This web mail archive is frozen.
This page is part of a frozen web archive of this mailing list.
You can still navigate around this archive, but know that no new mails
have been added to it since July of 2016.
Click here to be taken to the new web archives of this list; it includes all the mails that are in this frozen archive plus all new mails that have been sent to the list since it was migrated to the new archives.
Hi. I don't think that this is a support emailing list. That's
definitely better suited for users_at_open-mpi.org.
But, to answer your question, the fastest way I can think of is
something along the lines of
(see docs: http://www-unix.mcs.anl.gov/mpi/www/www3/MPI_Comm_split.html)
That won't change the existing rank inside of MPI_COMM_WORLD (I don't
think that's possible), but it will create a new comm object (myNewComm)
you can use instead (where each node has a new rank that you helped
specify (though if two nodes specified the same key (myNewRank) then
there will be some tie-breaking and you'll have to lookup the new rank
Sean Ross-Ross srossross-at-gmail.com |openmpi/Allow| wrote:
> Is there a way to reorder the mpi rank at runtime or otherwise?
> I have a program that makes files on each local node in somewhat of a
> random fashion. I want to these files to correspond to a part of a
> PETSc Distributed array used in another software package. My problem
> is that PETSc expects these windows of data to be on a certain
> ranking node. It seems that an easy solution may be to change the
> rank of the nodes to correspond with the PETc array.
> Thank you for your help
> docs mailing list
[A dream that comes true can't really be called a dream.]