On Thu, Nov 10, 2011 at 12:30 PM, Bogdan Dita <span dir="ltr"><<a href="mailto:bogdan@lmn.pub.ro">bogdan@lmn.pub.ro</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<br>
Hello,<br>
<br>
Until a few days ago I've only be using PETSc in debug mode and when I<br>
switch to the optimised version(--with-debugging=0) I got a strange<br>
result regarding the solve time, what I mean is that it was 10-15 %<br>
higher then in debug mode.<br>
I'm trying to solve a linear system in parallel with superlu_dist, and<br>
I've tested my program on a Beowulf cluster, so far only using a single<br>
node with 2 quad-core Intel processors.<br>
From what I know the "no debug" version should be faster and I know it<br>
should be faster because on my laptop(dual-core Intel) for the same<br>
program and even the same matrices the solve time for the optimised<br>
version is 2 times faster, but when I use the cluster the optimised<br>
version time is slower then the debug version.<br>
My quess is that this has something to do with MPI. Any thoughts?<br></blockquote><div><br></div><div>Performance questions are meaningless without the output of -log_summary for all cases.</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;">
Best regards,<br>
<font color="#888888"> Bogdan Dita<br>
</font></blockquote></div><br><br clear="all"><div><br></div>-- <br>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<br>