Yesterday, we mentioned a thread in which someone was getting up to about 3.5GHz using high-end air.
Well, he’s doing better now.
Granted, he’s doing it with a lot of voltage, about 1.58V, which is about 30% more than default. I would not like to insure the life of that CPU. 🙂
On the other hand, not all the news is good. Intel’s Conroe EE will not be a 3.33GHz/1333MHz, but rather a 2.93GHz/1066MHz part.
Launch date has slipped a bit to late July (7/23, to be exact). Intel’s estimate as to the percentage of their desktop production that will be Conroes has also slipped a bit, from 40% to 35% in Q1 2007, which is not what Intel wants at all.
Conroe prices look like they’ll stay where there are until spring 2007.
How To Square The Circle
On the one hand, Intel’s lowering the clock, on the other hand, overclockers are raising it. What does that say about the state of Conroe?
Up to now, the standard explanation would be that the chips really can’t overclock that much, and any chips currently being used are somewhat cherry-picked, or at least being handled by people who really know what they’re doing.
We can’t dismiss that until more people get the CPU, but there may be a kinder explanation for both this contradiction, and some of AMD’s actions, too.
Power saving is becoming a mantra in the corporate world. AMD has been making a big push in the business-oriented magazines about how much power can be saved using their chips.
Intel is in a position to trump that with Conroe-generation processors, and to trump as much as possible, they may be willing to leave some performance on the table for the moment to keep the wattage down.
There may be other reasons for the pullback in XE frequency, most notably rock-solid stability at relatively low system voltages at a 1333MHz bus.
A while back, Intel also mumbled a bit about having power leakage licked at the 45nm level, and that some of that technology would eventually be incorporated into some of the latter 65nm chips.
Here’s what I think is happening: Intel will want to beat AMD up on every measurement they can the second half of the year. Let’s call them the three Ps: performance, price and power consumption.
They can do all three with their planned Conroe offerings, knowing that all AMD can’t do much of anything so long as they’re stuck with 90nm processors. There’s no free lunch for AMD; whatever they could do is going to hurt them one way or the other.
Make a faster processor? Bye, bye power consumption claims. Lower power? Lower performance. Lower prices? Forget about any profit momentum. Don’t lower prices? Forget about sales momentum.
Yes, there will be some 65nm out by the end of the year, but the low initially announced speeds of these processors indicate that AMD may feel power-constrained, too, and their power constraints are a bit tighter than Intel’s. However, we may find those AMD CPUs end up being more overclockable than expected, too, for much the same reasons as Conroes.
Eventually, AMD will come out with K8L, but by then, Intel will likely be able to grab another free lunch by incorporating some of that low-leaking power technology into the latter-day Conroes, and still hold a lead in the three Ps.
A Great Time For Overclockers
It’s too early to say for sure this is going to happen, but there are signs that for those not too finicky about power consumption (like us), CPUs are going to become more overclockable, simply because both CPU companies are going to leave performance on the table to keep the power down.
Not a displeasing prospect. 🙂