Brian, I had compiled 1.0.1 which seemed to be the latest stable version. But just to be sure, I downloaded the latest, rebuilt, and I'm getting the same problem. I ran strace on the run and mpirun is hanging on a futex() call. Any notion of what might cause this?
Thanks Mike > On Dec 30, 2005, at 4:01 PM, Michael L. Stokes wrote: > >>> On Dec 30, 2005, at 1:05 PM, Michael L. Stokes wrote: >>> >>>> I have compiled and installed OpenMPI on a SUSE based SGI PRISM (8 >>>> processor model). I configured the build with --with- >>>> device=ch_shmem, >>>> otherwise I took all the defaults. I installed as root using the >>>> default >>>> install prefix. SO good so far. Then I compile my hello world >>>> project >>>> using mpiCC, and again this looks good. I run ldd on the executable >>>> and I >>>> was convinced that is is using the OpenMPI libs and not the MPT >>>> libs. > > I noticed you are using a fairly old version of Open MPI (from before > we officially released v1.0.1). Could you try either v1.0.1 or a new > nightly tarball and see if you get the same results. I'm also a > little concerned about the output from mpirun -d -- can you double > check that you are using the mpirun from the right installation path > (and not one from an older version of Open MPI or LAM or MPICH or > something?). > > Thanks, > > Brian > > > -- > Brian Barrett > Open MPI developer > http://www.open-mpi.org/ > > > _______________________________________________ > users mailing list > us...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/users > --------------------------------------- Michael L. Stokes, Ph.D. Integrated Modeling & Test On-site Lead RTTC/WST&E Support Room 27 Building 4500 Redstone Arsenal, AL 35898 Ohio Supercomputer Center 1224 Kinnear Road Columbus, OH 43212-1163 256.876.5967 phone 256.876.7918 fax msto...@osc.edu
