The more interesting article was linked at the bottom of the Otillini one: This article articulates a lot of what we have been saying here (in our discussions with Doug):
I love this one:
Analysis: One small step for Dell, one giant leap for AMD
tgdaily.com
[A lot more at the link. I'm posting the paragraphs that espouse the exact thing we have been saying here]
<snip>
Why Opteron, and why now?
Where Opteron has the advantage at present, analysts believe, is in multi-processor capability. Intel Xeon processors are somewhat competitive in the single- and dual-processor (1P and 2P) arena; but for customers that need to scale even further upward, not only is Opteron perceived as the clear leader, but it doesn't really have competition from Xeon.
"Even if you're in 1P, 2P today, if you ever perceive a need of going to [multiprocessing], you may want to stay with AMD just because you want that scalability."
Jim McGregor, In-Stat Microprocessor Report
[Wow. An analyst with a clue]
The hole in Intel's product line in the multi-processor field was translating directly into a hole in Dell's product line. This is where AMD seized the initiative, as the only supplier that could possibly fill the gap. "Finally there was enough pressure on Dell," explained Jim McGregor, "that Dell almost had to go to AMD, because there was a hole in their roadmap, especially in their high-end solutions.
"Obviously, Dell's getting market pressure from two areas: from the market, and some of their customers that are requesting these high-end systems," McGregor continued. "Intel's still got a hole in their roadmap, and AMD's had a kind of a lead in this market for quite some time. Intel is fighting back, but their products that they're coming in with in Q3, still don't really address [their key segment]." It could be next year, he implied, before Intel catches up.
But doesn't that catch-up process have a big "Woodcrest" title painted all over it? Not in the "MP" field, McGregor believes. "AMD today has had a performance lead in server processors," he told us. "Woodcrest is coming up here in Q3, and may make a more competitive situation on the 1P and 2P servers for x86 solutions, but AMD's still going to have a performance lead in multi-processor solutions. So the real question is, what are your processing needs, and how scalable are they? Even if you're in 1P, 2P today, if you ever perceive a need of going to that, you may want to stay with AMD just because you want that scalability."
[Me:Two comments: It may be obvious to folks that troll the "extreme performance" message boards- but the common concensus is that AMD has the clear performance lead and Intel lags. Even with conroe and Woodcrest actually being on top- it'll take a while for that perception to change. It's surprising how clueless the IT folks are.]
In other words, AMD's perceived lead in MP could give the company a significant advantage in 1P and 2P, just on account of the scalability factor. Customers often buy scalable systems, with the objective of leaving the door open for themselves to add more processing power later. Even if Intel's 1P and 2P Xeons hold their own in performance against Opterons, if they can't scale up, many customers may reject them...even if they end up never scaling up after all.
[Me: Wow! Could not have said it better. What, it is possible that a 20% woodcrest "advantage" may not immediately turn the tables on AMD immediately?]
Of course, Dell can only capitalize on AMD's scalability advantage if it also offers 1P and 2P Opteron-based systems. If it only plans to offer Opteron on MP servers, as some comments from yesterday's conference call suggested, then at least for the time being, HP may be perceived as the vendor with the broadest range of customer options. Giving HP the lead in a market that Dell just now cast a broader spotlight upon, may pressure Dell into offering a 1P or 2P Opteron-based server, either immediately or eventually.
"I think if you look at customer choices in enterprise spaces, [or] even in any technology space," explained Doug Freedman, "if you have a supplier that's shown an ability to produce a product that's superior - whether this generation is superior or the next generation - you want to have a relationship with them so you have access to that technology when and if it does end up turning out that the next generation is that way. In a way, they should have a relationship with AMD just because the past performance of their products was better."
"In a way, [Dell] should have a relationship with AMD just because the past performance of their products was better."
Doug Freedman, American Technology Research
In a best-case scenario for Intel, Freedman suggested, Woodcrest processors could close the 1P and 2P performance gap with Opteron. [Me: I guess that Freedman- even being a CPU analyst does not troll the extreme systems message boards all day long] That might trigger the next wave in the performance battle, with AMD introducing its own next-generation architecture later this summer. Again, weighting the scenario in favor of Intel, AMD would only gain a slight performance advantage over its predecessors. But if Dell didn't offer consumers that choice, it could still suffer, Freedman explained, "because they haven't had a relationship [with AMD], is always behind the eight ball, never able to ramp and introduce those products because they're just not part of the supply chain.
[Me: BANG! This is the bottom line why I (and others) feel that this game HAS CHANGED forever! AMD is now a tier 1 supplier in this market. Intel will never (in the near future) go back to monopoly status- thus, along with the loss of Mopo status will come a loss of multiple and further share loss as other suppliers also buy into this concept and rebalanc their sockets to also include AMD]
"I think what [Dell is] accepting is the fact that performance, in the future, could possibly flip-flop back and forth between the two," continued Freedman. "And that's very healthy for the market, and the consumer wins. If that's the case, you need to have a relationship with both, so that you choose the processor that works the best for the application that you're trying to serve."
<snip>
[From the 2nd page in the link/article]
In moving to more upscale markets, AMD made the right first choices, including aiming not for enterprise PCs but for servers, McGregor explained - and it was this choice that won AMD the Dell contract. "It's probably the smartest thing they did," he said, "because that puts them right in the face of IT managers. It's a lot easier to possibly get into the enterprise PC [market] once you've got the servers, than the other way around."
I'll say it again: The game has changed for ever. A 20% advantage here or there is no longer an issue- specially given that the Mhz race is over and now CPU suppliers are adding features to provide more value.
Welcome to the new era.
TG |
