Dear All,
I'll add to this that as recently presented at the ACS and SBDD 2017 conferences there is an officially supported GPU-TI implementation, authored by Dan Mermelstein, Charles Lin and myself coming in AMBER 18 that works identically to the current CPU implementation, includes full support for TI and MBAR and provides performance equivalent to that seen with regular GPU AMBER. Since it will be part of the official release it will also work with any GPU model and performance updates added to AMBER as new hardware is released and updates/bugfixes will be included as part of the regular AMBER update procedure. A paper describing this work is currently in review and beta versions of the code are being extensively tested on live projects by several major pharma companies and academic groups. Any of the existing workflow tools that support the CPU TI code in AMBER 16 will support this GPU implementation. There is also work in progress in collaboration with the Chemical Computing Group to provide GUI support for TI/FEP calculations with AMBER 18 through their MOE software.
Here's a taste of the performance it offers for a ~ 50K atom system (1xTitan-X GPU vs 2 x E5-2650V4 CPUs 20 cores).
If you are interested in being part of the beta test and have a current copy of AMBER 16+AmberTools 17 please contact me.
All the best
Ross
> On Sep 12, 2017, at 9:01 AM, Darrin York <york.chem.rutgers.edu> wrote:
>
> Hi all,
>
> We wanted to give everyone an update regarding our GPU-accelerated free
> energy methods in Amber, and in particular, our implementation of
> thermodynamic integration <http://theory.rutgers.edu/?q=node/288>
> reported very recently in JCTC
> (http://pubs.acs.org/doi/abs/10.1021/acs.jctc.7b00102). The code
> typically delivers over 2 orders of magnitude of speed-up relative to a
> single CPU core for the calculation of ligand–protein binding affinities
> with no statistically significant numerical differences and thus
> provides a powerful new tool for drug discovery applications.
>
> The code (pmemdGTI) is ready for beta release as a patch to Amber 16,
> and is available as through the Laboratory for Biomolecular Simulation
> Research <http://lbsr.rutgers.edu/>, a new core laboratory at Rutgers
> dedicated to the development and application of innovative biomolecular
> simulation tools, on the download page:
> http://lbsr.rutgers.edu/software-downloads
>
> Also available are new tools for free energy analysis such as the the
> vFEP method <http://theory.rutgers.edu/?q=node/106> for robust
> determination of free energy profiles.
>
> We hope you find it useful, and look forward to getting your feedback,
> discussion, and suggestions for further extension and next sets of features.
>
> cheers!
>
> Darrin
>
> --
> --
> ======================================================================
> Darrin M. York : Henry Rutgers University Professor
> : Director, Cyberlearning
> Center for Integrative Proteomics : Innovation & Research Center
> Research and Department of : Director, Laboratory for
> Chemistry & Chemical Biology : Biomolecular Simulation Research
> :
> Rutgers, the State University : Darrin.York.rutgers.edu
> of New Jersey : fax: +1-732-445-4320
> 174 Frelinghuysen Road : phone: +1-848-445-5199
> Piscataway, NJ 08854 USA : http://theory.rutgers.edu
> ======================================================================
>
> _______________________________________________
> 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 Tue Sep 12 2017 - 07:00:02 PDT
