On Mar 3, 2014, at 1:48 PM, Beichuan Yan <beichuan....@colorado.edu> wrote:
> 1. After sysadmin installed libibverbs-devel package, I build Open MPI 1.7.4 > successfully with the command: > ./configure > --prefix=/work4/projects/openmpi/openmpi-1.7.4-gcc-compilers-4.7.3 > --with-tm=/opt/pbs/default --with-verbs=/hafs_x86_64/devel/usr > --with-verbs-libdir=/hafs_x86_64/devel/usr/lib64 > > 2. Then I rebuild and run my job in hybrid MPI/OPENMP mode: each compute node > only runs 1 process (this 1 process runs 16 OPENMP threads), it can get > initialized and run well each time with $TCP setting as follows, this is > great: > TCP="--mca btl_tcp_if_include 10.148.0.0/16" > mpirun $TCP -np 16 -hostfile $PBS_NODEFILE ./paraEllip3d input.txt If you're using the native verbs API, you don't need that TCP clause. Also, if you're running in a PBS job, you don't need the -hostfile clause. And if you're running one process per core in the allocated PBS job, you can skip the -np clause, too. You should be able to run with: mpirun ./paraEllip3d input.txt If you want one process per server, then mpirun -np <num_servers> --map-by node ./paraEliip3d input.txt > 3. Then I test pure-MPI mode: OPENMP is turned off, and each compute node > runs 16 processes (clearly shared-memory of MPI is used). Four combinations > of "TMPDIR" and "TCP" are tested: > case 1: > #export TMPDIR=/home/yanb/tmp > TCP="--mca btl_tcp_if_include 10.148.0.0/16" > mpirun $TCP -np 64 -npernode 16 -hostfile $PBS_NODEFILE ./paraEllip3d > input.txt > output: > Start Prologue v2.5 Mon Mar 3 15:47:16 EST 2014 > End Prologue v2.5 Mon Mar 3 15:47:16 EST 2014 > -bash: line 1: 448597 Terminated > /var/spool/PBS/mom_priv/jobs/602244.service12.SC > Start Epilogue v2.5 Mon Mar 3 15:50:51 EST 2014 > Statistics > cpupercent=0,cput=00:00:00,mem=7028kb,ncpus=128,vmem=495768kb,walltime=00:03:24 > End Epilogue v2.5 Mon Mar 3 15:50:52 EST 2014 It looks like you have two general cases: 1. The job fails for no apparent reason (like above), or 2. The job complains that your TMPDIR is on a shared filesystem Right? I think the real issue, then, is to figure out why your jobs are failing with no output. Is there anything in the stderr output? -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/