XBitLabs reports that AMD will be coming out with a cheap CPU codenamed Applebred on it.
Once upon a time, AMD planned on having a Duron successor called Appaloosa. They currently make Thoroughbreds, so I guess where the name comes from.
To Circumcize Or Not To Circumcize?
This is obviously going to based on the Thoroughbred core. The only question is, “Will it be gelded (by knife or design) or not?”
On the one hand, if they don’t, it’s hard to see how a 1800MHz Appalbred would be any worse than a Thornton 2000+; it probably would be better.
On the other hand, the XBitLabs article says that a 1.4GHz Applebred would beat a 2GHz Celeron. That claim would be solid with an effective renamed
ThoroughAppalbred. It would become more questionable with one that had a 128K or smaller L2 cache (though the performance dropoff would be less than with the PIV/Celeron circumcision, presuming a castrated Applebred retained the 128K L1 cache).
We’ll just have to see.
NOW You Drop PR???
The XBitLabs article says that these processors will be sold at their core frequencies. No PR. It also says that these processors are supposed to compete against Celerons.
This is insane.
People have bitched and moaned about PR since AMD brought it back, and maybe they bent it a little when they put out the 3000+ and 3200+.
Yet when Intel put out a processor with 1/4 the cache of a PIV which drop performance around 20-25%, and gave it the same performance “number” as a PIV, practically no one says anything about that.
Which is worse and more misleading? Only the conceptually challenged would pick AMD; those who think two pints of milk is more than one liter.
Yet just at the moment when PR could erase this nonsense, AMD drops it. So Joe Sixpack goes to Best Buy and sees a 1.4GHz Appalbred and a 2.0GHz Celeron for maybe $50 or $75 more. Guess what he is likely to pick.
I’d have no problem at all if AMD called a 1.4GHz Appalbreds a 2000C+ or something like that. If that’s the competition and comparison, and the people doing the comparison are especially likely to go just by the MHz tag, why suddenly handicap yourself 40%????
The $25 Processor
The offical AMD price for the 1400MHz Appalbred is supposed to be $32. Given what we know about official AMD pricing, we’ll see $25 sooner rather than later.
Even if that $32 figure holds, what does that do to Via? You look at the other things AMD is doing, and you have to wonder if Via isn’t really the target. I don’t care how much AMD castrates this new horse, it’s going to blow the doors off anything Via has, and cost less, too.
It can’t quite compete purely on cheapness until we see a nForce2 board with built-in video for about $75, but that shouldn’t be too hard to do.
It would probably be rougher to build a car stereo computer with one, but a very underclocked Appalbred could form the basis of that.
Then again, if you actually wanted to . . like . . . play a game with a computer, getting a decent-enough CPU and motherboard and 256Mb of decent RAM for maybe $150 leaves more money for a video card.
And this $150 wonder, even unoverclocked can give you most of the performance of the latest whiz-bang system for about 25% of the cost, and “most” is “enough” for the vast majority of computer users.
Bet that’s going to sell a lot of Athlon64 systems. 🙂
Not Exactly A Money Maker
Estimates of what it costs to make a CPU these days hovers around $20 a pop. Even if AMD can do it for somewhat less than that, how much money can AMD, and the distributor and the reseller make from a $30 chip?
At these dollar levels, a success is more dangerous than a failure. This is a success than could cause failure.
More importantly, though, it’s a sign of where the mainstream computer market is going.
Revenge Against The Nerds
The CPU industry isn’t quite ready yet (even if they wanted to, which they certainly don’t) to seriously sustain a world where a $30-40 processor is the norm. The financial statements show how badly AMD does when $60 is the norm.
The technology and economics will come within a generation or two. The bigger challenge will be changing the computer paradigm. Eventually, though, it’s going to come to that once people decide they don’t need “more” computing power. Instead, they’ll want “enough” for less. That day is fast arriving, and news items like $25 processors and car-stereo computers are signs of it.
For the average person, computers will look, act, and be priced like a consumer electronics good. The typical computer eight-ten years from now probably won’t be much more powerful than today’s beast, but it will be much more reliable and much easier to use than today’s model, and will be more than good enough for Joe Sixpack.
This change is what can, and probably in the long run will wreck the likes of an Intel and a Microsoft; it’s hard to see them changing with the new times. The problem for AMD pursuing such a strategy is that it will wreck them long before Intel, and it’s hard to see who would buy them in the short-term and back such a strategy. It’s not in IBM’s interest to create a “stack them high and sell them cheap” CPU market.
That’s what Joe really wants, and that’s what he’s going to end up getting from somebody. It will be Revenge Against The Nerds.
Those who want more will still be able to do so; but they’ll be back to paying a few thousand a pop for it. The geeks will end up pay for all the R&D and fab costs for the hot goodies.