<html><head><style type='text/css'>p { margin: 0; }</style></head><body><div style='font-family: Arial; font-size: 12pt; color: #000000'><font class="Apple-style-span" face="'times new roman', 'new york', times, serif"><br>>----- Original Message -----<br>>From: "Rahul Nabar" <rpnabar@gmail.com><br>>To: "Jan Heichler" <jan.heichler@gmx.net><br>>Cc: "Beowulf Mailing List" <beowulf@beowulf.org>, "Mark Hahn" <hahn@mcmaster.ca><br>>Sent: Wednesday, May 13, 2009 10:21:06 AM GMT -06:00 US/Canada Central<br>>Subject: Re: Re[2]: [Beowulf] recommendations for cluster upgrades<br>><br>>On Wed, May 13, 2009 at 1:55 AM, Jan Heichler <jan.heichler@gmx.net> wrote:<br>>> I'm really surprised that everyone just screams "Nehalem" - of course the<br>>> platorm is the fastest that you can buy for money at the moment. It is the<br>>> youngest design so that is not suprising.<br>>><br>>> But clustering is always about price/performance. And AMD doesn't look so<br>>> bad there.<br>><br>>Absolutely. The price/performance ratio is what I'm really interested in.</font><div><font class="Apple-style-span" face="'times new roman'"><br></font></div><div><span class="Apple-style-span" style="font-family: 'times new roman'; ">Of course, neither the numerator or denominator here are atomic. In particular,</span></div><div><font class="Apple-style-span" face="'times new roman', 'new york', times, serif">performance must refer to sustained performance on your application(s). I would</font></div><div><font class="Apple-style-span" face="'times new roman'">expect bandwidth intensive applications to show enough of an advantage on</font></div><div><font class="Apple-style-span" face="'times new roman'">Nehalem to compensate for the difference in processor price; on the other hand</font></div><div><font class="Apple-style-span" face="'times new roman'">if you are planning a socket-only upgrade, the result might be different. This </font></div><div><font class="Apple-style-span" face="'times new roman'">is an occasion to break out Excel, and crunch your numbers. In addition,</font></div><div><font class="Apple-style-span" face="'times new roman'">I would consider the average processor count of the your work load. If it typically</font></div><div><font class="Apple-style-span" face="'times new roman'">does not exceed your current system size, running two clusters with different</font></div><div><font class="Apple-style-span" face="'times new roman'">architectures as a throughput engine might make sense. </font></div><div><font class="Apple-style-span" face="'times new roman'"><br></font></div><div><font class="Apple-style-span" face="'times new roman'">Regards,</font></div><div><font class="Apple-style-span" face="'times new roman'"><br></font></div><div><font class="Apple-style-span" face="'times new roman'">rbw</font></div><div><font class="Apple-style-span" face="'times new roman'"> </font></div><div><font class="Apple-style-span" face="'times new roman', 'new york', times, serif">_______________________________________________<br>Beowulf mailing list, Beowulf@beowulf.org sponsored by Penguin Computing<br>To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf</font><br></div></div></body></html>