A couple days ago, to his general incredulity, Joe Citarella posted this link to an article about an Intel-designed water cooler.
It would not be at all surprising to find some Intel engineers moonlighting a bit to put together a watercooler. Many technological advances in the past (for just one example, flash memory) have been the result of such moonlighting.
However, from the article, this water-cooling project was actually a regular work, Intel approved project.
Why would Intel want to do such a thing? I can see two good reasons for it:
Faster CPUs: Intel isn’t interested in having overclockers run CPUs faster; they are very interested in selling faster CPUs, though.
After all, when it comes to overclocking, who determines what’s normal? Intel (or AMD), that’s who. If Intel finds that they can make a Conroe run 10-20% faster with a watercooler (and perhaps a touch more voltage), who’s going to stop them from selling one that way, like the next Extreme Edition chip?
Obviously, there will be some much reluctance on the part of some OEMs to go to water, but we now have the high-end PC OEM market. They ought to be far more willing to play guinea pig on this (and you ought to include Apple’s PowerMacs as a potential customer).
That alone would be worth doing, but the real benefit would come later on, after mass-production has knocked down the price of these units, in:
HTPCs: You don’t want a noisy HTPC. Water-cooling is quieter than air cooling. Build an idiot-proof, quiet and cheap liquid cooler, and you could sell millions of them in HTPC boxes.
And if you can finagle the deal to sell more of your CPUs, so much the better.
I suspect this project originated primarily as a way to cope with Prescott heat. However, as you can see, even if the original reason for such a project is passing away, there are other reasons to continue pursuing it.