[gmx-users] MPICH and scaling performance

[Anton Feenstra]

Hi,


I just finished installing gromacs here on our linux cluster 
(1GHz Athlon 1 CPU/node) running 2.4.6 kernel (RH 7.0) and using
MPICH 1.1.2 and 'standard' ethernet connections. 

With a protein/ligand/water system (4171/63/89043 atoms) I get 
terrible scaling, compared to the scaling benchmark on the web
(I get: 100% - 70% - 40% for 1-2-4 nodes). Of course, that is 
a vastly different type of system (both the hardware and the 
simulated system), but then again there might be some optimization 
possible in the MPI setup. Can anyone give me some pointers for 
that? (preferably specific for MPICH?)

-- 
Groetjes,

Anton
 ________ ___________________________________________________________
|        | Anton Feenstra                                            |
| .      | Dept. of Pharmacochemistry - Free University Amsterdam    |
| |----  | De Boelelaan 1083 - 1081 HV Amsterdam - The Netherlands   |
| |----  | Tel: +31 20 44 47608 - Fax: +31 20 44 47610               |
| ' __   | Feenstra at chem.vu.nl - http://www.chem.vu.nl/afdelingen/FAR|
|  /  \  |-----------------------------------------------------------|
| (    ) | Dept. of Biophysical Chemistry - University of Groningen  |
|  \__/  | Nijenborgh 4 - 9747 AG Groningen - The Netherlands        |
|   __   | Tel +31 50 363 4327 - Fax +31 50 363 4800                 |
|  /  \  | K.A.Feenstra at chem.rug.nl - http://md.chem.rug.nl/~anton   |
| (    ) |-----------------------------------------------------------|
|  \__/  | "May the Force Be With You" (Star Wars)                   |
|________|___________________________________________________________|