<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Tue, Nov 7, 2017 at 8:43 AM, Bernardo Rocha <span dir="ltr"><<a href="mailto:bernardomartinsrocha@gmail.com" target="_blank">bernardomartinsrocha@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div style="font-family:arial,helvetica,sans-serif;font-size:small">Hello everyone,</div><div style="font-family:arial,helvetica,sans-serif;font-size:small"><br></div><div style="font-family:arial,helvetica,sans-serif;font-size:small">I have a general question about the performance of the PCFieldSplit</div><div style="font-family:arial,helvetica,sans-serif;font-size:small">that I'm not sure if I understood properly.</div><div style="font-family:arial,helvetica,sans-serif;font-size:small"><br></div><div style="font-family:arial,helvetica,sans-serif;font-size:small">Consider a simple Poisson problem discretized by FEM into a system Ax=b </div><div style="font-family:arial,helvetica,sans-serif;font-size:small">which is then solved by CG and Jacobi.</div><div style="font-family:arial,helvetica,sans-serif;font-size:small"><br></div><div style="font-family:arial,helvetica,sans-serif;font-size:small">Then, I create a "vectorial Poisson" problem by simply adding another block</div><div style="font-family:arial,helvetica,sans-serif;font-size:small">of this problem to create a block-like version of it. </div><div style="font-family:arial,helvetica,sans-serif;font-size:small">Something like</div><div style="font-family:arial,helvetica,sans-serif;font-size:small">[ [A, 0]</div><div style="font-family:arial,helvetica,sans-serif;font-size:small">  [0, A]] </div><div style="font-family:arial,helvetica,sans-serif;font-size:small">then I create a PCFieldSplit with CG and Jacobi for each block.</div><div style="font-family:arial,helvetica,sans-serif;font-size:small"><br></div><div style="font-family:arial,helvetica,sans-serif;font-size:small">Either with additive or multiplicative fieldsplit, the PC is much better and solves it </div><div style="font-family:arial,helvetica,sans-serif;font-size:small">with fewer iterations than the scalar case. However, the execution time taken by</div><div style="font-family:arial,helvetica,sans-serif;font-size:small">the PCFieldSplit is much bigger than the simple Jacobi for the scalar case.</div><div style="font-family:arial,helvetica,sans-serif;font-size:small"><br></div><div style="font-family:arial,helvetica,sans-serif;font-size:small">(From -log_view I see all the time difference in PCApply)</div><div style="font-family:arial,helvetica,sans-serif;font-size:small"><br></div><div style="font-family:arial,helvetica,sans-serif;font-size:small">Why is this happening?</div></div></blockquote><div><br></div><div>1) This is block-Jacobi, why not use PCBJACOBI? Is it because you want to select rows?</div><div><br></div><div>2) We cannot tell anything without knowing how many iterates were used</div><div><br></div><div>  -ksp_monitor_true_residual -ksp_converged_reason -pc_fieldsplit_[0,1]_ksp_monitor_true_residual</div><div><br></div><div>3) We cannot say anything about performance without seeing the log for both runs</div><div><br></div><div>  -log_view</div><div><br></div><div>   Thanks,</div><div><br></div><div>     Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div style="font-family:arial,helvetica,sans-serif;font-size:small">Best regards,</div><div style="font-family:arial,helvetica,sans-serif;font-size:small">Bernardo</div><div style="font-family:arial,helvetica,sans-serif;font-size:small"><br></div></div>
</blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div>What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.<br>-- Norbert Wiener</div><div><br></div><div><a href="http://www.caam.rice.edu/~mk51/" target="_blank">https://www.cse.buffalo.edu/~knepley/</a><br></div></div></div></div></div>
</div></div>