> On Dec 18, 2014, at 7:41 AM, David A Case <case.biomaps.rutgers.edu> wrote:
>
> I'm hoping someone can look into this: for me, all the pmemdTI tests are
> failing. The jobs run fine, but the final breakdown of DV/DL into pieces
> (section headed "DV/DL, AVERAGES OVER n STEPS") gives all zeroes.
> This on the master branch, and a Mac with gcc 4.9. If it works for you on
> some other platform, please post info, since that would be a help. Or, if
> there is someone more familiar with the code than I am, that would also be
> great.
Works for me on OS X 10.10 with GCC 4.8. I *think* I recall seeing this before quite recently when I tried gcc 4.9 (I can’t remember if that was on my Linux box or on my Mac... I’ll look into it).
> In general, we need to start cleaning up the test cases, in preparation for
> the upcoming developers' meeting. There are also test failures (at least for
> me) in the ti_decomp sander tests and the softcore/dnylmb tests.
The TI failures I *think* come from the changes that Kevin Hauser sent to me just before the Amber 14 release. Those changes were just fixing the backbone atom definitions for nucleic acid residues, if I recall correctly, so those failures have been around for a long time. I’m copying him here in case he has time to look into this and make sure that it’s OK to update the test case output. I was nervous doing it since the differences looked larger than I naively expected at first.
I also see a number of other failures when I use the clang/gfortran compilers... most of which seem to be slightly larger roundoff errors than other compilers display. When should we open a new spot on the Wiki to report test results?
All the best,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Dec 18 2014 - 05:30:06 PST
