A Second Word on Quadcore . . .

The other day, we pointed out that quad-cores were a waste
were a waste for just about everyone now and anytime soon.

Well, during the IDF, Intel did another captive benchmark series on its foursome, and could only find three real apps that got any real advantage from it (and two of them are really specialist, not mainstream apps).

For the vast majority of people, buying a quadcore shortly is like watching your child be born, then immediately ordering a PC for him/her. It’s not going to do him/her any good any time soon, and by the time it does, it will be outmoded.

Quad cores aren’t going to hit their stride until 45nm becomes old hat, which will probably mean late 2008/early 2009. It’s not the chips themselves so much as waiting for software to catch up, and unlike duallies, saying that quads will give you a smoother ride begins to sound faintly absurd. What are we striving for: a core for every app and utility you have?

Even if you’re antsy, waiting a while will have its benefits. Digitimes reports that the first 45nm quadcores will come out in 3Q 2007.

As the article points out, the Yorkfield chips will be designed as quad-cores, rather than two Conroes slapped together, plus support nice little things like PCI-E 2.0 and DDR3. It should also be pointed out that besides at least a little benefit from the process shrink, 45nm Intel chips should leak hardly any power at all. Both developments ought to make these chips rather faster than Kentsfields.

Even these chips may be somewhat ahead of the curve (I suspect the average reader will wait until the Nehalem generation before going 4-core), but wouldn’t you rather have a Yorkfield than a Kentsfield, especially since all you’re likely to do with a Kentsfield is run DivX, some synthetic benchmarks, and, if you’re lucky, maybe two games?

Your attitude about quads ought to be, “If I have to ask how one is going to help me, it’s not for me.” When software that’s critical to you comes out that takes serious advantage of this, then it will be time to look into this, and only then.

Ed


Be the first to comment

Leave a Reply