Re: [AMBER] Attempting to compile fresh version of Amber 11, bugfixes > 17 won't apply cleanly

From: Jason Swails <jason.swails.gmail.com>
Date: Wed, 2 Jul 2014 09:47:50 -0700

On Wed, Jul 2, 2014 at 8:22 AM, Bob Healey <healer.rpi.edu> wrote:

> Hello.
>
> One of the researchers I support has requested a fresh build of Amber 11
> with the GPU binaries enabled. (most recent version this site has a
> license for). Digging out my archived source tarball, and a copy of
> Amber Tools 1.5, I am attempting to apply all the bugfixes before
> beginning to build. The bugfixes for Amber Tools apply cleanly, but the
> big apply_bugfix script and bugfix.all.tar.gz will not. I've snipped
> the successes out of the apply bugfix output. Should I ignore all the
> patch errors, or do I have a bad set of sources/patches? I'm currently
> building in the /scratch space on the machine since there's a non-gpu
> enabled working copy in /usr/local/amber. AMBERHOME is properly set
> before applying any patches. I have tried it applying the tools fix
> before and after the main amber fixes. Doesn't change outcome.
>

​To save some time, I don't know that pmemd.cuda from Amber 11 would even
be usable with the current GPU hardware ​and the current nVidia CUDA
toolkit. I wouldn't even recommend trying. That code is no longer
supported and quite a few bugs have been fixed since support for Amber 11
ended.

If obtaining an Amber 14 license is not an option, ParmEd can be used to
run GPU-accelerated MD through OpenMM using the same input files (and
writing the same trajectory file formats) as pmemd.cuda. It is described
in the Amber 14 manual.

HTH,
Jason

P.S. I wrote a mini how-to for compiling Amber 11 with AmberTools 1.5
awhile back, but still after all of the bug fixes have been released so it
should still be valid.
http://jswails.wikidot.com/installing-amber11-and-ambertools-1-5

Also, the patching is not clean (there will be numerous failures) that
result from the rather clumsy marriage of AmberTools 1.5 and Amber 11
(AmberTools 1.4 was originally released with Amber 11, and 1.5 was not as
compatible as we had hoped it would be), so ignore patching errors.

-- 
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jul 02 2014 - 10:00:03 PDT
Custom Search