[gmx-users] GROMACS API?

[Erik Lindahl]

Hi Ben,

We will share our ideas before deciding on the final form, but it  
will likely be another month or two until then - we want a more or  
less consistent core sketch first to avoid ending up in a huge  
discussion that leads nowhere - that's what happened when we and  
others tried to standardize force fields a couple of years ago :-)

The other issue is that the API initially will be quite limited and  
different from all current programs. We will isolate the internal  
state of the system completely and load/read coordinates with  
routines. The reason for this is that we want the API to support  
hardware acceleration such as running simulations on Graphics cards,  
FPGAs, and any other accelerator you can think of, and if the user  
was to read/write coordinates and forces on the main CPU every single  
step it would just kill the performance due to bandwidth issues (e.g.  
PCI-E to the GPU).

Anyway - my plan is to have a document describing our thoughts and an  
initial (slow) reference implementation done later this spring, and  
then we'll setup mailing lists and stuff at www.openmm.org.

Cheers,

Erik


On Feb 7, 2007, at 5:31 PM, Ben FrantzDale wrote:

> Erik,
> Interesting. I may want to get involved with that. I've worked on  
> extending the API for LAMMPS and much of my MS thesis will be about  
> appropriate API functionality for MD code. Is there an OpenMM  
> mailing list?
>
> —Ben
>
> On 2/7/07, Erik Lindahl <lindahl at cbr.su.se> wrote:
> Hi Ben,
>
> On Feb 6, 2007, at 11:57 PM, Ben FrantzDale wrote:
>
> > I am interested in using GROMACS as a library. I get the impression
> > from its use in Folding at Home that this is possible, but I don't see
> > any documentation on the subject. Is there a C/C++ API for GROMACS
> > so that I could initialize GROMACS with an MPI communicator then
> > send it jobs to process? If so, did I just not find the  
> documentation?
> >
>
> At the moment there isn't really an standardized (read: stable) API.
> It changes as functionality is added to the calls, which in turn
> aren't really well-documented either.
>
> For F at H we're simply using a normal Gromacs binary I ported to
> windows a couple of years ago, and then add a couple of data
> encryption routines. Thus, it's called on a script rather than API
> level.
>
> However, there is some good news: We recently started working on a
> standard "OpenMM" interface that will not only be well documented,
> but we will also make the entire specification public so that it can
> be implemented by anybody, much like OpenGL. And just like OpenGL we
> can hopefully convince people to use it by supporting hardware
> acceleration.
>
> Cheers,
>
> Erik
>
> _______________________________________________
> 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.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>
> _______________________________________________
> 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.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20070208/6165c7ef/attachment.html>