We wrote an article yesterday in which we said that if an XBit Labs article were correct, that it wouldn’t be a good idea to buy 2400+ and 2600+ immediately but rather wait until the October date purported to be when the revised TBred would come out.
XBit Labs has hardly been the only ones hinting at a 166MHz FSB and a resultant increase in the PR numbers; CNet has an article saying much the same thing, and there were articles written as late as Sunday indicating that at least the 2800+ would be a 333MHz FSB chip
Since then, there seems to be some information which makes the contentions in the article at the least questionable (and in all likelihood incorrect, at least for right now).
The primary piece of evidence are some benchmarking results from what appears to be an AMD processor “officially” rated at 2GHz with 1) a 133MHz bus and 2) the ability to be overclocked to almost 2.5GHz.
It’s not clear yet exactly what this processor is, or details like level of cooling. It may be what will come out two days from now. It might be the chips that were according to XBit Labs supposed to be out in October. It’s not inconceivable it might even be a Barton.
Somehow, I get the feeling we’re going to find out on Wednesday. 🙂
It’s probably safest to assume right now that this 2GHz CPU capable of 2.5 is likely to be an advance soldier of what will be announced two days from now, and it got 2.5GHz without special cooling measures.
Why Not the 2800?
The reason why I was inclined to accept XBit Labs explanation on the origins of the 2400+ and 2600+ and the reason why I have some questions about this 2GHz processor are the whereabouts of the 2800+.
It makes no sense for AMD to hold off on a 2800+ processor if it can deliver one now. No matter how you think Intel might respond, AMD will make more money putting out a 2800+ now rather than later, and money is something AMD could rather use now.
This 2GHz processor capable of 2.5GHz (if that is typical of what AMD can do now) would seem to be good enough to allow AMD to release a 2800+ now. There seems to be enough headroom for AMD to do so.
I see a few possibilities:
1) AMD is ready to surprise Intel with a 2800+, either right away, or shortly thereafter.
2) AMD needs extra time to get a 2800+ because the yields at that speed really aren’t good enough yet, and the CPU used was cherry-picked.
3) AMD needs some extra time to get a 166MHz FSB processor ready (it may have to be one capable of running at two separate multipliers for 133/166 MHz FSB). Or just wait a bit for mobo makers to update BIOSes.
4) What was being tested was a processor that won’t be available immediately.
I find items one and two the most likely, and item four the least likely.
Shooting Oneself In The Foot
Intel glides through its schedule with methodical regularity. It has no problem telling people well in advance what it’s going to do, and even its surprises almost always consist of moving events up a quarter.
Why can’t AMD? What does it gain from being so secretive?
Forget people like me and other little websites, that’s miniscule. There’s a much greater issue than that here.
Predictability is an important asset. If you pretty much know what your supplier is going to do, you can plan around that schedule.
That can be more important to the average developer or corporate buyer than a few extra percentage points of performance.
AMD, on the other hand, leaves the impression of an organization run by the seat of its pants, and the pants are worn by Secret Agent Man. Neither attribute increases confidence in the organization.
Look at the CNet article linked above; it’s full of “We may do this, we might do that. The X processor may run at this speed, it may not. Maybe we’ll go to 333, maybe just for Barton, maybe not at all.”
You read this and you think to yourself, “Don’t you know what you’re going to be doing a couple weeks from now?”
They need to hide secrets from Intel? Point one, hide what? Clawhammer? You put out technical papers available to a ton of CPU design engineers on the other side equipped with the most powerful circuit simulators around, and they’re not going to pretty much figure it out?
Point two, a sizable chunk of AMD employees are ex-Intel employees. I’m sure there’s a few moles among them.
Intel probably has a far better idea of what Clawhammer will do than any of its potential customers.
I could understand it if they were ready to spring bombshells to stun Intel, but what bombshells has AMD ever thrown? Outside of a couple precedent-breaking price cuts a couple years ago, I can’t think of any secret AMD kept which proved to be a positive bombshell (I can think of a few negative ones, though).
Even a rapid 2800+ introduction really would help AMD more than stagger Intel.
Let’s not even talk about pricing games.
AMD spends a lot of time and effort keeping secrets that don’t seem to be worth keeping. The last year or so, it seems like AMD is shadow boxing against Intel, and every once in a while, Intel notices and gives them a real wallop.
I think AMD loses a lot more due to its secrecy and lack of predictability (especially in the corporate market) than it gains from any competitive advantage.
Now if you think I’m wrong, fine, but then please tell a specific instance where secrecy helped AMD.