<div style="font-family: 'Times New Roman'; font-size: 16px;"><br /><br /><span>On 02/13/10, <b class="name">Jess Bromley </b> <jess.bromley@googlemail.com> wrote:</span><blockquote cite="mid:69649d441002121001tcddce6ak4a471ab60d427102@mail.gmail.com" class="iwcQuote" style="border-left: 1px solid rgb(0, 0, 255); padding-left: 13px; margin-left: 0pt;" type="cite"><div class="mimepart text plain">> I haven't actually run 4.0.7 against git gmxtest, but I made the last commit<br />> to the latter, and it was routine for correctly-functioning 4.0.x versions<br />> to fail the tests you mention. I think your installation is fine.<br /><br />Thanks Mark, that's very encouraging!<br /><br />Just a quick followup: when I run the tests in parallel mode I'm also<br />seeing checkvir related failures in complex/field (~250-275 errors)<br />and complex/tip4flex (1 or 2 errors), although strangely the latter<br />PASSES for 4 CPUs and fails for other numbers. Is this kind of<br />behaviour also expected?</div></blockquote>Yes. The virial is computed from a very large summation over floats and is extremely sensitive to architecture and compilation conditions. Thus, erratic failures.<br /><blockquote cite="mid:69649d441002121001tcddce6ak4a471ab60d427102@mail.gmail.com" class="iwcQuote" style="border-left: 1px solid rgb(0, 0, 255); padding-left: 13px; margin-left: 0pt;" type="cite"><div class="mimepart text plain"><br />PS I get fails in several other tests but they just seem to be due to<br />an inability to do domain decomposition on those tests.</div></blockquote>Yes, that's normal. A suitably small case for single-cpu tests need not be able to undergo arbitrary DD.<br /><blockquote cite="mid:69649d441002121001tcddce6ak4a471ab60d427102@mail.gmail.com" class="iwcQuote" style="border-left: 1px solid rgb(0, 0, 255); padding-left: 13px; margin-left: 0pt;" type="cite"><div class="mimepart text plain"><br />PPS Strangely in parallel mode all the pdb2gmx tests pass, unlike in<br />single-CPU mode.</div></blockquote>Hmm, not sure about that, but pdb2gmx isn't parallel, so forget about it.<br _moz_dirty="" /><br _moz_dirty="" />All these kinds of issues are why the wiki no longer encourages the use of the regression tests..<br _moz_dirty="" type="_moz" /><div _moz_dirty="" class="mimepart text plain"><br _moz_dirty="" />Mark<br _moz_dirty="" /></div><blockquote cite="mid:69649d441002121001tcddce6ak4a471ab60d427102@mail.gmail.com" class="iwcQuote" style="border-left: 1px solid rgb(0, 0, 255); padding-left: 13px; margin-left: 0pt;" type="cite"></blockquote></div>