<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Ali,<br>
<br>
You might want to consider looking into Tyan's VX50 systems, that
system is able to handle dual core Opterons (8*2=16 cores, it does not
support the 9000 series quad core though) and it can go up to 128Gig of
RAM. Some advantages are:<br>
<br>
Each Opteron chip (datasheets:
<a class="moz-txt-link-freetext" href="http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_739_9003,00.html">http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_739_9003,00.html</a>):
<br>
- has its own internal MMU (Memory Management Unit),<br>
- is interconnected to another chip using Hypertransport<br>
- is a single system which consumes and generates very little heat
compared to multiple systems connected together<br>
- and many more...<br>
<br>
Though I would love it if the other that are showing speed results post
a link to the test program so that I could also post some results so we
could compare ;)<br>
<br>
Eric<br>
<br>
<br>
amjad ali wrote:
<blockquote
cite="mid:428810f20711270556x2bcf06c4me3d28b1b62dbd660@mail.gmail.com"
type="cite">
<div>Hello,</div>
<div>I planned to buy 9 PCs each having one Core2Duo E6600 (networked
with GiGE) to make cluster for running PETSc based applications.</div>
<div> </div>
<div>I got an advice that because the prices of Xeon Quadcore is
going to drop next month, so I should buy 9 PCs each having one
Quadcore Xeon (networked with GiGE) to make cluster for running PETSc
based applications.</div>
<div> </div>
<div>Which is better for me to get better performance/speedup?</div>
<div> </div>
<div>My question is due to following as given in PETSc-FAQ:</div>
<div> </div>
<div>
<p><a moz-do-not-send="true" name="computers"><strong><font
color="#ff0000">What kind of parallel computers or clusters are needed
to use PETSc?</font></strong><br>
</a><br>
PETSc can be used with any kind of parallel system that supports MPI.<span
style="font-weight: bold;"> BUT </span>for any decent performance one
needs </p>
<ul>
<li>a <span style="font-weight: bold;">fast, low-latency
interconnect</span>; any ethernet, even 10 gigE simply cannot provide
the needed performance. </li>
<li><span style="font-weight: bold;">high per-CPU memory performance</span>.
Each CPU (core in dual core systems) needs to have its <span
style="font-weight: bold;">own</span> memory bandwith of roughly 2 or
more gigabytes. For example, standard dual processor "PC's" will <span
style="font-weight: bold;">not</span> provide better performance when
the second processor is used, that is, you will not see speed-up when
you using the second processor. This is because the speed of sparse
matrix computations is almost totally determined by the speed of the
memory, not the speed of the CPU. </li>
</ul>
</div>
<div> </div>
<div>regards,</div>
<div>Amjad Ali.</div>
<pre wrap="">
<hr size="4" width="90%">
_______________________________________________
Beowulf mailing list, <a class="moz-txt-link-abbreviated" href="mailto:Beowulf@beowulf.org">Beowulf@beowulf.org</a>
To change your subscription (digest mode or unsubscribe) visit <a class="moz-txt-link-freetext" href="http://www.beowulf.org/mailman/listinfo/beowulf">http://www.beowulf.org/mailman/listinfo/beowulf</a>
</pre>
</blockquote>
<br>
</body>
</html>