[gmx-users] Performance

[Anton Feenstra]

Ran Friedman wrote:
> Hi,
> 
> Rough data:
> 
> 1ns simulation of ~22,000 atoms on 8cpu (R12000, 400MHZ) with the gromacs FF, with PME, time step 2fs, takes me 44-48 hours.

a 22k atom system on 8 CPU's may scale badly. Did you happen to test scaling for
your system, by comparing runs on 1, 2, 4 and 8 CPU's? I'd guess that 8 CPU's
may be only 20-30% faster than 4 CPU's (but I'd like to know if I'm wrong!).

> ehab at biof.ufrj.br wrote:
> 
>>Please, somebody can tell me how is the time performance of GROMACS in a
>>Origin 2000? I just need to know the system size, the time of the simulation
>>(like 1ns) and the real time it took to run (like 3 days) on this machine.

There are benchmarks on www.gromacs.org, surely there is also (something
like) an Origin tested, or at least an r12k or r14k machine.


-- 
Groetjes,

Anton
  _____________ _______________________________________________________
|             |                                                       |
|  _   _  ___,| K. Anton Feenstra                                     |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|(   |   )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands   |
| \_/ \_/ | | | Tel: +31 20 44 47608 - Fax: +31 20 44 47610           |
|             | Feenstra at chem.vu.nl - www.chem.vu.nl/~feenstra/       |
|             | "If You See Me Getting High, Knock Me Down"           |
|             | (Red Hot Chili Peppers)                               |
|_____________|_______________________________________________________|