Thanks, for a quickly reply. I have four nodes (GigaEthernet), each node with two Quad Core E5410@2.33GHz, Mem 16Gb - DDR2 667Mhz.<span id="result_box" class="short_text"><span style="background-color: rgb(235, 239, 249);" title="definitivamente">, and definitely I'm not have</span></span> the enough memory bandwidth for a reasonable speedup. <br>
<div dir="ltr" style=""><span id="result_box" class="short_text"><span style="background-color: rgb(235, 239, 249);" title="razonable"><br>This may be a dummy question but in the second bullet says "</span></span>its <span style="font-weight: bold;">own</span>
memory bandwith of roughly 2 or more gigabyte<span id="result_box" class="short_text"><span style="background-color: rgb(235, 239, 249);" title="razonable">s", this means gigabytes/seconds, or refers to amount of memory per core?. Thanks a lot. <br>
<br>Pedro<br><br></span></span></div><div class="gmail_quote">2010/4/14 Barry Smith <span dir="ltr"><<a href="mailto:bsmith@mcs.anl.gov">bsmith@mcs.anl.gov</a>></span><br><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<br>
Second bullet at <a href="http://www.mcs.anl.gov/petsc/petsc-as/documentation/faq.html#computers" target="_blank">http://www.mcs.anl.gov/petsc/petsc-as/documentation/faq.html#computers</a><br><font color="#888888">
<br>
Barry</font><div><div></div><div class="h5"><br>
<br>
On Apr 14, 2010, at 4:10 PM, Jed Brown wrote:<br>
<br>
<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
On Wed, 14 Apr 2010 18:02:11 -0300, Pedro Torres <<a href="mailto:torres.pedrozpk@gmail.com" target="_blank">torres.pedrozpk@gmail.com</a>> wrote:<br>
<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
Hello,<br>
<br>
Sorry if this questions its not appropiate for the petsc-list, but I really<br>
want to known what happen when I'm getting differente KSP time results. For<br>
example, allocating two process in the same node I get 6.24 sec, and when<br>
allocating two process in two nodes (1 process per node) I get 4.7sec. Is<br>
there a memory contention problem in my node?? The problem get worst when<br>
increase the number of process.<br>
</blockquote>
<br>
Sparse matrix kernels are primarily limited by memory bandwidth which<br>
does not increase much with multicore hardware (vendors rarely mention<br>
this). When you use multiple cores per socket, they have to share the<br>
available bandwidth, so you get lower performance. It's *usually* still<br>
faster to use the available cores, but the per-core performance is<br>
definitely lower than when using only one core per socket.<br>
<br>
Jed<br>
</blockquote>
<br>
</div></div></blockquote></div><br><br clear="all"><br>-- <br>Pedro Torres<br>GESAR/UERJ<br>Rua Fonseca Teles 121, São Cristóvão<br>Rio de Janeiro - Brasil<br>