If you look at the Abit page for its new product, the IC7-MAX3, it boasts of two new features.
This article will speak about the feature of more interest to overclocking; the OTES cooling system.
Is It Just A Matter of Cooling Chips?
For the moment, at least, it looks like the reason why Prescotts may not work too well in current 865/875 boards is due to Prescotts requiring 10-15% more power than had been anticipated.
The voltage regulators responsible for providing PC circuitry with the voltages it can use heat up as they handle more power. When they get too hot (and they can get hotter than boiling water), they stop working.
There’s a couple ways to address that problem.
You can use beefier voltage regulators built to handle more power. The problem with that is you still have a different version of the heat problem. The circuits getting hot may not care, but other parts of the computer do, most specifically the resin out of which you make mobos.
When you use components surface-mounted into the mobo, the resin has to stand much of the heat. Designing a mobo to handle big VRDs transforming a whole lot of power is expensive.
You can split up the burden. You can have more VRDs doing the work, so each has to do less. The problem with that is if you use six of the same capacity VRDs, it costs you more. If you try to save money using less beefy ones, you may not be much better off with six than with three.
What Abit is trying to do with the MAX3 is to actively cool the circuits so they can handle more power.
It will be interesting to see whether or not Abit beefed up the VRDs on the MAX3, or whether they’re using the same circuits and just chilling them a bit.
It also remains to be seen if this is a solution or a band-aid. Will this form of cooling be enough to let overclockers crank up Prescott as much as they like, or is it just enough to let the board get enough power for a Prescott, and little more?
Don’t know yet.
A New Limiting Factor?
Besides the CPU, and the memory, and the northbridge, the VRDs and associated circuits are going to become the next suspect in the eternal “Why can’t I overclock more?” This will be especially so in motherboards that use relatively few VRDs to transform power.
The emphasis is on the word “suspect.” It seems likely that cooler VRDs should help overclocking at some point, we just don’t know what that point is, or whether it will make any real difference in a real-life situation. For instance, if your CPU inherently can’t do X, cooler VRDs won’t help. They’ll only help if your CPU can’t do X because the VRDs are overheating.
Cooler VRDs may help now, but they may well not. It might well be the situation that these circuits work perfectly fine under current conditions, but will start blowing up with Prescott overclocking. It may make not the slightest bit of difference to these chips if they run at 50 or 100C, but start blowing up at 120C.
In short, this is virgin territory for overclockers.
Even if cooling VRDs doesn’t help you overclock more; it may turn out to be a very good idea for longevity. What we may find with current 865/875 boards is that they may run Prescotts, may be even overclock then fine for a while, then stop working.
It might even be possible that phenomena like Sudden Northwood Death Syndrome may have been caused in part by overtaxed transformer circuits providing damaging power to CPUs.
A Call For A New Product
Cooling these chips must do some good, otherwise Abit wouldn’t have bothered. The question is “How much good?”
At a time when there isn’t going to be much in the way of advances in CPU technology for some time to come, it may be a good idea for amateurs and cooling pros to start tinkering with cooling solutions for these circuits.
This will present some challenges. There’s no standardized layout for these chips, so cooling pros may find it tough to create “one-size-fits-all” solutions.
It may be possible that cooling the back end of the mobo may prove desirable or even necessary.
Water and electrocooling advocates face even bigger challenges. It seems like a flexible cooling block is really what’s called for, a block that can mold itself to the surface area in question, and again, the backend may need help, too.
The current Prescott situation appears to be a problem. Abit suggests a possible answer.
Those who tinker should explore the range of possibilities.