[gmx-users] how to perform analysis in faster with multicore/processor

[Justin Lemkul]

On 10/29/19 6:53 AM, Christian Blau wrote:
> Hello Nikhil,
>
>
> Currently there is no "in-built" way to do this. (we thought of this 
> before and there is some basic infrastructure for this in the 
> trajectory analysis framework, but in short - no)
>
> However, if you have many trajectories you can run analysis of all 
> trajectories in parallel to speed up the time to obtaining a result. 
> One possible way to automate this parallelisation would be to set up a 
> makefile for your analysis and then run with -j (numberOfProcessors), 
> see for a rough outline on how this would work 
> https://stackoverflow.com/questions/19822435/multiple-targets-from-one-recipe-and-parallel-execution/ 
> <https://stackoverflow.com/questions/19822435/multiple-targets-from-one-recipe-and-parallel-execution/19822767#19822767> 
>
>

To build onto this, if analyzing only one trajectory, just about every 
GROMACS analysis tool has -b and -e flags, so one can perform analysis 
on multiple time intervals of a given trajectory easily, e.g.

gmx program -f -s -b t0 -e t1 &
gmx program -f -s -b t1+10 -e t2 &
gmx program -f -s -b t2+10 -e t3 &
etc.

-Justin

>
> Best,
>
> Christian
>
>
> On 10/29/19 11:35 AM, Nikhil Maroli wrote:
>> Is there any way to speed up the gromacs analysis or any way to use
>> multiprocessor for gromacs trajectory analysis.? I have to analyze
>> trajectories of 10 ps intervals upto 1 microsecond. (I do not want to
>> reduce time interval for many reason)

-- 
==================================================

Justin A. Lemkul, Ph.D.
Assistant Professor
Office: 301 Fralin Hall
Lab: 303 Engel Hall

Virginia Tech Department of Biochemistry
340 West Campus Dr.
Blacksburg, VA 24061

jalemkul at vt.edu | (540) 231-3129
http://www.thelemkullab.com

==================================================