<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN"><HTML><head><META content="text/html; charset=utf-8" http-equiv="Content-Type">
<META content="text/html; charset=utf-8" http-equiv="Content-Type">
<STYLE TYPE="text/css"><!-- /* Style Definitions */p.b506152b-047c-412d-8126-2cf8b483f1d5, li.b506152b-047c-412d-8126-2cf8b483f1d5, div.b506152b-047c-412d-8126-2cf8b483f1d5, table.b506152b-047c-412d-8126-2cf8b483f1d5Table {margin:0cm; margin-bottom:.0001pt;} div.Section1 {page:Section1;} --></STYLE>
<META content="text/html; charset=utf-8" HTTP-EQUIV="Content-Type">
<META CONTENT="MS Exchange Server version 6.5.7638.1" NAME="Generator">
<TITLE>hypre preconditioners</TITLE>
<META content="text/html; charset=utf-8" http-equiv="Content-Type">
</head><BODY>
<FONT FACE="Arial" SIZE="2">
<P><FONT FACE="Arial" SIZE="2">
<!-- Converted from text/plain format -->
<P><FONT SIZE=2>I'm solving the steady incompressible Navier-Stokes equations (discretized with FV on unstructured grids) using the SIMPLE Pressure Correction method. I'm using Picard linearization and solve the system for the momentum equations with BICG and for the pressure equation with CG. Currently, for parallel runs, I'm using JACOBI as a preconditioner. My grids typically have a few million cells and I use between 4 and 16 cores (1 to 4 quadcore CPUs on a linux cluster). A significant portion of the CPU time goes into solving the pressure equation. To reach the relative tolerance I need, CG with JACOBI takes about 100 iterations per outer loop for these problems.<BR>
<BR>
In order to reduce CPU time, I've compiled PETSc with support for Hypre and I'm looking at BoomerAMG and Euclid to replace JACOBI as a preconditioner for the pressure equation. With default settings, both BoomerAMG and Euclid greatly reduce the number of iterations: with BoomerAMG 1 or 2 iterations are enough, with Euclid about 10. However, I do not get any reduction in CPU time. With Euclid, CPU time is similar to JACOBI and with BoomerAMG it is approximately doubled.<BR>
<BR>
Is this what one can expect? Are BoomerAMG and Euclid meant for much larger problems? I understand Hypre uses a different matrix storage format, is CPU time 'lost in translation' between PETSc and Hypre for these small problems? Are there maybe any settings I should change?<BR>
<BR>
Chris<BR>
<BR>
<BR>
<BR>
<BR>
<BR>
<BR>
</FONT>
</P>
</FONT></P><FONT FACE="Arial" SIZE="2">
<TABLE BORDER="0" CELLPADDING="0" CELLSPACING="1" STYLE="WIDTH: 475px; HEIGHT: 180px" WIDTH="471">
<TBODY>
<TR>
<TD COLSPAN="3"><IMG ALIGN="left" ALT BORDER="0" HSPACE="0" SRC="cid:imageaa6970.jpg@92b4de85.c8e74f36" STYLE="VERTICAL-ALIGN: bottom"><IMG ALIGN="center" ALT BORDER="0" HEIGHT="26" HSPACE="0" SRC="cid:image8cdbf6.jpg@c2563e89.62a94add" STYLE="VERTICAL-ALIGN: top; WIDTH: 410px; HEIGHT: 26px" WIDTH="404"></TD></TR>
<TR>
<TD COLSPAN="3"><FONT COLOR="#828282" FACE="Arial" SIZE="2">dr. ir. Christiaan Klaij</FONT></TD></TR>
<TR>
<TD COLSPAN="3"><FONT COLOR="#828282" FACE="Arial" SIZE="2">CFD Researcher</FONT></TD></TR>
<TR>
<TD COLSPAN="3"><FONT COLOR="#828282" FACE="Arial" SIZE="2">Research & Development</FONT></TD></TR>
<TR>
<TD></TD>
<TD><FONT COLOR="#828282" FACE="Arial" SIZE="2"><STRONG>MARIN</STRONG></FONT></TD></TR>
<TR>
<TD WIDTH="45%"></TD>
<TD WIDTH="30%"><FONT COLOR="#828282" FACE="Arial" SIZE="2">2, Haagsteeg</FONT></TD>
<TD></TD></TR>
<TR>
<TD WIDTH="45%"><FONT FACE="Arial" SIZE="2" STYLE="TEXT-TRANSFORM: lowercase"><a href="mailto:C.Klaij@marin.nl">C.Klaij@marin.nl</a></FONT></TD>
<TD WIDTH="30%"><FONT COLOR="#828282" FACE="Arial" SIZE="2">P.O. Box 28</FONT></TD>
<TD WIDTH="25%"><FONT COLOR="#828282" FACE="Arial" SIZE="2"><FONT COLOR="#828282">T </FONT>+31 317 49 39 11</FONT></TD>
<TR>
<TD WIDTH="45%"><FONT COLOR="#828282" FACE="Arial" SIZE="2"></FONT></TD>
<TD WIDTH="30%"><FONT COLOR="#828282" FACE="Arial" SIZE="2">6700 AA Wageningen</FONT></TD>
<TD WIDTH="25%"><FONT COLOR="#828282" FACE="Arial" SIZE="2">F +31 317 49 32 45</FONT></TD></TR>
<TR>
<TD WIDTH="45%"><FONT COLOR="#828282" FACE="Arial" SIZE="2">T <FONT COLOR="#828282" FACE="Arial" SIZE="2">+31 317 49 33 44</FONT></FONT></TD>
<TD WIDTH="30%"><FONT COLOR="#828282" FACE="Arial" SIZE="2">The Netherlands</FONT></TD>
<TD WIDTH="25%"><FONT COLOR="#828282" FACE="Arial" SIZE="2"><FONT COLOR="#828282">I </FONT><A HREF="http://www.marin.nl/"><FONT COLOR="blue">www.marin.nl</FONT></A></FONT></TD></TR></TR></TBODY></TABLE>
<P></P></FONT><FONT FACE="Arial" SIZE="2"><FONT FACE="Arial" SIZE="2"></FONT></FONT>
<P></P>
<P><FONT COLOR="#828282" FACE="Arial" SIZE="2"><A HREF="http://www.marin.nl/web/show/id=46836/contentid=2324"><FONT FACE="Arial" SIZE="2">MARIN webnews:</FONT></A><FONT FACE="Arial" SIZE="2"> <SPAN LANG="NL" STYLE="FONT-SIZE: 10pt; FONT-FAMILY: Arial">First AMT'09 conference, Nantes, France, September 1-2</SPAN> <BR></FONT></FONT><BR></P></FONT><SPAN STYLE="FONT-SIZE: 10pt; FONT-STYLE: italic; FONT-FAMILY: Arial"><FONT COLOR="#828282" SIZE="1"><FONT COLOR="gray" FACE="Arial" SIZE="1"><SPAN LANG="EN-GB" STYLE="FONT-SIZE: 8pt; COLOR: gray; FONT-FAMILY: Arial">This e-mail may be confidential, privileged and/or protected by copyright. If you are not the intended recipient, you should return it to the sender immediately and delete your copy from your system.<BR></SPAN></FONT></FONT></SPAN><BR></BODY></HTML>