NAMD 2.7 performance issues and Questions about Compiling

From: David Lee Braun (dlbraun_at_umich.edu)
Date: Wed Nov 03 2010 - 15:27:58 CDT

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

I am trying to address a performance issue with NAMD 2.7 compiled from
source.

The problem I am having is that NAMD under-preforms. Top will show ~30
user, 70% sys, and 0 io wait for most jobs. For some it runs at ~100%
user and does .15 days/ns ( 100000 atoms). Jobs that are using large
percentages of system run >15x slower.

It is possible that I did not compile charmm++ and NAMD correctly. Can
someone suggest the proper compilation instructions for the following
configuration:

OS openSuSE 11.1
Ethernet Gigabit
Compiler Intel 11.1
MPI OFED Intel openmpi-1.4.1

charmm++ was built with the following command:

env MPICXX=mpicxx ./build charm++ mpi-linux-x86_64 --no-build-shared
- --with-production

NAMD was build with the instructions from the notes.txt file included
with the source.

The cluster nodes are dual quad core nodes with one of the following
processors:

Intel(R) Xeon(R) CPU E5440 @ 2.83GHz
Intel(R) Xeon(R) CPU E5520 @ 2.27GHz

All nodes have either 2GB RAM or 3GB RAM per core. Jobs will run on
either the 2.83GHz or the 2.27GHz processors, but not a mixture.

Thanks

David
- --
David Lee Braun
Manager of Computational Facilities
for Dr Charles L. Brooks, III Ph. D.
930 N. University Ave
Chemistry 2006C
Ann Arbor, MI 48109-1055
dlbraun_at_umich.edu
(734) 615-1450
Skype: david.lee.braun
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.15 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/

iEYEARECAAYFAkzRxc4ACgkQjvKLiEz1ziI6dgCfZXgUrEGAnklSvHtrC/zt/xRH
AP4AoMLbfnuZKJW/0UUAEJe2S4HBBM5J
=VxM2
-----END PGP SIGNATURE-----

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:54:43 CST