FYI a list of actions/analyses in cpptraj that benefit from OpenMP
parallelization is given in section 28.2.6 of the Amber 14 manual.
-Dan
On Wed, Apr 23, 2014 at 8:29 PM, Jason Swails <jason.swails.gmail.com> wrote:
> On Wed, Apr 23, 2014 at 7:21 PM, Eugene Yedvabny <eyedvabny.berkeley.edu>wrote:
>
>> Hello Amber community,
>>
>> I am compiling Amber14 with gcc/gfortran and want to build the OpenMP
>> version of cpptraj. It seems that GNU OpenMP and MKL libs are not mutually
>> compatible, so I was wondering if cpptraj used MKL
>
>
> Yes. It uses MKL routines in the rotdif (rotational diffusion), and
> matrix diagonalization routines. But as far as I can tell, that's it.
> (Basically I tried compiling WITHOUT linking the MKL libraries and just
> took note of the unresolved symbols, which just came from action_rotdif and
> dataset_mods).
>
> and if compiling an
>> OpenMP version would have an impact on the speed of non-threaded parts of
>> cpptraj?
>>
>
> It might for the rotdif and diagonalization routines if the MKL speeds
> those calculations up (on my AMD-based machine, MKL actually runs _slower_
> than several standard ifort-optimized blas/lapack/arpack routines). Other
> than that, there won't be any impact on using OpenMP parallelization.
> (Where threading is not supported, it's not used and simply defaults to
> the serial code).
>
> HTH,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
--
-------------------------
Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 201
Salt Lake City, UT 84112-5820
http://home.chpc.utah.edu/~cheatham/
(801) 587-9652
(801) 585-6208 (Fax)
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Apr 24 2014 - 07:30:04 PDT
