[gmx-users] Gromacs scaling problem

[David van der Spoel]

M.N. Manoj wrote:
> Dear friends,
> 
> We have observed the following scaling problem on our cluster:
> 
> Cluster config:
> Nodes = 3, Processor = Xeon quad core 1.86 Ghz, 2 processors per node.
> Network= gigabit
> 
> We have used the "dppc" system from gromacs website to benchmark this 
> cluster.
> The following are the results, with gromacs 3.3.2  ( configured with 
> MPICH2 )
> 
> 1. With np 8, we get 1.29 ns/day ( all running on 1 node)
> 2. With np 12 we get 1.50 ns/day
> 3. With np 16 we get 1.433 ns/day
> 4. With np 24 we get 0.517 ns/day
> 
> We understand the scalability decreases with number of processors, but 
> why the ns/day comes down with higher job distribution?
try grompp -shuffle
version 4 will improve scaling dramatically.

> 
> Kindly advice whats going wrong...
> Thanks in advance...
> 
> M.N. Manoj
> CEO
> Jigsaw Bio Solutions,
> Bangalore - INDIA
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php


-- 
David van der Spoel, Ph.D.
Molec. Biophys. group, Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205. Fax: +4618511755.
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://folding.bmc.uu.se