The Inquirer has an article about the cooling needs of a dual-core Montecito. Bottom line: you need a TEC to do it.
If you don’t know what a TEC is, it stands for ThermoElectric Cooler. If that doesn’t help, it’s a fancy name for a Peltier. If that doesn’t help either, a Peltier is a type of electrical refrigeration unit.
Overclockers have used TECs for a long time, well, a few of them. They’ve never been mainstream overclocking components because they’re rather more expensive and finicky to deal with than a heatsink/fan. Because they cool so much compared to a air system, condensation can gather around the cooling area, and if CPU insulation isn’t done right, well, CPUs and water don’t play well together.
Yes, it can be done, but it’s not a brain-dead operation, and it’s going to have to become one for this to hit the mainstream.
True, the article talks about a Montecito rather than a dual-core Hammer or PIV, but it’s really hard to see how a 90nm dual core processor, unless it’s running rather slowly, isn’t going to be in the general thermal neighborhood of the Montecito, especially since initial dual core designs will essentially just slap two single cores together and chew up about double the power of a single core.
Shrinking the process down to 65nm may help matters, and adopting the Dothan core (for Intel) may help more, but given the heat crisis in the CPU world, one’s attitude ought to be “Show Me.” This is especially so when both manufacturers are being rather coy about parameters like speed and (in the case of AMD) power consumption.
Another Split Conclusion
As has been happening so often lately, the interests of overclockers and Joe Sixpacks diverge yet again.
Joe Sixpack (and the OEM companies supplying him) aren’t going to want to hear anything about TECs until/unless they cost no more than a heatsink/fan, and more important, are just as “set it up and forget it” reliable as air coolers.
These are substantial challenges. So far, products on the market come close or meet one criteria or the other, not both. They either cost a lot, or are DIYers.
Overclockers are a little more flexible on this. They’ll pay somewhat more, and accept a bit more finickiness than, say Dell, but (generally) not a whole lot more. (Yes, there are DIYers who think Peltiers are a snap right now, but the average overclocker still uses a heatsink/fan, and that’s whom I’m talking about.)
If TECs can be made to replace current air cooling for roughly the same cost and hassle, all well and good for both Joe and us. Overclockers will want more powerful ones than Joe Sixpack, of course, but if TEC manufacturers can make Dell happy, this shouldn’t be too much of a problem.
If they can’t, though, the OEMs are likely to say, “You keep your dual-cores until you can cool them the old-fashioned way.” which ought to keep dual cores away from the desktop mainstream until some 65nm Dothan comes out (God knows what AMD would do under such circumstances; they don’t have anything like Dothan to fall back upon).
That’s no big deal for Joe, but it leaves most overclockers on a limb. If overclocking dual cores requires a significant step-up in investment/effort, a lot of overclockers are going to step out.
If overclocking becomes a hobby that requires $150 rather than less than $50 to start doing, there’s going to be fewer overclockers, at least until dual-cores come out that Joe Sixpack will get the chance to buy.
However this plays out, though, over the next couple years, cooling is going to go from being an afterthought to becoming one of the primary thoughts in computing.