<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
<style type="text/css" style="display:none;"><!-- P {margin-top:0;margin-bottom:0;} --></style>
</head>
<body dir="ltr">
<div id="divtagdefaultwrapper" style="font-size:12pt;color:#000000;font-family:Calibri,Helvetica,sans-serif;" dir="ltr">
<p style="margin-top:0;margin-bottom:0"><br>
</p>
<p style="margin-top:0;margin-bottom:0">Hi Steffen,</p>
<p style="margin-top:0;margin-bottom:0"><br>
</p>
<p style="margin-top:0;margin-bottom:0">Thanks for the note.</p>
<p style="margin-top:0;margin-bottom:0"><br>
</p>
<p style="margin-top:0;margin-bottom:0">I would recommend fixed dt at a value near your target CFL.</p>
<p style="margin-top:0;margin-bottom:0"><br>
</p>
<p style="margin-top:0;margin-bottom:0">There are several reasons -</p>
<p style="margin-top:0;margin-bottom:0"><br>
</p>
<p style="margin-top:0;margin-bottom:0"></p>
<ol style="margin-bottom: 0px; margin-top: 0px;">
<li><span style="font-size: 12pt;"></span><span style="font-size: 12pt;"></span>I doubt you gain that much with variable dt</li><li>I personally don't like Nek's auto-dt selection scheme</li><li>You have to reorthogonalize the projection basis every time dt changes (expensive)</li></ol>
Paul
<p></p>
<p style="margin-top:0;margin-bottom:0"><br>
</p>
</div>
<hr style="display:inline-block;width:98%" tabindex="-1">
<div id="divRplyFwdMsg" dir="ltr"><font face="Calibri, sans-serif" style="font-size:11pt" color="#000000"><b>From:</b> Nek5000-users <nek5000-users-bounces@lists.mcs.anl.gov> on behalf of nek5000-users@lists.mcs.anl.gov <nek5000-users@lists.mcs.anl.gov><br>
<b>Sent:</b> Monday, September 10, 2018 9:19:00 AM<br>
<b>To:</b> nek5000-users@lists.mcs.anl.gov<br>
<b>Subject:</b> Re: [Nek5000-users] Proper setup for AMG solver</font>
<div> </div>
</div>
<div class="BodyFragment"><font size="2"><span style="font-size:11pt;">
<div class="PlainText">Dear Nek experts,<br>
<br>
<br>
I wanted to add some remarks on my performance tests conducted last week <br>
mostly for the turbulent pipe flow at Re_b=5300<br>
<br>
- The bad performance of the AMG solver was indeed due to an old version <br>
of amg_hypre, as Stefan mentioned already.<br>
For Re_b=5300 XXT and AMG showed the same performance.<br>
<br>
- Additionally, I could increase the performance by about 20% using the <br>
following settings:<br>
-- lxd=10 (instead of 12, for lx1=8)<br>
-- lx2=lx1-0 (PN-PN)<br>
-- lower tolerances for p=1e-5 (instead of 1e-8) and 1e-6 for velocity <br>
and passive scalars<br>
-- turning of projection except for the fields of low Prandtl number<br>
<br>
- Going from BDF3/EXT3 with a variable DT and targetCFL=0.5 to BDF2/OIFS <br>
with variable DT and targetCFL=0.5 resulted in a longer time / timestep <br>
(3x) but a also a larger average DT (6x).<br>
Assuming the collected statistics require a similar averaging time in <br>
wash-outs, BDF2/OIFS is advantageous (even for 10 passive scalars).<br>
Before this discussion, I was running at constant DT and collected <br>
statistics every 10th step. Now with a larger DT using characteristics, <br>
should I collect statistics for each step?<br>
<br>
- Regarding the filterWeight and filterCutoffRatio, I found out (with <br>
Stefan's help) that the divergence error in L2 norm (when using PN-PN) <br>
is affected by filterWeight.<br>
For the setup at Re_b=5300 a filterWeight of 27 seemed to be OK <br>
considering a divergence error in L2 norm of L2=3e-2, whereas a weight <br>
of 54 resulted in L2=1e-1.<br>
<br>
<br>
How these settings will affect the statistics, I am testing now.<br>
<br>
Thank you all for your help and suggestions.<br>
<br>
<br>
Cheers,<br>
Steffen<br>
<br>
_______________________________________________<br>
Nek5000-users mailing list<br>
Nek5000-users@lists.mcs.anl.gov<br>
<a href="https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users">https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users</a><br>
</div>
</span></font></div>
</body>
</html>