I tested and the number of ranks in world comm is correct. I couldn't find the bug that causes the program to produce erroneous answers when this scheme is used, though.
On Fri, Jul 29, 2016 at 3:38 PM, Saliya Ekanayake <esal...@gmail.com> wrote: > Thank you, that's good to know. > > Yes, testing this now. I'll let you know > > On Fri, Jul 29, 2016 at 3:33 PM, Ralph Castain <r...@open-mpi.org> wrote: > >> FWIW: I just tested it on master, and all the procs are in the same >> comm_world. However, that was with a C application, not Java. >> >> Saliya: can you please check to see if Java is behaving differently? A >> simple app that gets and prints out its rank is all that is required. >> >> >> On Jul 29, 2016, at 12:18 PM, Ralph Castain <r...@open-mpi.org> wrote: >> >> Actually, what Saliya describes sounds like a bug - those procs must all >> be assigned to the same comm_world. >> >> Saliya: are you sure they are not? What ranks are you seeing? >> >> >> On Jul 29, 2016, at 12:12 PM, Udayanga Wickramasinghe < >> uswic...@umail.iu.edu> wrote: >> >> Hi, >> I think orte/ompi-mca foward number of environment variables (ie:- $ >> OMPI_*) to its ranks. So I believe you may use >> $OMPI_COMM_WORLD_LOCAL_RANK to specifically filter out parameters within >> the script. >> >> Regards >> Udayanga Wickramasinghe >> Research Assistant >> School of Informatics and Computing | CREST >> Indiana University, Bloomington >> >> >> >> >> >> On Fri, Jul 29, 2016 at 2:50 PM, Saliya Ekanayake <esal...@gmail.com> >> wrote: >> >>> Hi, >>> >>> I am invoking my Java MPI program as, >>> >>> mpirun opts script.sh >>> >>> Now, I want to pass some rank specific argument to this script. For >>> example, I need to enable GC logs for one of the processes only, so I need >>> to pass an argument to the JVM if that's rank0 to do GC logs. >>> >>> I've asked the same question a while back and Ralph suggested to use >>> >>> mpirun opts -np 1 java args-for-rank0 app : -np 1 java args-for-others >>> .. style. >>> >>> It works for 1 node, but beyond that, there's something that goes wrong >>> with how processes are assigned to communicators. In the code, the comm >>> world is split into different communicators and this pattern breaks that >>> logic. >>> >>> Is there an alternative approach to doing this? >>> >>> Thank you, >>> Saliya >>> >>> -- >>> Saliya Ekanayake >>> Ph.D. Candidate | Research Assistant >>> School of Informatics and Computing | Digital Science Center >>> Indiana University, Bloomington >>> >>> >>> _______________________________________________ >>> users mailing list >>> users@lists.open-mpi.org >>> https://rfd.newmexicoconsortium.org/mailman/listinfo/users >>> >> >> _______________________________________________ >> users mailing list >> users@lists.open-mpi.org >> https://rfd.newmexicoconsortium.org/mailman/listinfo/users >> >> >> >> >> _______________________________________________ >> users mailing list >> users@lists.open-mpi.org >> https://rfd.newmexicoconsortium.org/mailman/listinfo/users >> > > > > -- > Saliya Ekanayake > Ph.D. Candidate | Research Assistant > School of Informatics and Computing | Digital Science Center > Indiana University, Bloomington > > -- Saliya Ekanayake Ph.D. Candidate | Research Assistant School of Informatics and Computing | Digital Science Center Indiana University, Bloomington
_______________________________________________ users mailing list users@lists.open-mpi.org https://rfd.newmexicoconsortium.org/mailman/listinfo/users
