kapinos - High Performance Computing,
> RWTH Aachen University, Center for Computing and Communication
> Seffenter Weg 23, D 52074 Aachen (Germany)
> Tel: +49 241/80-24915
> users mailing list