Next Generation CPU Refrigerators

Iddo Genuth writes "Researchers at Purdue University are developing a miniature refrigeration system, small enough to fit inside laptop computers. According to the researchers, the implementation of miniature refrigeration systems in computers can dramatically increase the amount of heat removed from the microchips, therefore boosting performance while simultaneously shrinking the size of computers."

Re:Hotter?

[treeves]

Yes, but heat flow != temperature.

Re:Revolutionary

[Anonymous Coward]

Yes, if you take advantage of the extra heat absorption by overclocking the CPU to run faster.

It costs something like $300/gallon

[localroger]

I remember a piece linked here where a couple of morons immersed a computer in the stuff and cooled it with liquid nitrogen, oblivious to the fact that liquid nitrogen was cold enough to freeze the stuff. I was thinking "one small room air conditioner..." Apparently the miniaturized and practical version of that is what TFA is, although I say that as conjecture since I haven't read TFA.

Re:Revolutionary

[Anonymous Coward]

So it's the overclocking (i.e., increasing the clock frequency) that makes your CPU run faster, not the fact that it's cooler, as the article implies. And some CPUs generate more heat than other CPUs with lower clock speeds, so that relationship isn't a linear one, either.

Also, most modern high-end CPUs can't be overclocked by much, regardless of how cold you make them. The problem isn't heat, the problem is how fast the transistors can switch while remaining in sync. Sure, if you buy a low-end CPU from a high-end "family", you can usually overclock it a lot, because it's basically identical to a high-end models (just set to a lower speed at the factory). But, again, that has nothing to do with temperature, and temperature itself does not have any influence on a CPU's performance.

Re:Condensation?

[Smidge204]

Only because they cool below the dew point - which, in turn, is dependent on the humidity levels.

People who build active cooling into their computers (for overclocking) typically insulate the chip(s) and cooling block to keep air-exposed surfaces at or above ambient temperatures for just that reason.

Also, even if it does produce condensation I'd say there's little reason to worry... just recycle the condensate to provide evaporative cooling on the (much hotter) heat sink side of the system.
=Smidge=

Re:Side Question???

[jaxtherat]

You mean mineral oil immersion?

linkage: http://www.engadget.com/2007/05/12/puget-custom-computers-mineral-oil-cooled-pc/ [engadget.com]

Re:Side Question???

[SQL Error]

I think he's probably thinking of Fluorinert [wikipedia.org], which was used to cool the Cray 2.

Actually yea...

[Kaeles]

You need to remember that 90% of laptop CPUs will automatically downclock themselves if they are overheating (or over a certain temperature threshold.) They also do this if the cpu is more idle.

Re:Nitrogen costs less than beer

[billcopc]

You're right, liquid nitrogen does not cost anywhere near $300/gallon, but the GP wasn't talking about nitrogen, they were talking about 3M Fluorinert, which does indeed cost an arm and a leg.

The problem with these fluids is they can't keep up with today's processors. Immersing a PC in a vat of mineral oil won't magically cool the damned thing. You still need to extract the heat from that big pool of sludge; natural convection just doesn't cut it anymore. In fact, the fluid acts kind of like an insulator, because it moves so slowly that heat builds up right on your processor. You'd need propellers to move the flooz around, probably pump it through some sort of radiator.

On the plus side, I could use my overclocked PCs to cook me some french fries for my poutine :)

Re:Hotter?

[0111 1110]

No they are called phase change systems. Much more expensive than water cooling.

Re:Revolutionary

[NeilTheStupidHead]

It's been my experience with many modern laptops that they tend to slow down when they near their shutdown temperatures. I don't know if it's a direct result of the heat or some software actually slowing the processor to try and generate less heat or something else, but when both my Dell and Toshiba get near their shutdown temperatures (somewhere around 90C) they both slow noticably. They both speed up again if their internal temperatures start to drop.

Re:Revolutionary

[Anonymous Coward]

If a chip can be kept cool, all things being equal,

All things being equal, it will run at exactly the same speed. What makes a chip run faster is increasing its clock speed, not keeping it cooler. In other words, it's other things being different, not all things being equal.

A chip that doesn't overheat can (sometimes) be made to run faster by increasing its clock speed (which often also requires an increase in voltage, to maintain stability). No question about that. But lowering a CPU's temperature does not make it run faster. Understood?

The GP is absolutely correct, and this is a common misconception. It was probably meant as a joke, but I know some people who have actually put their system right in front of the AC, and swore it ran faster (no, they didn't overclock it), simply because it reported a slightly lower temperature.

Re:Hotter?

[drinkypoo]

We already deal with this with heat pipes. This is just using a heat pump instead of a heat pipe, active rather than passive. The goal is to move the heat away from the CPU core, to someplace you have more room for a heat sink and fan (like way over at a corner of the system instead of the middle of the motherboard.)