[gmx-users] GPU CUDA version does not support improper dihedrals?

[Erik Lindahl]

Hi,

We discussed this a bit off-list, and just figured it would be good to include the conclusions here too so people find it if they search the archives.

First, we definitely support all improper torsions of periodic form, such as those in amber, charmm, and opls in OpenMM too, while Gromos uses harmonic torsions.

The torsions themselves would actually be straightforward to implement with the customforce classes in OpenMM (slightly slower, but that doesn't matter for torsions). However, this is presently only needed for Gromos force fields, and those won't work with the GPU code anyway since they rely on a full N^2 (where N=natomtypes) matrix for nonbonded parameters that would hurt performance quite a bit on GPUs. 

This doesn't mean we've ruled out Gromos for future GPU ports (on the contrary, we want to and plan to support it eventually), but it's not priority 1A right now.

Cheers,

Erik




> Read and respond to this message at: 
> https://simtk.org/forum/message.php?msg_id=4741
> By: Peter Eastman
> 
> I discussed this with the Gromacs developers to get the full story.  It appears
> that they currently do not support the Gromos force field on GPUs, and it will
> take a significant amount of work to do so.  Improper torsions would actually
> be easy to handle using a CustomTorsionForce.  The real challenge is nonbonded
> interactions, since Gromos doesn't use consistent combining rules.
> 
> Hopefully that will happen in a later release.
> 
> Peter
On Aug 10, 2010, at 5:50 PM, chris.neale at utoronto.ca wrote:

> This surprised me since most of the systems that I am interested in
> have improper dihedrals. Is it possible that I did something wrong here?
> 
> ### Error message from mdrun-gpu:
> 
> Program mdrun-gpu, VERSION 4.0.99-dev-20100315
> Source code file:
> /project/pomes/cneale/GPC/exe/intel/GPUgromacs/gromacs-4.5-beta2/src/kernel/openmm_wrapper.cpp, line:
> 610
> 
> Fatal error:
> OpenMM does not support (some) of the provided interaction type(s)
> (Improper Dih.)
> For more information and tips for troubleshooting, please check the GROMACS
> website at http://www.gromacs.org/Documentation/Errors
> 
> <I'm posting again the same topic as http://lists.gromacs.org/pipermail/gmx-users/2010-August/053018.html in case my first post was overlooked in the batch that came through when the mailing list was fixed>
> 
> 
> -- 
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use thewww interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> 

----------------------------------------------------------
Erik Lindahl <lindahl at cbr.su.se>
Professor, Computational Structural Biology
Center for Biomembrane Research & Swedish e-Science Research Center
Department of Biochemistry & Biophysics, Stockholm University
Tel: +468164675 Cell: +46703844534

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20100813/fb0dc2e3/attachment.html>