<div dir="ltr"><div class="gmail_default" style="font-family:arial,helvetica,sans-serif">The fix to PetscWrappers::VectorBase::all_zero seems to have done the trick.  However, I am still seeing large iteration counts on the advection systems.  Overall, this is a small(ish) portion of the runtime, though it is curious.  It seems to be the same whether I use one or several processors.  Can you reproduce that?</div>
</div><div class="gmail_extra"><br><br><div class="gmail_quote">On Tue, Jan 21, 2014 at 6:20 AM, Timo Heister <span dir="ltr"><<a href="mailto:heister@clemson.edu" target="_blank">heister@clemson.edu</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">> SolverCG<LinearAlgebra::Vector> solver(solver_control);<br>
> with<br>
> PETScWrappers::SolverCG solver(solver_control);<br>
<br>
</div>Can you please check if deal.II r32242 fixes this problem for you?<br>
<br>
Note that I still don't quite understand the differences in number of<br>
iterations, especially because the composition solver uses ILU(0) and<br>
the deal.II GMRES so they should be very similar. Could you check if<br>
this changes if we are on 1 cpu?<br>
<div class="HOEnZb"><div class="h5"><br>
--<br>
Timo Heister<br>
<a href="http://www.math.clemson.edu/~heister/" target="_blank">http://www.math.clemson.edu/~heister/</a><br>
_______________________________________________<br>
Aspect-devel mailing list<br>
<a href="mailto:Aspect-devel@geodynamics.org">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>
</div></div></blockquote></div><br></div>