I ran hello program which return the host name when I run it using mpirun -np 8 hello all the 8 answer returned from the same machine when I run it using mpirun -np 8 --host host1,host2,host3 hello I got answers from all the machines but it is not from all processors because I have 8 processors host1=4, host2=2, host3=2 the answer was 3 from host1, 3 from host2 and 2 from host3.
> From: email@example.com > Date: Wed, 28 Mar 2012 16:42:21 +0200 > To: firstname.lastname@example.org > Subject: Re: [OMPI users] Can not run a parallel job on all the nodes in the cluster > > Hi, > > Am 28.03.2012 um 16:30 schrieb Hameed Alzahrani: > > > Hi, > > > > I mean the node that I run mpirun command from. I use condor as a scheduler but I need to benchmark the cluster either from condor or directly from open MPI. > > I can't say anything regarding the Condor integration of Open MPI, but starting it directly by mpirun and supplying a valid number of ranks and hostfile should start some processes on other machines as requested. Can you run a plain mpihello first and output rank and hostname? Do you have ssh access to all the machines in questions? You have a shared home directory with the applications? > > -- Reuti > > > > when I ran mpirun from a machine and checking the memory status for the three machines that I have it appear that the memory usage increased just in the same machine. > > > > Regards, > > > > > From: email@example.com > > > Date: Wed, 28 Mar 2012 15:12:17 +0200 > > > To: firstname.lastname@example.org > > > Subject: Re: [OMPI users] Can not run a parallel job on all the nodes in the cluster > > > > > > Hi, > > > > > > Am 27.03.2012 um 23:46 schrieb Hameed Alzahrani: > > > > > > > When I run any parallel job I get the answer just from the submitting node > > > > > > what do you mean by submitting node: you use a queuing system - which one? > > > > > > -- Reuti > > > > > > > > > > even when I tried to benchmark the cluster using LINPACK but it look like the job just working on the submitting node is there a way to make openMPI send the job equally to all the nodes depending on the number of processor in the current mode even if I specify that the job should use 8 processor it look like openMPI use the submitting node 4 processors instead of using the other processors. I tried also --host but it does not work correctly in benchmarking the cluster so does any one use openMPI in benchmarking a cluster or does any one knows how to make openMPI divids the parallel job equally to every processor on the cluster. > > > > > > > > Regards, > > > > _______________________________________________ > > > > users mailing list > > > > email@example.com > > > > http://www.open-mpi.org/mailman/listinfo.cgi/users > > > > > > > > > _______________________________________________ > > > users mailing list > > > firstname.lastname@example.org > > > http://www.open-mpi.org/mailman/listinfo.cgi/users > > _______________________________________________ > > users mailing list > > email@example.com > > http://www.open-mpi.org/mailman/listinfo.cgi/users > > > _______________________________________________ > users mailing list > firstname.lastname@example.org > http://www.open-mpi.org/mailman/listinfo.cgi/users