If you don’t specify a slot count, then we will automatically detect the #cores on the machine and set the slot count to that number.
You can add —map-by node to your cmd line and that will give you the prior behavior as we’ll round-robin one proc on each node at a time > On Aug 25, 2015, at 6:54 AM, Nicolas Niclausse <nicolas.niclau...@inria.fr> > wrote: > > > Hello, > > I'm trying to use openmpi 1.8.8 on a cluster managed by OAR, however i have > some troubles with the default slots number. > > I have reserved one core on two nodes (each has 12 cores): > > # cat $OAR_NODEFILE > nef097.inria.fr > nef098.inria.fr > > but: > mpirun -np 2 --mca plm_rsh_agent oarsh -hostfile $OAR_NODEFILE ./NPmpi > > runs only on the first node: > 0: nef097 > 1: nef097 > Now starting the main loop > 0: 1 bytes 7 times --> 0.00 Mbps in 12571.35 usec > [skip] > > > If i use a nodefile like this, it works: > nef097.inria.fr slots=1 > nef098.inria.fr slots=1 > > The doc says however that the default value is 1, and openmpi 1.6.4 works > fine (the OS is Centos 7, btw) > > Am i missing something ? > > -- > Nicolas NICLAUSSE Service DREAM > INRIA Sophia Antipolis http://www-sop.inria.fr/ > 2004 route des lucioles - BP 93 Tel: (33/0) 4 92 38 76 93 > 06902 SOPHIA-ANTIPOLIS cedex (France) Fax: (33/0) 4 92 38 76 02 > _______________________________________________ > users mailing list > us...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/users > Link to this post: > http://www.open-mpi.org/community/lists/users/2015/08/27487.php
