Re: [AMBER-Developers] reference os+compiler systems

From: Daniel Roe <daniel.r.roe.gmail.com>
Date: Thu, 10 Mar 2016 10:20:15 -0700

On Thu, Mar 10, 2016 at 10:03 AM, David A Case <david.case.rutgers.edu> wrote:
> Mostly, the OS should not matter. I hope that any gnu compiler after 4.4
> should be fine...we'll know more soon. My experience with Intel compilers
> is that they have been much more variable (and buggy) from version to version.
> It would be of interest to hear (from Dan, others?) about what version of
> compilers are available at national sites.

XSEDE Stampede:
intel/13.0.2.146 intel/14.0.1.106 intel/15.0.2 (overall default)
gcc/4.4.6 gcc/4.6.3 gcc/4.7.1 gcc/4.9.1(default)

XSEDE Comet:
intel/2013_sp1.2.144(overall default) intel/2015.2.164
gnu/4.9.2
pgi/15.7

NCSA Blue waters: (cray compilers are default here; that's a whole
'nother can of worms...)
intel/15.0.090 intel/16.0.0.109 intel/15.0.3.187(default) intel/16.2.2.181
gcc/4.5.3 gcc/4.7.1 gcc/4.8.1 gcc/4.9.2 gcc/5.3.0 gcc/4.6.1
gcc/4.7.2 gcc/4.8.2 gcc/4.9.3(default) gcc/4.6.2 gcc/4.7.3
gcc/4.9.0 gcc/5.1.0 gcc/4.6.3 gcc/4.8.0 gcc/4.9.1 gcc/5.2.0
pgi/11.10.0 pgi/13.9.0 pgi/14.4.0 pgi/15.7.0 pgi/13.10.0
pgi/14.10.0 pgi/14.7.0 pgi/13.3.0 pgi/14.2.0 pgi/14.9.0
pgi/13.6.0 pgi/14.3.0 pgi/15.10.0(default)

CHPC.Utah:
intel/2015.1.133 intel/2016.0.109 (overall default)
gcc/4.4.7(default) gcc/4.7.2 gcc/4.8.4 gcc/4.9.2
pgi/12.8 pgi/13.4 pgi/14.7 pgi/14.10 pgi/15.1
pgi/15.7(default) pgi/16.1

Note that although many compiler versions are listed at some sites,
they are not all fully supported in terms of what other libraries you
can load with them (e.g. netcdf, mpich, etc).

-Dan

>
> I would not worry obout cygwin or dragonegg. What works for those works,
> and some kind soul might want to try to squash bugs, but that's not required.
>
> (There *is* a virtue in trying lots of compilers, since it can expose code
> bugs that are somehow hidden otherwise.)
>
>>
>> I don't have experience with gcc5, but on my linux cluster, Intel
>> compiler gives faster results for MD, especially when the MKL is used.
>> I'm talking here of course about things that are not cuda-enabled (QM
>> and QM/MM for example).
>
> Could you say how much faster (for some example)? (Aside: you can use MKL
> + gnu compilers, if that makes any difference).
>
> ...thx...dac
>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers



-- 
-------------------------
Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 307
Salt Lake City, UT 84112-5820
http://home.chpc.utah.edu/~cheatham/
(801) 587-9652
(801) 585-6208 (Fax)
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Mar 10 2016 - 09:30:05 PST
Custom Search