Re: mail about pbs script

From: John Hamre (johnhamre3_at_gmail.com)
Date: Thu Mar 28 2019 - 08:49:41 CDT

I am trying to run REMD on NAMD with 3 GPU's and 12 cores. I'm using
NAMD_2.11 lib/replica/example job.0conf. Here is the command I am using,

*mpirun ./charmrun ./namd2 -n 12 +replicas 12 job0.conf +stdout
output/%d/job0.%d.log *

I've tried different replica and process #'s

The error I'm getting is an mpi issue. This is the CUDA mpi. "*Open MPI
tried to fork a new process via the "execve" system call but*
*failed." *

If I use mpi with no CUDA I get the error message, "mpirun noticed that the
job aborted, but has no info as to the process that caused that situation"

Any suggestions? Thank you.

Regards,
John

On Thu, Mar 28, 2019 at 7:37 AM Vidhya Sankar <
vidhyasankar7928_at_remove_yahoo.com> wrote:

> Dear namd user could any one briefly explain the following commands in
> pbs script
>
> mpiexec.hydra -machinefile $PBS_NODEFILE -n 3 -perhost 2 namd2 +ppn 24
> +commap 0,12 +pemap 1-11,13-23
> /scf-data/vidhya/svsnamd/newcplipidinitial.conf > proteinlipid.log
>
> I want to use more number of processor and more number of MPI process on
> host
>
> my lines of pbs script as follows
>
> #!/bin/bash
> #PBS -S /bin/bash
> #PBS -N svs_MD
> #PBS -l select=3:ncpus=24
> #PBS -l walltime=240:00:00
> #PBS -q workq
> #PBS -o out.txt
> #PBS -e err.txt
> #PBS -joe
>
> # ## export environment variable from current session to job run-time ...
> better to use this always.
> #PBS -V
>
> #module load intelimpi
> #module load intel/icc
> module load intel/2019.1.144
> module load intel/icc/2019.1.144
> module load intel/idb/2019.1.144
> module load intel/ifort/2019.1.144
> module load intel/mkl/2019.1.144
> module load mpi/intelmpi/2019.1.144
> module load apps/NAMD/2.10/mic/intel
>
> 1,1 Top
>
>
> cd $PBS_O_WORKDIR
> which mpirun
> NPROCS=`wc -l < $PBS_NODEFILE`
>
> echo $NPROCS
> export OMP_NUM_THREADS=16
> export I_MPI_PRINT_VERSION=1
> mpiexec.hydra -machinefile $PBS_NODEFILE -n 3 -perhost 2 namd2 +ppn 24
> +commap 0,12 +pemap 1-11,13-23
> /scf-data/vidhya/svsnamd/newcplipidinitial.conf > proteinlipid.log
>
>
> what is the meaning of +commap and +pemap
>
>
>
> with regards
> S.Vidhysankar
>
>

This archive was generated by hypermail 2.1.6 : Mon Nov 18 2019 - 23:20:35 CST