<html><head></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><div><div>On Jun 29, 2010, at 3:53 AM, Roland Schulz wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite"><br><br><div class="gmail_quote">On Mon, Jun 28, 2010 at 4:32 AM, Carsten Kutzner <span dir="ltr"><<a href="mailto:ckutzne@gwdg.de">ckutzne@gwdg.de</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div class="im">On Jun 28, 2010, at 10:23 AM, <a href="mailto:hess@sbc.su.se">hess@sbc.su.se</a> wrote:<br>
<br>
> Hi,<br>
><br>
> But does dlb get turned on at some point without -dlb?<br>
><br>
</div>Apparently not, at least not in the runs I just looked at.<br></blockquote><div><br></div><div>Does it crash immediately or after many steps?</div><div>If it is the later one: often it is possible to make it reproducible crash by setting GMX_DLB_FLOP to 1.</div>
<div><br></div></div></blockquote>Hi,<br>it crashes immediately and reproducibly in step 0, with various tpr files that all work</div><div>in 4.0.x, for different architectures/MPI libs. It's not a real problem, because I can as well</div><div>leave the -dlb flag away, but since this worked in 4.0.x, I guess it should still work now.</div><div><br></div><div>Carsten</div><div><br></div><div><blockquote type="cite"><div class="gmail_quote"><div>Roland</div><div><br></div><blockquote class="gmail_quote" style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0.8ex; border-left-width: 1px; border-left-color: rgb(204, 204, 204); border-left-style: solid; padding-left: 1ex; position: static; z-index: auto; ">
<br>
Carsten<br>
<div><div></div><div class="h5"><br>
> Berk<br>
><br>
>> Hi,<br>
>><br>
>> there appears to be an issue with the -dlb flag when turned on<br>
>> from the very beginning in mdrun. For at least two different<br>
>> MD systems (incl. dppc with PME) I get errors like<br>
>><br>
>> Fatal error:<br>
>> 248 particles communicated to PME node 3 are more than 2/3 times the<br>
>> cut-off out of the domain decomposition cell of their charge group in<br>
>> dimension y.<br>
>><br>
>> when started with<br>
>> mpirun -np 4 mdrun -s ../tpr/ap_1ps.tpr -dlb yes.<br>
>><br>
>> With -dlb no everything works fine.<br>
>><br>
>> Carsten<br>
>></div></div></blockquote></div></blockquote></div></body></html>