[Beowulf] Re: Beowulf of bare motherboards

[Florent Calvayrac]

Jack Wathey wrote:
> 
> 
> On Mon, 27 Sep 2004, Andrew Piskorski wrote:
> 
>>> http://www.abo.fi/~physcomp/cluster/celeron.html
>>
>>
>> I recently bought a bunch of old ECS P4VXMS motherboards cheap on
>> Ebay, in order to build my own personal home cluster for testing and
>> experimentation.  I plan to mount them on metal shelves somewhat like
>> the Celeron cluster shown above.  However, that raises the following
>> questions:
>>
>> Normally, motherboards are mounted to the metal case using metal
>> standoffs, which touch the motherboard ONLY at the designed mounting
>> holes.  Is this in fact necessary?  Desirable?
>>
>> Is there any reason the standoffs need to be conductive?
>>


Inspired by the discussions on this list, we built in March 2003
a Beowulf of bare motherboards. You can see it (with text in French) at


http://perso.wanadoo.fr/sitepatrick/Station/index.html


Everything is made of metal, using ready-made beams to which we attached
the motherboards with integrated Broadcom NICs. The cluster (diskless, 
headless) has been running
fine since then. We have one power supply per node in order
to maximize cooling : design was made with CATIA V5R6, calculations were 
done by graduate students
in engineering using Flo Therm, and they predicted the temperature
inside the cluster within 1 degree C : 27 degrees at steady state
for 19 ambient.

The only detail we had forgotten was to make buttons for power and 
reset, since they are only to be found in ready-made cases ; we lost
hours to find the proper socket size on Radiospares.

The design took about 20 hours, the fabrication about one week,
software installation (having to find the proper driver for the NICs for 
clustermatic, setting up PXE) two days,  we saved maybe 30 %  of the 
total cost (here about 4000 euros for 8 Athlon 2400 nodes with 1 Gb 
RAM), had a lot of fun and learning.









-- 
Florent Calvayrac  | Tel : 02 43 83 26 26  | Fax : 02 43 83 35 18
Laboratoire de Physique de l'Etat Condense | UMR-CNRS 6087
Inst. de Recherche en Ingenierie Moleculaire et Matx Fonctionnels
FR CNRS 2575                | http://www.univ-lemans.fr/~fcalvay
Universite du Maine-Faculte des Sciences   | 72085 Le Mans Cedex 9
!!!!!!! NOUVEAU : OFFICE HOURS de 8h00 à 12h00 si present !!!!!!!!