Dave -
Per my earlier mail, I am sold on bintraj as a useful feature, especially
for high scaling at supercomputer centers. At these locations, it is
often
true that a netCDF module is already available, built by the system s/w
support people. For pmemd, I think we may get 5-10% speedup from using
bintraj at high processor count rather routinely for realistic runs that
are
dumping coords every 0.25 psec or so. Anyway, I originally implemented
the
pmemd bintraj config option default as nobintraj, and was intending to
leave
it that way based on a desire to make configuration simpler for the
novices
(heck, I am not sure how to config this stuff without looking at it
hard...). So I concur with the decision in sander, but at the same time
have a lot of enthusiasm for this stuff (great going, John M. and whoever
else was involved). The bintraj capabilities in pmemd will be available
to
the amber dev team at large when I do a drop at the end of the week.
Best Regards - Bob
----- Original Message -----
From: "David A. Case" <case.scripps.edu>
To: <amber-developers.scripps.edu>
Sent: Tuesday, February 07, 2006 1:53 PM
Subject: amber-developers: reversing course on the bintraj option
> Hi everyone:
>
> Given the large numbers of problems people are having with bintraj
> compilation
> on some machines (see the wiki page), and the fact that it is quite
> difficult
> to find even what the problems are (much less the solutions) [again, in
> some
> cases], I have un-done the recent change to "configure", and made
> bintraj=no
> be the default. [There is no change in any code--just in what default
is
> set in configure.]
>
> I can just forsee many difficult questions about this from people who
> would actually never even need/use that feature, but for whom a failure
in
> configuration or compilation is a bad first introduction to our codes.
> In contrast, users with enough experience to know that they could use
this
> option can easily reconfigure with the "-bintraj" flag.
>
> Of course, there are tradeoffs both ways, but I would rather err on
> helping
> new users vs. making life a little easier for experienced ones. And
even
> experienced amber developers are becoming frustrated when netCDF fails
to
> compile....
>
> Since only a handful of lines in configure are involved, we can re-visit
> this
> issue before release, but only if we can get a better handle on
> things...perhaps a "stripped-down" version of the installation that only
> creates just what is needed for sander, and not a bunch of other stuff
as
> well?
>
> ....dac
>
> --
>
> ==================================================================
> David A. Case | e-mail: case.scripps.edu
> Dept. of Molecular Biology, TPC15 | fax: +1-858-784-8896
> The Scripps Research Institute | phone: +1-858-784-9768
> 10550 N. Torrey Pines Rd. | home page:
> La Jolla CA 92037 USA | http://www.scripps.edu/case
> ==================================================================
>
Received on Wed Apr 05 2006 - 23:49:43 PDT
