Smallpops, thanks for the response.
Re: The only possible advantage Hammer has over Itanium is Hypertransport which *might* allow it to scale better/easier.
The more obvious advantage is x86 compatibility. Expect AMD to capitalize on that most of all. Hypertransport is an obvious and eventual bus architecture, and I expect Intel to follow them soon with a unidirectional source synchronous protocol of their own. In fact, I'm surprised they haven't already. The only reason I can think of is the huge infrastructure support they have for their current front side bus. But it's only a matter of time before that runs out of gas....
Re: Hey if you have time I would like to get your thoughts about Itanium and its relatively short pipeline, what frequencies do you think it will reach etc. Thanks.
I once saw a rumor (probably on the Inquirer) that Madison will reach 1.5GHz. While it's usually rather aggressive to expect 50% frequency gains in the beginning with only a process shrink, it's worth keeping in mind that Intel has already had a long time to optimize their .13u process, and I think 1.5GHz is a very realistic frequency for Madison. After that, it's hard to say. I would need to see Intel's 90nm process on Pentium 4 before I could make any kinds of guesses on how that process would affect Itanium. Long term, I expect Itanium to eventually end up with longer pipelines, but it will still fall short of Intel's desktop products. That's because Itanium can perform well without a lot of frequency, but at the same time, increasing frequency over the course of several designs is a trusted and viable way to increase overall performance.
Re: IBM and Dell
Frankly, I don't see either of these companies showing interest in Opteron until it has a chance to prove itself. IBM, the ultimate hedge, still only operates as their customers demand. If Opteron has no market presence, why would you expect the end user to demand it? The only way they would know about Hammer is through AMD's hype which they do so well, and through high profile designs, like the one they won at Sandia National Labs. AMD is hoping for a miracle, that some way or another, they would create instant brand recognition through aggressive advertising. The reason why Intel doesn't counter their claims publicly is because even bad publicity is publicity. Even FUD against AMD is the same as good advertising in the position they are in. For IBM or Dell to consider them, they have to create a brand that they don't have. Without IBM and Dell, they have to rely on hype and tricky marketing, which puts them in a tough position. They seem to be gaining, though, however slowly.
In the short term, don't expect too many tier-1 partners. Dell deals in reliable partners, and AMD could never supply their volumes. IBM, like I said, is known to hedge, but only when their customers demand it. HPQ is more likely to go with AMD, since they have learned to segment their product lines. Case and point would be their ability to deliver PA-RISC, Alpha, Itanium, and Xeon technology, all currently available in multiple configurations. When PA-RISC and Alpha goes away, I can't help but see a void. Maybe they can cover it all with Xeon and Itanium, or maybe they will want Opteron as well. After all, they use AMD extensively in their desktop lines, so who knows....
Re: Newisys
You're reading too much in between the lines. People leave their respective companies routinely to create new startups. Newisys has made a big deal about the experience that they are starting with, as well they should. But don't confuse that with the experience necessary to create a top to bottom business model that is sustainable. Even great engineering falls victim to management stupidity. Those starry-eyed ex-IBM and Dell employees are going to get a dose of reality once they take the plunge next year with actual products. Hopefully for them, AMD will have a product for them to deliver, unlike all the Taiwanese chipset and motherboard makers that they screwed by delaying the desktop Hammer chip once again.
wbmw
