Purdue Miniature Cooling System Revealed

H

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
There is more information on that miniature refrigeration system we showed you last week from the gang at Purdue. This article rehashes a lot of what was said last week but there are pictures this time and a link to Purdue's research center.
 
I have a couple of problems with the whole concept:

They say that computers are only going to get hotter, but that is not where current chipsets are going. They are getting smaller, more energy efficient, and produce less heat. Many people thought that a super increase in heat would happen, but that has seemed to turn the other way in recent years.

Wouldn't these tiny refrigerators require a lot of energy? Moreso than, say, a fan? And with the improvements in passive cooling in the last few years (coupled with the fact that chipsets are being designed cooler) wouldn't this be a little backwards? We don't need more things in our laptops running down our batteries.

That's just the tip of things, I guess. I could be very wrong. But I have always been more of a fan of passive cooling and reducing energy use. That's the real future of computing, IMO.
 
Core clock speeds haven't really gone up in a few years due to thermal restrictions with current processor technology. We're adding more cores, upgrading cache sizes, faster memory, increasing integer range, basically anything but clock speed. In order to go higher, we're going to need beefier cooling solutions..until we get something like optical processors, but that's ways off.
 
When the entire compressor and condenser assembly can be packaged on chip, then this will not only be technically and economically viable, but open the door for denser chips. The best way to engineer cost out and increase reliability is through this type of integration - eliminate plumbing, couplers, and fragile external components by fitting the micro-channel evaporator, diaphragm compressors, and condenser all in one hermetic package.

Think of it as an active IHS ;)

Anyway, Total Design Power may be getting better as process technology makes transistors smaller, but the "hot spots" only get worse. Say for example you go from 45nm to 22nm on a die shrink, and the TDP of the part drops from 80 Watts to 40. Except now the rectangular die has 1/4 the surface area (it shrinks in both the X and Y axis), so you have twice the watts per square mm to dissipate - hence the need for active cooling as density increases.

A chip with integrated active cooling would allow you to maintain a reasonable die temperature, perhaps taking a 5-10% overhead per watt actively displaced, into a heat sink pressed up against the active IHS - just like they do today. There has to be an interface to air at some point, either with fins and a fan, or the case of your laptop.
 
They say it's for laptops... but they dont' say how they are planning to get the heat out of the laptop. Yes you've got it off the chip, but just like watercooling you're gonnna have to exchange it to the air at some point.
 
You must log in or register to reply here.
Back
Top