Re: processors benchmarks etc.

Search:

Re: processors benchmarks etc.

From: Drew McCormack <drewmccormack_at_email.domain.hidden>
Date: Fri, 31 Oct 2003 22:39:29 +0100
X-Mailer: Apple Mail (2.606)

Hi Georg,

Everyone that knows me will know I can't let an opportunity like this
go past without plugging an Apple ;-)

Apple is not traditionally a manufacturer you would consider for a
beowolf cluster, but recent events have changed that somewhat. Apple's
operating system, Mac OS X, is a UNIX system based on FreeBSD. A few
months back, Apple released the first G5 systems: these are computers
based on the 64 bit PPC970 chip from IBM, which is a cut-down Power4.
The chip is 64 bit.

Virginia Tech has just built the most powerful Beowolf cluster ever
made with 1100 of these G5 systems, each with dual 2Ghz G5 chips. This
system is the 3rd fastest computer in the world at this point in time.
It has a theoretical peak performance of 15.9Teraflops, and has
achieved real-world performance of 9.9Teraflops or so. See
http://www.macdevcenter.com/pub/a/mac/2003/10/29/osxcon_g5cluster.html
for more info about this system, and how they built it in just a few
months, costing only $5.2million US.

We haven't had much access to G5 systems for ADF testing and
optimization, but to give you some idea of what we have been able to
find out, a 2Ghz G5 processor runs ADF about as fast as a 2.4Ghz intel
xeon processor. The G5 scales a little better to dual processors than
the xeon. A dual processor G5 system begins at around $3000, but you
can check prices in detail at www.apple.com/store

Hope this helps,
Drew McCormack
Vrije Universiteit, Amsterdam

On Oct 31, 2003, at 7:03 PM, H. Georg Schreckenbach wrote:

> Dear list,
>
> does anybody have any recent benchmarks (for ADF and other QC codes) or
> anecdotal evidence about which computer systems are the best right now
> to build
> a Beowulf? In particular, I have two quotes from a vendor for the
> following
> dual-processor systems
> (i) Intel Xeon 2.8GHz
> (ii) Opteron AMDOPT242 1.6GHz
> The vendor claims that the latter might be better. (The price is
> comparable.)
> Any comments?
>
> Thank you, and best wishes, Georg
>
> --
> Dr. H. Georg Schreckenbach
> Department of Chemistry, University of Manitoba
> Winnipeg, MB, R3T 2N2, Canada
> phone: (+204) 474-6261
> FAX: (+204) 474-7608
>
> -------------------------------------------------
> This mail sent through IMP: http://horde.org/imp/
Received on 2003-11-03 14:32:57

This archive was generated by hypermail 2.2.0 : 2006-11-02 07:00:02 CET