Fixed (testing before pushing). Turns out to be one of the recent
additions that didn't account for how parallelization in minimization was
handled differently than parallelization in MD, so it was wired to hang for
every parallel minimization. (More details are in the log message).
The parallel test suite hung on the 5th test (and it would on any platform)
after "make parallel" was run, so running the test suite would be very
useful in catching errors like this.
All the best,
Jason
To Ross: This further demonstrates the necessity for cruise control (when
it does start running the tests) to have some mechanism for breaking out of
a test that appears to hang forever or it'll never really be automated.
On Thu, Dec 15, 2011 at 12:10 AM, Jason Swails <jason.swails.gmail.com>wrote:
> One of the recent changes to sander has broken parallel minimizations such
> that they simply hang when more than one thread is used per group. I think
> this has happened within the past day and a half or so.
>
> Any ideas? (I'd rather like to avoid having to hunt down errors on
> multiple threads using gdb if possible :)).
>
> All the best,
> Jason
>
> --
> Jason M. Swails
> Quantum Theory Project,
> University of Florida
> Ph.D. Candidate
> 352-392-4032
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Dec 15 2011 - 07:00:03 PST
