[gmx-users] Running with > 5 NODES

Mehmet Suezen suzen at theochem.tu-muenchen.de
Fri Jan 4 11:53:19 CET 2002


Hi All,

FFTW has its own test  script. It's successful, at least in my case. But
even after
David`s suggestions, I do have still problems with running Gromacs with
MPICH.  
Does any body know, do I have to run in different manner for Linux SMP
cluster?
I've just try to test MPICH with given water example

$ more spc216.gro
$ more water.top
$ more water.mdp
$ grompp_d -np 2 -f water.mdp -p water.top -c spc216.gro -o water.tpr
$ gmxdump_d -s water.tpr | more
$ /users/suzen/mpich-1.2.2.1/bin/mpirun -np 2 -machinefile file_hosts
/home/suzen/MDcodes/gromacs/linux/bin/mdrun_d -s water.tpr -o water.trr
-c water_out.gro -v -g water.log

In case of single cpu, it works fine. Setting LD_LIBRARY_PATH doesn`t
help. After I execute last command
mpirun appeared very short time in process table and then droped without
producing any .log file.
It is puzzling like that running with > 5 node problem of choon peng.
Even I tried to recompile it several times carefully.

Many Thanks,

Mehmet




David van der Spoel wrote:
> 
> On Fri, 4 Jan 2002, Chng Choon Peng wrote:
> 
> >Hi Mehmet,
> >
> >   My MPICH implementation should be working fine as I tested it with
> >other MPI programs. All CPUs in each cluster node are reacheable to MPI
> >via rsh.
> >Also, I could obtain the various output files that David mentioned in
> >his latest reply.
> >Quite puzzling...
> Yes but how about FFTW? Isn't there a test program that comes with 
> Either way, scaling of the PME code is quite poor over more than a few
> nodes (2-3)
> 
> Groeten, David.
> ________________________________________________________________________
> Dr. David van der Spoel,        Biomedical center, Dept. of Biochemistry
> Husargatan 3, Box 576,          75123 Uppsala, Sweden
> phone:  46 18 471 4205          fax: 46 18 511 755
> spoel at xray.bmc.uu.se    spoel at gromacs.org   http://zorn.bmc.uu.se/~spoel
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.



More information about the gromacs.org_gmx-users mailing list