Slightly faster Phenom processors will be delayed until the second quarter, according to mobo makers.
They blame it on the TLB. Is that true? Will B3 be the only answer to that, or is TLB cover for B3 problems? Who knows and, furthermore, who cares? AMD can fix the TLB fairly quickly, but they can’t fix the low performance or overclockability of these chips.
We view Phenom the way we viewed Prescott, and before that, Willamette: fatally flawed, don’t buy it. Eventually, Phenom may become a good idea for some, especially as an upgrade to current AMD systems, but not until the performance gets better and/or the price gets a lot lower.
Of course, the fatal flaw isn’t TLB. Lack of overclocked performance, especially for the buck, is the real reason, and one that will linger after the TLB is resolved.
TLB is just an extra kick, though I’m not too sure who is getting kicked.
At the least, it doesn’t look good when a flaw is big enough to stop mainstream sales of Barcelonas, enough to have a sometimes stunningly performance-killing BIOS fix installed by default in Phenom systems, but no big deal at all and certainly no reason for any enthusiast to think twice about buying one and turning off the fix. Seems like enthusiasts are being treated as some lesser species.
Do as I say, not as I do.
Is this FUD, like some are saying? Part of my job is to decrease FUD among my real employers, those who read this. No part of my job is to sell slow AMD processors for high prices. Which is more likely to cause FUD, buying a relatively slow processor now at a relatively high price with a known problem that you need to strip the fix from, or buying a (hopefully) faster processor at a (hopefully) lower price with no such problem at all later?
So much for that.
PS, AMD refused to decrease FUD by refusing to comment on whether or not R700 would be out in 2008. Given that AMD didn’t say a single word about R700 at the analyst conference, and had only a single slide mentioning R700 as a 2009 product, it’s pretty safe to say R700 (as in a real next-generation video, not some variation of R600s called an R700) is a 2009 product. The only reason why AMD ever keeps its cards close to its chest is because they have nothing in their hands.
How’s Intel doing?
There are reports that Penryns start acting funny when you get above 4.25GHz, but let’s face it, AMD would love to have that problem with Phenom.
PC Watch Japan reports that the Penryn Problem is actually a slight revision to get Yorkfields running on cheaper 4-layer mobos
The explanation fits the facts and answers questions like, “Why is the mystery problem not a problem with the expensive Penryns already out?” so if it isn’t the truth, at least it’s a much better untruth than we’ve been getting elsewhere.
Nonetheless, Intel PR should be ashamed of stonewalling a rather minor matter. You know, it really ought to be the purpose of PR to reduce FUD, not increase it.
Regardless of whether it’s Intel or AMD, the vast majority of the advance information that comes out comes from information these companies release to their partners.
If everybody else involved knows, why shouldn’t consumers?
Tags: Systems & Components