From: Nicholas M Glykos (glykos_at_mbg.duth.gr)
Date: Fri Sep 25 2009 - 11:09:40 CDT
> I am running a simulation with 67005 atoms (lipid bilayer with water).
> Using openmpi, infiniband, and 86 CPU's, I've got the following performance.
> Is this reasonable?
In the absence of any information about the actual hardware it is
difficult to say whether you are doing well (or otherwise). The number of
CPUs you've mentioned is probably wrong (should have been 68 corresponding
to 17 quad core machines).
With 68 cores and infiniband you get 5.4 nsec per day. For a 60,000 atom
problem and using only 32 cores (8 quads) on gigabit we get² 3.2 nsec per
day. So, it would appear that there may be some room for improvement.
Having said that, the first thing to do is to study your parallel scaling:
start from 4 cores, then 8, 16, 32 and 64 (stay with powers of two for
starters). Don't forget to read the material on
http://www.ks.uiuc.edu/Research/namd/wiki/?NamdPerformanceTuning
Nicholas
² http://norma.mbg.duth.gr/index.php?id=about:benchmarks:namdv27b1
--
Dr Nicholas M. Glykos, Department of Molecular Biology
and Genetics, Democritus University of Thrace, University Campus,
Dragana, 68100 Alexandroupolis, Greece, Tel/Fax (office) +302551030620,
Ext.77620, Tel (lab) +302551030615, http://utopia.duth.gr/~glykos/
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:53:18 CST