[gmx-developers] gmxtest and GROMACS cvs again

[Carsten Kutzner]

Hi Rafael,

the CVS version is by it's very nature changing from day to day. I guess
that there is no use to update the test suite until the CVS version is
in a stable state.

The 'simple' neighboursearching type will work on a single processor.
For the tests with domain decomposition, try setting ns-type to 'grid'
in the mdp file.

The fourier grid in x- and y-dimension (nkx, nky in the tpr dump /
fourier_nx, fourier_ny in the mdp file) has to be a multiple of the
number of PME processors - which is equal to the total number of
processors you use for a simulation unless you specify separate PME
nodes with the -npme flag. So you could also set fourier_nx = fourier_ny
= 28 in the mpd file for the nacl test to be able to run on 2
processors. OR use the 3.3 grompp to produce the mdp files since it will
automatically set the fourier grid to fit the number of processors.

Carsten



Rafael R. Pappalardo wrote:
> Dear All,
> 
> I am still fighting with GROMACS CVS and Altix. I am now convinced that the 
> gmxtest version is not compatible with the current CVS version. If I run in 
> serial almost everything is OK (I will report later). But if I run in 
> parallel I obtain several complains about obsolete features and incompatible 
> options.
> 
> After solving the problem with the grompp flag
> ./gmxtest.pl -np 2 simple complex reports:
> 1 out of 16 simple tests FAILED
> 8 out of 14 complex tests FAILED
> 
> All kernel test finished without problem.
> 
> Fatal error:
> ns type Simple is not supported with domain decomposition
> 
> appears in several tests:
> simple/bham
> complex/water
> complex/urea
> complex/tip4p
> complex/sw
> complex/argon
> complex/aminoacids
> complex/acetonitrilRF
> 
> Fatal error:
> Could not find an appropriate numbers of separate PME nodes that is a multiple 
> of the fourier grid x (27) and y (27) components.
> 
> appears in complex/nacl
> 
> Changing -np 2 to -np 3 solve the problem with the nacl test.
> 
> 
> I am doing something wrong?
> 
> Sincerely,

-- 
Dr. Carsten Kutzner
Max Planck Institute for Biophysical Chemistry
Theoretical and Computational Biophysics Department
Am Fassberg 11
37077 Goettingen, Germany
Tel. +49-551-2012313, Fax: +49-551-2012302
http://www.mpibpc.mpg.de/research/dep/grubmueller/
http://www.gwdg.de/~ckutzne