[gmx-users] Best step to run simulation over GPU?

Kutzner, Carsten ckutzne at gwdg.de
Mon Jul 27 12:41:52 CEST 2015


> On 27 Jul 2015, at 12:35, Rajiv <rajiv at kaist.ac.kr> wrote:
> 
> 
> p{margin:0;padding:0;}
> 
> 
> 
> 
> Yes i noticed them. Since i am not using MPI-enabled threats ( I use single PC contains 24 threads with 1 GPU and OpenMP installed), the -multi command doesn't work. 
> 
> 
> I wonder is there any other way without using MPI-enabled or should i have reconfigure with MPI? 
For running multi simulations you indeed need an MPI-enabled mdrun.

Carsten


> 
>> 
>> Thanks for the info. 
>> 
>> 
>> Can i put the command as below if i wanna run 3 simulation which has 24 processor with 1 GPU 
>> 
>> 
>> mdrun -deffnm first -multi 1 
>> 
>> 
>> mdrun -deffnm second -multi 2 
>> 
>> 
>> mdrun -deffnm third -multi 3 
>> 
>> 
>>> 
>> or i have to specify -multi 3 to all cases ? I am bit confused with this. thanks. 
> There is an example command line in the PDF file I pointed to. 
> The idea is to start all three simulations with one command line, 
> either using -multi 3 or -multidir dir1 dir2 dir3 
> 
> Carsten 
> 
>> 
>> 
>>> On 24 Jul 2015, at 08:48, RJ <rajiv at kaist.ac.kr> wrote: 
>>> 
>>> Dear gmx, 
>>> 
>>> 
>>> I have a single PC contains 24 threads with GTX 980Ti. 
>>> 
>>> 
>>> I would like to know how do i run the 3 or 2 simulation in same time with above mentioned PC would have similar speed. 
>> Simply try it out :) 
>> 
>> It could be beneficial to use the -multi options of gmx mdrun 
>> to start the multiple simulations. This way GROMACS would automatically 
>> take care of proper process placement and pinning. 
>> 
>> You could make a quick check with, e.g., mdrun -nsteps 5000 -resethway -multi ? 
>> to get a performance estimate for 2 and for 3 simulations. 
>> 
>> There is more information about this on page 7 of 
>> http://arxiv.org/src/1507.00898v1/anc/supporting.pdf 
>> 
>> Carsten 
> 
> -- 
> Gromacs Users mailing list
> 
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
> 
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> 
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-request at gromacs.org.

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



More information about the gromacs.org_gmx-users mailing list