[gmx-developers] libgromacs vs libgromacs_core?
szilard.pall at cbr.su.se
Sat Sep 28 20:44:45 CEST 2013
On Sat, Sep 28, 2013 at 8:06 PM, Mark Abraham <mark.j.abraham at gmail.com> wrote:
> On Sat, Sep 28, 2013 at 7:48 PM, Szilárd Páll <szilard.pall at cbr.su.se> wrote:
>>> This is an important point, but doesn't Jenkins help by compiling with older
>>> compilers as well as state of the art?
>> Not really - at least not in the current setup. We do not and probably
>> will not include in our builds compilers like pgi, pathscale, xlc,
>> Fujitsu C compiler, and other embedded compilers. It's not entirely
>> trivial to predict what the future is e.g. if Tilera come up with a
>> promising HPC platform or the IBM + NVIDIA (ref: Open Power) marriage
>> results in a new fancy machine (which are not out of the question, not
>> even in the 12 months time-frame), it may take quite some effort to
>> strip away the portability bottlenecks just to compile mdrun on such a
>> machine, let alone start optimizing for it.
> I think our portability and language requirements are so conservative
> that discussing the hypothetical scenario of putative non-portable
> code in GROMACS coping badly with unknown non-standard-compliant
> hardware/compiler/libraries is not going to help us at this time. The
> general problem is unsolvable. We're good at solvable problems when we
> know we have them.
We are drifting off-topic, but hey...
I believe, it is not a hypothetical scenario. Compared to the amount
of C in HPC expertise in the current developer team (AFAIK) only a
small a fraction of such C++ in HPC knowledge exists. Hence, stating
that all concerns raised are hypothetical is a bit like sticking our
hands in the sand - exactly because IMHO the lack of concreteness
stems from the lack of experience, and because "C++" and "easy" are
not on the same page, perhaps not even in the same chapter.
Let me close with a Stroustrup quote which probably most people know:
"C makes it easy to shoot yourself in the foot; C++ makes it harder,
but when you do it blows your whole leg off."
> gmx-developers mailing list
> gmx-developers at gromacs.org
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-developers-request at gromacs.org.
More information about the gromacs.org_gmx-developers