Hi David,
Thanks a lot for your answer.
mpirun -np 84 sander ....
Runs perfectly OK (although from what you say seems not very effective) produces output etc. I just thought it runs very slowly, is 1/2 ns x day a a good or a bad pace?
On the other hand tests I have done with sander.MPI job looks like it is running, but it does not produce any output, so I tried your test suggestion and .... it worked.
So what is the right way to do it?
I need to fill the following information in my script:
#PBS -l select=4:ncpus=12:mpiprocs=12
Where select is the number of nodes, ncpus the number of CPU and mpiprocs the number of processes...
And then
mpirun -np 48 sander.MPI ....
Where X = nodes * ncpus = 48 in the example above.
Can you suggest the ideal use?
Thanks!!!
Fabian
_______________________________
Fabian Glaser, PhD
Bioinformatics Knowledge Unit,
The Lorry I. Lokey Interdisciplinary
Center for Life Sciences and Engineering
Technion - Israel Institute of Technology
Haifa 32000, ISRAEL
fglaser.technion.ac.il
Tel: +972 4 8293701
Fax: +972 4 8225153
On Nov 28, 2012, at 5:29 PM, David A Case wrote:
> On Wed, Nov 28, 2012, Fabian Glaser wrote:
>>
>> I am successfully but very slowly running MD on 84 processors, about a
>> rate of 1/2 ns a day, with sander using the following command:
>>
>> mpirun -np 84 sander -O -i prod.in -p protein.prmtop -c
>> protein_equil.rst -o protein_prod.out -x protein_prod.mdcrd -r
>> protein_prod.rst
>
> This is way wrong: you are running 84 copies of the same (single-cpu) sander
> program. All the outputs are writing over each other.
>
>>
>> When I try the same with sander.MPI, nothing seems to happen, I mean the
>> job runs, but no output files are ever written, I guess I am doing some
>> trivial mistake....
>
> Do the parallel test cases run? Have you tried a small example with your
> inputs (e.g. -np 2, set nstlim to 10 and ntpr=1)?
>
> Even when you get everything running, sander.MPI is unlikely to scale well to
> 84 threads. You need to run pmemd.MPI to get good parallel scaling (unless
> this is a GB calculation; then sander.MPI in principle can scale well).
>
> ...dac
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 28 2012 - 23:00:02 PST
