From: Thomas Evangelidis (tevang3_at_gmail.com)
Date: Fri Dec 03 2010 - 05:08:28 CST
Dear NAMD users,
I'm operating a ROCKS cluster with 1 master node (curie) and 2 slave nodes
(compute-0-0 and compute-0-1). All three nodes possess 24 CPUs. Connection
to compute-0-0 and compute-0-1 is possible via passwordless ssh from curie
master node. I've tried several combinations of qualifiers in the .nodelist
file but haven't managed to run NAMD on more than one node concurrently. It
is worth noting that NAMD can run on every node separately, and that I have
managed to run successfully GROMACS on 2 nodes concurrently.
My .nodelist file looks like this:
group curie
host compute-0-0 ++shell ssh ++cpus 24
host compute-0-1 ++shell ssh ++cpus 24
To run NAMD I connect to curie and issue the following command line:
/usr/local/bin/charmrun +p 48 ++nodelist /home/thomas/.nodelist ++nodegroup
curie ++mpiexec ++remote-shell /opt/openmpi/bin/mpirun
/export/home/zoe/namd/NAMD_2.7_Linux-x86_64-TCP/namd2 apoa1.namd >
apoa1.48proccessors.log
What am I doing wrong? I would greatly appreciate any advice.
thanks,
Tom
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:56:26 CST