Hi,<br><br>I'm trying to carry out an MD simulation on 4 processors but am having trouble with the MPI ("Can't read MPIRUN_HOST"). My error is:<br><br>-catch_rsh /opt/gridengine/default/spool/compute-0-32/active_jobs/176501.1/pe_hostfile<br>
compute-0-32<br>compute-0-32<br>compute-0-32<br>compute-0-32<br>Got 4 slots.<br>compute-0-32<br>compute-0-32<br>compute-0-32<br>compute-0-32<br>OSU MVAPICH VERSION 0.9.5-SingleRail<br>Can't read MPIRUN_HOST<br># Cleaning local host (compute-0-32) for lpeng (lpeng)... done.<br>
<br><br>My submission script is:<br><br># Define locations of MPIRUN, MDRUN, and grompp<br>MPIRUN=/usr/local/topspin/mpi/mpich/bin/mpirun<br>MDRUN=/share/apps/gromacs-3.3.1/bin/mdrun-mpi<br>grompp=/share/apps/gromacs-3.3.1/bin/grompp<br>
<br>cd /nas2/lpeng/nexil/gromacs/cg_setup<br><br># Running Gromacs: read TPR and write output to /nas2 disk<br>$MPIRUN -v -machinefile $TMPDIR/machines -np $NSLOTS $grompp -f md.mdp -c confout.gro -o md3.tpr -n index.ndx<br>
<br># Run MD<br>$MDRUN -v -nice -np $NSLOTS -s md3.tpr -o md3.trr -c confout.gro -g md3.log -x md3.xtc<br><br><br>Does anyone have any insight on what the problem could be? Is the problem with the MPI or the submission script?<br>
<br>Thanks,<br>Lili<br><span class="gI"></span><br>