[Beowulf] Re: Beowulf Digest, Vol 10, Issue 32

Maurice Hilarius maurice at harddata.com
Sun Dec 19 22:00:45 PST 2004


>----- Original Message -----
>From: "Douglas Eadline, Cluster World Magazine" <deadline at linux-mag.com>
>To: "Jim Lux" <jimlux at earthlink.net>
>Cc: <beowulf at beowulf.org>
>
>With our best regards,
>
>Maurice W. Hilarius        Telephone: 01-780-456-9771
>Hard Data Ltd.  FAX:       01-780-456-9772
>11060 - 166 Avenue         email:maurice at harddata.com
>Edmonton, AB, Canada       http://www.harddata.com/
>   T5X 1Y3
>
>
>This email, message, and content, should be considered confidential, 
>and is the copyrighted property of Hard Data Ltd., unless stated otherwise.
>Sent: Sunday, December 19, 2004 10:58 AM
>Subject: Re: [Beowulf] $2500 cluster. What it's good for?

...

>Several times, I've contemplated a cluster to solve some problem, and then,
>by the time I had it all spec'd out and figured out and costed, it turned
>out that I'd been passed by AMD/Intel, and it was better just to go buy a
>(single) faster processor.  There are some interesting power/MIPS trades
>that are non-obvious in this regime, as well as anomalous application
>environments where the development cycle is much slower (not too many "Rad
>Hard" Xeons out there).

That process has changed with the Opterons and dual versus single CPU.
With the CPU's prior to that there were lots of cases we did not get very great computational gains on dual CPU machines.
That is because a dual SMP was around 50% more computationally powerful than a single CPU implementation.
Also older kernels were not very efficient in SMP compared to single.

In many cases the only real reason for going dual was it was still slightly more of a gain than building lots more single CPU machines, especially if we used expensive interconnects.
Further we often had to use SMP machines simply because that was often the only way to get the better PCI interfaces, PCI-64 or PCI- versus PCI32.

Now however the Opteron offers SMP at about 90% efficiency, and that  certainly skews the calculation of performance single versus SMP dual machines.
Add to that the better scheduler in the newer 2.6 kernels and it is a vastly different model than we used to see.






More information about the Beowulf mailing list