On 02/13/10, Jess Bromley <[email protected]> wrote:
Yes. The virial is computed from a very large summation over floats and is extremely sensitive to architecture and compilation conditions. Thus, erratic failures.> I haven't actually run 4.0.7 against git gmxtest, but I made the last commit
> to the latter, and it was routine for correctly-functioning 4.0.x versions
> to fail the tests you mention. I think your installation is fine.
Thanks Mark, that's very encouraging!
Just a quick followup: when I run the tests in parallel mode I'm also
seeing checkvir related failures in complex/field (~250-275 errors)
and complex/tip4flex (1 or 2 errors), although strangely the latter
PASSES for 4 CPUs and fails for other numbers. Is this kind of
behaviour also expected?
Yes, that's normal. A suitably small case for single-cpu tests need not be able to undergo arbitrary DD.
PS I get fails in several other tests but they just seem to be due to
an inability to do domain decomposition on those tests.
Hmm, not sure about that, but pdb2gmx isn't parallel, so forget about it.
PPS Strangely in parallel mode all the pdb2gmx tests pass, unlike in
single-CPU mode.
All these kinds of issues are why the wiki no longer encourages the use of the regression tests..
Mark
-- gmx-users mailing list [email protected] http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [email protected]. Can't post? Read http://www.gromacs.org/mailing_lists/users.php
