Archives


- Beowulf
- Beowulf Announce
- Scyld-users
- Beowulf on Debian

[Beowulf] The fan from hell

Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.

Search

Eray Ozkural examachine at gmail.com
Fri Jun 10 12:13:49 PDT 2005


Hi David,

Are you certain that replacing the fans will solve your heat
problem in the long run? My parallel database code runs
continuously for >1day, up to 40 processors. In the end
we measured that the system consumed more energy
than what the cooling system consumed! When a friend arrived
the other day he found the system room at 45 celcius,
with hot steam in the air :) I don't know what the board
temperature was, but I'm sure they got pretty hot.
Luckily nothing broke down.

Cheers,

On 6/10/05, David Mathog <mathog at mendel.bio.caltech.edu> wrote:
> Today I distributed database fragments out from the main server
> to the 20 compute nodes.  This required the server to run gunzip
> pretty much continuously for a couple of hours. The net result of
> which was that CPU1's temperature rose from 55C to 60-61C.
> Normally the fan on CPU1 sits happily at 5033 RPM. However
> today it started to slow down.  In fact it slowed down below
> 3000RPM, which triggered an automatic shutdown of the server.
> 
> After a bit of experimentation I've now determined that
> this wasn't a coincidence. This CPU fan really does run
> slower when the CPU gets hotter.  It takes a
> few minutes before it starts to slow down, but slow down it
> does.  Not exactly the desired response for a fan on a hot CPU!
> This is on an Athlon MP 2000+ processor and the fan is Dynatron
> model that AMD used with the boxed CPU.
> 
> Any of the system vendors (or anybody else) out there have a
> preferred brand of 60mm fan that doesn't slow down at 60C?
> 
> Thanks,
> 
> David Mathog
> mathog at caltech.edu
> Manager, Sequence Analysis Facility, Biology Division, Caltech
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
> 


-- 
Eray Ozkural (exa), PhD candidate
Comp. Sci. Dept., Bilkent University, Ankara  KDE Project: http://www.kde.org
http://www.cs.bilkent.edu.tr/~erayo  Malfunction: http://www.malfunct.com




More information about the Beowulf mailing list