I would have agreed with you until I started messing with that TBred. The 10-15% extra cooling that water provides is basically all that is keeping it under 60C. The heat DENSITY is going up up up, and THAT is an issue. Could be that silver inserts may become necessary; or jet on die cooling.
In the end, water cooling trades an additional junction (block to water then water to rad) for enormous surface area (can't fit the kinda surface area of a heatercore on a CPU). This is a good tradeoff as long as efficient heat transfer from die to block can take place. Looking at some of Bill's data, the TIM joint is often the limit.
The other thing about future cooling needs is that we need to focus on Intel more because AMD may not be around in any real sense. (I hate to say it, but they are in trouble)
Phase change is a very robust technology, but I don't see an affordable phase change solution being released any time soon. I CAN visualize a mass marketed water chiller though that would use existing water cooled parts to push performance limits. That doesn't marginalize water cooling though in any way.
In an environment where people spend $100 on LIGHTING for their cases, it is hard to imagine that water cooling is going to vanish. More and more systems put together for "show" rather than "go" are being seen. If manufacturers can accomodate this (and they seem to be), there will be a market.
Heat pipes are good for some apps, but they require CPU temps higher than I typically want to really work. I would be scared to see how hot my notebook's chip gets under load...
A rhetorical question: Has watercooling EVER been anything other than a hobbiest activity? People have always gone back to heatsinks when they are too busy to maintain a water-cooled system and stepped up to extreme cooling when they wanted bigger o/cs.
|