[Beowulf] /. Cooler room or cooler servers?

Robert G. Brown rgb at phy.duke.edu
Fri Apr 8 04:00:30 PDT 2005

On Thu, 7 Apr 2005, David Mathog wrote:

> > From: "Robert G. Brown" <rgb at phy.duke.edu>
> > > Thats dual 32bit xeons, the dual em64t xeon systems are more like 420W.
> > > 
> > > Michael
> > 
> > Wow, that's Hot.  That's actually getting up there to the old alpha
> > boxen, per CPU.  So 40 1U systems are order of 16 KW/rack?  Five tons of
> > AC per rack?  $400/year/box for power and cooling (or thereabouts)?
> > Oooo.
> Convert 16KW to horsepower and find 16000 * .00134  => 214 hp.
> The point being that car engines generate this much heat in an
> even smaller space (1/3 to 1/4 of a rack volume, including the
> block and not just the cylinders).  Automobile engineers resolve
> this issue by dumping the heat as fast as they
> possibly can.   It's interesting to compare and contrast the
> way a car motor is cooled with the way a computer is cooled.
> 1.  Car motors are cooled by ambient input air - I've never seen
> a car that prechills its cooling air.    Nor is the cooling air
> blowing over the radiator particularly clean.

Car motors are water cooled, sometimes water cooled with ethelyne glycol
added to improve thermal transfer.  The water is then recirculated into
a large heat exchange device and (yes ambient) air forced through it.
Water has all sorts of nifty properties -- large heat capacity (relative
to air), intermediate boiling point, LARGE latent heat of vaporization
-- that are not present in "just" air cooled engines such as lawnmowers,
that tend to be much more power-restricted.  The air they dump into is
constantly flow-renewed by motion -- they don't operate in one place at
high power (if they do they overheat).

Car motors also generate almost all their heat in a single localized
core of cylinders that run at a much higher temperature (at the
operational point where thermal efficiency and output energy are
generated in the cylinder), apropos your very interesting discussion of
thermodynamics.  Heat transfer being proportional to \Delta T (all
things being equal) they can thus transfer at a higher power.  Finally,
a significant fraction of the engine's waste heat is rejected in the

Car motors don't generally run on full bore, and if they do they don't
run for long as they often overwhelm they cooling capacity of the water
and force it to boil.  In summary, the engine's cooling system has a
significantly different design and set of operational requirements than
does a cluster with its many different points of heat generation and
static environment.

But I get the idea, and your point is well taken.

> Perhaps we should be looking forward to chimneys and/or cooling towers
> in future computer room designs?

I like cooling towers.  Give terrorists a pause -- "hey, is that cooling
tower a nuclear power plant or David's cluster...?"  "I dunno, bomb it



> Regards,
> 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

Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu

More information about the Beowulf mailing list