What To Do About Prescott-64? . . . .

1) Well, there’s hardly any point in buying Prescott-32 now, is there?

2) There’s good reason at the moment to be underwhelmed by Prescott-64. Going to 64-bits hardly solves Prescott’s most pressing problem: heat. You don’t fix an overheated engine by widening the fuel lines.

3) Intel’s going to fix that, they say? Fine. Show me.

4) We don’t know a number of devilish details yet. One big one is “Will Prescott-64 be socket T only, or will it include socket 478, too?” I would not spend any money on socket 478 equipment assuming it will eventually include x86-64.

5) The next big matchup is going to be between Prescott-64 and 90nm Hammers. Unless AMD assassins killed your mother or something like that, I would not buy a Prescott-64 until I knew what was going on with the 90nm Hammers. If you put a gun to my head and made me choose between the two right now, I’d go with the Hammers. Then again, a lot can happen (or not happen, as in further 90nm delays) between now and this fall.

6) 2005 still looks like a much better time to buy a new computer and/or do a major upgrade at a reasonable price than 2004. Any socket T platform is likely to be overpriced for what you get for most if not all of 2004. Items like DDR-II memory will probably not hit mainstream levels until 2005; video cards probably won’t face that much of a pricing problem, but an extra generation of development from ATI/nVidia will probably make it easier to justify shelling out for a new one. Before I laid out big for a socket T system, I’d want BTX and Tejas compatibility from the system.

On the AMD side, what the Intel move will probably do is to devalue x86-64 as a feature. It’s hard to see how an overclocker isn’t going to end up with a much better deal from AMD in 2005, both from a performance and pricing standpoint, than they will in 2004.

7) Windows isn’t everything. Having a 64-bit Windows is essential but not sufficent. Given that the price/performance picture is likely to improve significantly as we go into 2005; it will be a waste of money to buy x86-64 until you can make fairly full use of it.

Ed

Be the first to comment

Leave a Reply