[OMPI users] Mumps Parallel version hanging with OpenMPI 1.8.1
Hi, I recently upgraded from openmpi 1.5.4 to openmpi 1.81 and built an application which uses parallel version of Mumps (http://mumps.enseeiht.fr/) . I am noticing process gets hung with np > 2 but some time it works also. I am not sure if this because of the openmpi upgrade or some problem with our code. It used to work fine with 1.5.4. Strace shows that process is polling some resource while it is hanged. Any pointers on how to debug this ? thanks, Kishor
Re: [OMPI users] Mumps Parallel version hanging with OpenMPI 1.8.1
My first suggestion would be to try using 1.8.8 instead to get all the bug fixes since 1.8.1 was released On Fri, Aug 7, 2015 at 10:34 PM, kishor sharma wrote: > Hi, > > I recently upgraded from openmpi 1.5.4 to openmpi 1.81 and built an > application which uses parallel version of Mumps ( > http://mumps.enseeiht.fr/) . > > I am noticing process gets hung with np > 2 but some time it works also. I > am not sure if this because of the openmpi upgrade or some problem with our > code. It used to work fine with 1.5.4. > > Strace shows that process is polling some resource while it is hanged. Any > pointers on how to debug this ? > > thanks, > Kishor > > ___ > 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/2015/08/27410.php >
Re: [OMPI users] Mumps Parallel version hanging with OpenMPI 1.8.1
if your application still hangs, you can use padb http://padb.pittman.org.uk to the state of the application. if it hangs in a MPI collective subroutine, you can try to mpirun --mca coll basic and see if the hang disappear. (the default tuned coll module is known to be broken in some cases) based on your report, we might recommend some tuning for the tuned module (as you can guess, the basic coll module is not optimized) Cheers, Gilles On Saturday, August 8, 2015, Ralph Castain wrote: > My first suggestion would be to try using 1.8.8 instead to get all the bug > fixes since 1.8.1 was released > > On Fri, Aug 7, 2015 at 10:34 PM, kishor sharma > wrote: > >> Hi, >> >> I recently upgraded from openmpi 1.5.4 to openmpi 1.81 and built an >> application which uses parallel version of Mumps ( >> http://mumps.enseeiht.fr/) . >> >> I am noticing process gets hung with np > 2 but some time it works also. >> I am not sure if this because of the openmpi upgrade or some problem with >> our code. It used to work fine with 1.5.4. >> >> Strace shows that process is polling some resource while it is hanged. >> Any pointers on how to debug this ? >> >> thanks, >> Kishor >> >> ___ >> 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/2015/08/27410.php >> > >
Re: [OMPI users] OpenMPI Fortran with Intel 15 and OpenMP
Hi Thomas, On Fri, Aug 07, 2015 at 06:39:10PM +0200, Thomas Jahns wrote: > Hello, > > On Aug 7, 2015, at 14:36 , James Spencer wrote: > >The Intel forum thread alleges that this is (at least for MVAPICH2) > >because incorrect Intel runtime sources are included in an MPI > >library, which is backed up by looking at the OpenMPI and Intel libraries: > > > this is already addressed, see [1]. It's meant to be fixed in 2015 Update 5, > but Georg gave some hints how to work around this with earlier compiler > versions by changing LDFLAGS/LIBS (although I guess one needs to decide on > whether to link libraries statically or dynamically). > [1] https://software.intel.com/en-us/forums/topic/549547 Thanks for the information; my own searches had not turned it up. Best wishes, --James -- James Spencer http://www.cmth.ph.ic.ac.uk/people/j.spencer/ http://www.github.com/jsspencer
