<div dir="ltr">On Fri, Oct 18, 2013 at 4:49 PM, Torquil Macdonald Sørensen <span dir="ltr"><<a href="mailto:torquil@gmail.com" target="_blank">torquil@gmail.com</a>></span> wrote:<br><div class="gmail_extra"><div class="gmail_quote">
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi!<br>
<br>
You probably hit the nail on the head with the debugging suggestion. After reinstalling petsc using --with-debugging=0, I'm getting more encouraging results. For the same problem as before, but with a larger matrix (now 10000x10000), I'm now getting the following scaling:<br>
<br>
# n_core time (sec)<br>
1 1540<br>
2 870<br>
3 660<br></blockquote><div><br></div><div>Here you are very likely running into the memory bandwidth bottleneck. Unfortunately, there is not</div><div>a whole lot you can do about this. This is not something vendor's like to acknowledge.</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">
4 571<br>
<br>
which is much more promising than before. So thanks!<br>
<br>
- Torquil<br>
<br>
On 18/10/13 20:49, Jose E. Roman wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
No, you don't need MUMPS for ex1.c with default options.<br>
Then I don't know what is happening. Did you build PETSc --with-debugging=0 ? You should send performance data obtained with -log_summary, otherwise we cannot guess what happens.<br>
<br>
Jose<br>
<br>
<br>
El 18/10/2013, a las 17:21, Torquil Macdonald Sørensen escribió:<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
It is a standard eigenvalue problem, and I'm only using default algorithm options, so I don't think I'm using any nontrivial spectral transformation.<br>
<br>
The program is just ex1.c from the documentation, available here:<br>
<br>
<a href="http://www.grycap.upv.es/slepc/documentation/current/src/eps/examples/tutorials/ex1.c" target="_blank">http://www.grycap.upv.es/<u></u>slepc/documentation/current/<u></u>src/eps/examples/tutorials/<u></u>ex1.c</a><br>
<br>
FAQ #6 mentions MUMPS, which I don't have the moment, so I'll have to look into that.<br>
<br>
- Torquil<br>
<br>
<br>
<br>
On 18 October 2013 16:46, Jose E. Roman <<a href="mailto:jroman@dsic.upv.es" target="_blank">jroman@dsic.upv.es</a>> wrote:<br>
<br>
El 18/10/2013, a las 16:42, Torquil Macdonald Sørensen escribió:<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi!<br>
<br>
I'm just starting to use SLEPc on a quad-core workstation computer (Intel Core i5-2500 CPU @ 3.30GHz).<br>
<br>
But I'm not getting the speedup that I was expecting, when increasing the number of cores. E.g. when running ex1.c from the documentation, on 1, 2, 3 and 4 cores. When asking for the 500 eigenvalues of smallest magnitude, for a 5000x5000 matrix, I'm getting the following approximate timings:<br>
<br>
# n_core time_usage<br>
1 480s<br>
2 400s<br>
3 500s<br>
4 580s<br>
<br>
Should I not expect a speedup for n_core > 2?<br>
<br>
The example is using the krylovschur algorithm. For all four runs, the program states "Number of iterations of the method: 33".<br>
<br>
I'm using PETSc 3.4.3, SLEPc 3.4.3, OpenMPI 1.6.5, and everything is compiled using GCC 4.8.1 or 4.8.2.<br>
<br>
Best regards<br>
Torquil Sørensen<br>
<br>
</blockquote>
Is it a generalized eigenproblem and/or are you doing shift-and-invert? In that case, have a look at FAQ #6 <a href="http://www.grycap.upv.es/slepc/documentation/faq.htm" target="_blank">http://www.grycap.upv.es/<u></u>slepc/documentation/faq.htm</a><br>
<br>
Jose<br>
<br>
<br>
</blockquote></blockquote>
<br>
</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
</div></div>