Spin Unspun

AMD has put out a good deal of nonsense about the Palomino delay, and unfortunately, the media has bought it hook, line and sinker.

I find it disgusting that AMD treats the media like clowns. I’m even more disgusted when they’re proven right.

“This Is A Great Move That Saves AMD Money and Gets The Most Out Of The TBird Investment”

You normally buy a new car every three years. That means that in six years, you buy two cars.

Let’s say you hold on to car number one for four years, and car number two for two years.

Have you saved any money over six years? No, you still bought two cars.

Did you get more for your money as a result? No, you got more use for your money (at the risk of a few extra breakdowns) from car number one, but less from car number two. You use car number one more, but car number two less.

This is just what AMD said it was going to do. It pushed back Palomino, but did NOT push back its successor chips. So instead of TBirds being around nine months, and Palominos twelve; TBirds will be around fifteen months, and Palominos six.

AMD still has to bear the cost of revamping twice. It sure will get more use out of any TBird equipment, but so long as those Thoroughbred/Clawhammer release dates stay the same, they’ll just get less use out of any Palomino equipment.

I cannot believe noticing that is genius. If it is, I need to find a better species to join. 🙂

An Unspoken Delay?

There is one way this approach could save AMD money.

If you decide not to buy two cars in six years, but rather buy two cars in seven years, then you do get more use for your money.

But that’s not the roadmap says. Not yet, anyway.

I’ve spoken in the past about AMD making a virtue out of necessity with the Palominos; I think they at least had problems getting them to run at high speeds.

For a number of reasons, I suspect those roadmap predictions of Clawhammer/Thoroughbred introduction are going to crumple like cards when the time approaches.

For one thing, the infrastructure for a next-generation mobo is not likely to be in place in time. Clawhammer will require
an entirely different type of motherboard. PC2100 is supposed to be replaced by PC2600, and there’s early engineering samples of PC3200 floating around. PC2600 has a different pin count than PC2100, that means yet new mobos for that even with Thoroughbred. Given the difficulties we’ve had getting PC2100 and mobos for it, it’s probably safe to assume delays there, and PC2100 doesn’t make a whole lot of sense for 2.5Ghz-3Ghz processors.

So even if AMD is ready on time, everybody else is unlikely to be.

Now this could have been really bad news for AMD in 2002 if Willy2 were a rip-snorting processor, but this is beginning to look unlikely. Some more L2 cache, and that’s pretty much it. A 2.5Ghz unimproved Willy2 (which we’re likely in see in early 2002) will still give AMD some uncomfortable moments in the Mhz race, but it looks manageable, especially if Intel insists on skyhigh pricing.

So either AMD is not being too truthful about saving money, or they’re being truthful about saving money but not truthful as to why.

Email Ed