<div dir="ltr"><div class="gmail_default" style="font-family:arial,helvetica,sans-serif">Hi Timo,</div><div class="gmail_default" style="font-family:arial,helvetica,sans-serif">I just tried undoing that change, and am still getting large iteration counts.</div>
<div class="gmail_default" style="font-family:arial,helvetica,sans-serif">Ian</div><div class="gmail_default" style="font-family:arial,helvetica,sans-serif"><br></div><div class="gmail_extra"><br><br><div class="gmail_quote">
On Mon, Jul 1, 2013 at 9:06 AM, Timo Heister <span dir="ltr"><<a href="mailto:heister@math.tamu.edu" target="_blank">heister@math.tamu.edu</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><p dir="ltr">Dear Ian,</p>
<p dir="ltr">Could this be related to r29539 ? <a href="http://dealii.org/websvn/revision.php?repname=deal.II+Repository&path=%2F&rev=29539&peg=29539" target="_blank">http://dealii.org/websvn/revision.php?repname=deal.II+Repository&path=%2F&rev=29539&peg=29539</a></p>



<div class="gmail_quote"><div><div>On Jul 1, 2013 5:50 PM, "Ian Rose" <<a href="mailto:ian.rose@berkeley.edu" target="_blank">ian.rose@berkeley.edu</a>> wrote:<br type="attribution"></div></div>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div>
<div dir="ltr">H<div style="font-family:arial,helvetica,sans-serif;display:inline" class="gmail_default">i all,</div><div><div style="font-family:arial,helvetica,sans-serif;display:inline" class="gmail_default"></div></div>




<div><div style="font-family:arial,helvetica,sans-serif;display:inline" class="gmail_default">Here's a bit of a doozy of a problem.  I noticed the other day a sharp increase in the number of iterations aspect required for the Stokes solver to converge.  As in, the cheap solver never converges, and the expensive one takes ~50-200 iterations.  After a bunch of experimentation, I think it may be due to some recent changes in deal.ii.  If I run simple_2d_shells.prm on 8 cores with deal.ii 7.3.0, I get the following representative output: </div>



</div><div><div style="font-family:arial,helvetica,sans-serif;display:inline" class="gmail_default"><br></div></div><div><div style="display:inline" class="gmail_default"><div><font face="arial, helvetica, sans-serif">*** Timestep 148:  t=4.6211e+06 years</font></div>



<div><font face="arial, helvetica, sans-serif">   Solving temperature system... 13 iterations.</font></div><div><font face="arial, helvetica, sans-serif">   Solving Stokes system... 13 iterations.</font></div><div style="font-family:arial,helvetica,sans-serif">



<br></div><div style="font-family:arial,helvetica,sans-serif">If, however, I run it using deal.ii r29892 I get the following</div><div style="font-family:arial,helvetica,sans-serif"><br></div><div><div><font face="arial, helvetica, sans-serif">*** Timestep 27:  t=2.68009e+06 years</font></div>



<div><font face="arial, helvetica, sans-serif">   Solving temperature system... 13 iterations.</font></div><div><font face="arial, helvetica, sans-serif">   Solving Stokes system... 30+219 iterations.</font></div><div style="font-family:arial,helvetica,sans-serif">



<br></div></div><div style="font-family:arial,helvetica,sans-serif">I have no idea what may be causing this, or how reproducible it is across different systems (though I have noticed the same increase in iterations on TACC Stampede).  In both cases, I am using trilinos 11.2.3 and p4est 0.3.4.  Everything is built with icc-12.1, mkl math libraries, and using myrinet MPI.  Any insight would be greatly appreciated.</div>



<div style="font-family:arial,helvetica,sans-serif">Best,</div><div style="font-family:arial,helvetica,sans-serif">Ian</div><div style="font-family:arial,helvetica,sans-serif"><br></div></div></div></div>
<br></div></div>_______________________________________________<br>
Aspect-devel mailing list<br>
<a href="mailto:Aspect-devel@geodynamics.org" target="_blank">Aspect-devel@geodynamics.org</a><br>
<a href="http://geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel" target="_blank">http://geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel</a><br></blockquote></div>
<br>_______________________________________________<br>
Aspect-devel mailing list<br>
<a href="mailto:Aspect-devel@geodynamics.org" target="_blank">Aspect-devel@geodynamics.org</a><br>
<a href="http://geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel" target="_blank">http://geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel</a><br></blockquote></div><br></div></div>