On Apr 23, 2012, at 8:22 AM, David A Case <case.biomaps.rutgers.edu> wrote:
> On Sun, Apr 22, 2012, Jason Swails wrote:
>
>> My personal feeling is that we should
>> actually use the real8 constant everywhere, since selected_real_kind() is
>> more portable than assuming real(kind=8) is an 8-bit real.
>
> I'm against such a change. Amber routines often call libraries (e.g. blas,
> but many others as well) that expect double precision arguments. These would
> all start to fail if people start inventing their own real types. You have
> already pointed out that the "real8" variable that existed in chamber was
> being translated into different types with different compilers. So, in my
> view, your suggested change is actually less portable than other alternatives.
>
> I prefer the designation "double precision" for double precision variables,
> as is done in most of Amber. This is easily understood by both people and
> compilers, and has a 50-year history.
I agree, I was just suggesting one as the superior and more portable of the 2 used options.
I assumed there was a deliberate reason for avoiding double precision and 'rolling your own' real derived from its relationship with CHARMM. But I agree that your suggestion is safer and more portable barring a good reason to avoid it.
All the best,
Jason
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Apr 23 2012 - 06:00:03 PDT
