liquid nitrogen cooling a possibility?

Bob Drzyzgula bob at drzyzgula.org
Tue Jun 12 09:54:22 PDT 2001


Given the speed of existing commodity systems, it seems to
me that in order to make sigificant improvements in system
performance, you'd soon run into limitations besides heat.
For example, signal integrity on the system motherboard,
quartz crystal behavior at low temperatures, and even
the lack of higher clock multiplier settings on the clock
generator chip or the commodity processors. Supercooling
processors might still be a viable approach, but I
imagine that this will most likely be true largely
for custom-designed high-performance systems such as
a mainframe or supercomputer, where one can afford to
do the engineering to make sure that *only* the CPU is
subjected to the cooling, and the AC characteristics of
the CPU-memory bus transmission lines can be carefully
controlled.

--Bob

On Tue, Jun 12, 2001 at 05:16:50PM +0200, Frank Joerdens wrote:
> On Tue, Jun 12, 2001 at 10:41:52AM -0400, Velocet wrote:
> [ . . . ]
> > For the money it may well be cheaper to just buy more nodes, wouldn't you
> > think? Run them at a slightly slower than dangerous speed and you've got more
> > computing power - you can place these extra nodes in the floor space otherwise
> > occupied by the chiller for your complex LN system. :)
> 
> Most likely, yes. I was just curious and had only been able to find
> rather wacky experiments on various overclocker's pages. David's info:
> 
> --------------------------- snip ---------------------------
> In St Paul MN in the late 80s? was ETA that went bankrupt and Cray
> absorbed the employees.
> --------------------------- snap ---------------------------
> 
> is intriguing though, in that there appears to have been at least one
> serious attempt. Anyone got more info on this?
> 
> Thanks all,
> Frank
> 
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf




More information about the Beowulf mailing list