On Sun, Aug 7, 2016 at 9:50 AM, Thomas Evangelidis <tevang3.gmail.com>
wrote:
> Thanks Jason! 'xargs' looks very practical to run multiple jobs
> concurrently on a PC or laptop. However, I would rather use 'scoop' from
> Python to run the jobs on a multi-node cluster since the database may
> contain several thousands or millions of compounds.
> Just wanted to know if
> something similar has been made before for antechamber in order not to
> re-invent the wheel :)
>
>
Since I frequently use massive cpus for energy minimization of thousands of
structures (and a bunch of batch jobs that can be easily run in parallel),
I just made a super simple package for those who just want to copy, paste
and edit a bit: https://github.com/Amber-MD/map_mpi#examples
(require numpy and mpi4py: those two packages are already installed during
amber installation).
Hai
> Thomas
>
>
>
>
> > Another question: is anyone aware of any effort made to parallelize
> > > antechamber computations? I want to screen a database using bcc
> charges.
> > >
> >
> > Not parallelize within antechamber. But there are ways of doing this at
> > the shell level. If you are doing some kind of database for this,
> "xargs"
> > is actually a great way to parallelize. You can tell it how many
> > processors to use, and it will run through a list of arguments and
> dispatch
> > it to the target program. Once one job finishes, it will pull the next
> one
> > from the queue and dispatch it to the program (making sure never to have
> > more than the requested number of jobs active at a time).
> >
> > The man page and googling for examples will show you how to use it. I've
> > personally used it in AF-NMR to process each snapshot in a multi-model
> PDB
> > file in parallel. You can see this here:
> > http://casegroup.rutgers.edu/shifts-5.1.tar.gz (look for
> 'mulitafnmr.sh').
> >
> > You will probably also have to wrap antechamber in a shell script to make
> > sure that each antechamber run runs in a separate directory to avoid
> > collisions of intermediate files.
> >
> > There is another way to do it through the shell, but it's not as
> > efficient. You can do something like this:
> >
> > cd dir1
> > antechamber <args> &
> > cd ../dir2
> > antechamber <args> &
> > cd ../dir3
> > antechamber <args> &
> > cd ../dir4
> > antechamber <args> &
> > cd ..
> >
> > wait
> >
> > cd dir5
> > antechamber <args> &
> > ...
> >
> >
> > This has the effect that 4 jobs will run at a time, but the next group
> of 4
> > will only proceed when each of the previous group finishes (rather than
> > xargs, where it uses a queue-like approach to make sure processors are
> > rarely idle).
> >
> >
> > HTH,
> > Jason
> >
> > --
> > Jason M. Swails
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
>
>
> --
>
> ======================================================================
>
> Thomas Evangelidis
>
> Research Specialist
> CEITEC - Central European Institute of Technology
> Masaryk University
> Kamenice 5/A35/1S081,
> 62500 Brno, Czech Republic
>
> email: tevang.pharm.uoa.gr
>
> tevang3.gmail.com
>
>
> website: https://sites.google.com/site/thomasevangelidishomepage/
> _______________________________________________
> 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 Mon Aug 08 2016 - 12:30:03 PDT
