As a historical note, at one point on Blue Waters the PGI compilers
were the only ones that were able to build working versions of sander
and pmemd (CPU). This was due to a bug in pubfft that has since been
fixed, but it illustrates a potential benefit of supporting multiple
compilers.
-Dan
On Mon, Apr 3, 2017 at 3:44 PM, Gerald Monard
<Gerald.Monard.univ-lorraine.fr> wrote:
> Hello,
>
> About PGI compilers, what is nice is that they can be used for free now
> (community edition). For those who can't/won't pay for the Intel compilers,
> it could be a nice alternative for a "commercial" compiler. It includes also
> some versions of cuda and OpenACC which could be handy.
>
> However, I've just quickly try to compile AmberTools17 on two different
> machines: my laptop (pgi 2016 community edition) and at a HPC center (pgi
> 16). On both, it fails to compile for different reasons (see attachements),
> but I'd be interested to test the speed of pmemd and cpptraj :-).
>
> Gerald.
>
>
> On 04/03/2017 05:26 PM, David Case wrote:
>>
>> Hi everyone:
>>
>> 1. The second release candidate for AmberTools17 is here:
>>
>> http://ambermd.org/downloads/AmberTools17.02apr17.tar.bz2
>>
>> This *should* be the same as what is in the amber17-with-patches branch
>> at git.ambermd.org. But use the tarball for testing, since that is
>> what users will actually download.
>>
>> 2. Once we are ready to release, the "update_amber --upgrade" script will
>> be functional, but this is not ready yet for testing.
>>
>> 3. Please post testing results here, indicating that you are using the
>> "02apr17" tarball:
>>
>> http://ambermd.org/pmwiki/pmwiki.php/Main/AmberTools17Test
>>
>> It would be helpful for someone to try Intel compilers without MKL
>> (neither setting MKL_HOME nor using the -mkl flag.)
>>
>> Not sure what we should do with all the PGI errors. Two questions:
>> a. do the pmemd tests pass with PGI compilers?
>> b. what are the arguments in favor of continuing to support PGI?
>> e.g.: some machines have only those compilers(?)
>> better performance for cpptraj and/or pmemd(?)
>> something else?
>>
>> ...thx...dac
>>
>> _______________________________________________
>> AMBER-Developers mailing list
>> AMBER-Developers.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber-developers
>>
>
> --
> ____________________________________________________________________________
>
> Prof. Gerald MONARD
> Directeur du mésocentre EXPLOR
> Université de Lorraine
> Boulevard des Aiguillettes B.P. 70239
> F-54506 Vandoeuvre-les-Nancy, FRANCE
>
> e-mail : Gerald.Monard.univ-lorraine.fr
> phone : +33 (0)372.745.279
> web : http://www.monard.info
>
> ____________________________________________________________________________
>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
--
-------------------------
Daniel R. Roe
Laboratory of Computational Biology
National Institutes of Health, NHLBI
5635 Fishers Ln, Rm T900
Rockville MD, 20852
https://www.lobos.nih.gov/lcb
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Apr 04 2017 - 05:30:06 PDT
