Last message on previous page:

I know that the architecture of the AMD Phenoms is better than Intel as it is a true Quad Core where Intel is a Dual Core doubled up, but Intels model just seems to work alot better.....

Ok so It is now said that there server market is where AMD should be, their processor to processor communication is better with hypertransport but so what... the "consumer" market buys more processors than the professional market as this is where AMD would get them some much needed cash.

Remember most people who buy machines at the office are home enthusiasts too.

Now we all know the Phenom has had more misses than hits, but surely these should of been sorted out well before it being released, this surely wasnt rushed out like Hellgate London for example on a processor that can litrally make or break a company.

AMD needed a cash donation which for a while and still has been bleeding fast from one of its jugulars.
AMD blame it down to the ATI aqusition, what makes me wonder is how AMD can make such mistakes..

All I can say it is down to the complete incompetence of the current management team....

It had taken over a year of Core 2 Duo to be available for the Phenom to be released.... Know and only now have there been technical issues with motherboards which dont work well with high end Phenoms and the lower ones dont make much more benefits than the unfortunate out going x2 processors...

Yes the technology maybe better but its how it is executed that makes the difference and AMD has certainly executed the release of this processor...

The best bit is to give us a tri core based on a fact that its a messed up quad core, I mean come on.. Who would want that really a duff cored chip..Whats stopping the third, or second core packing up... Mmm seems reliable

Will Deneb be any better, well its not that AMD can wow us with trust - is it, I mean you have to earn respect and not get lucky like they did with the Athlon 64's... And what they have done of as late is hardly been trustworthy.

Welcome to the digruntled AMD fan club! I think Baron will be joining us soon enough, his internet-wide silence seems to be a start. Sharikou will always be an AMD lover no matter what AMD does.


I just worry that we'll soon be mentioning AMD and VIA in the same subjects rather than AMD and Intel.

Well I disagree in part there.

Core2 is a wider issue cpu and has a much faster cache and prefetch system ... though it is little communication constrained beyond 2 core interconnects.

Though it might be bandwidth limited beyond 2 sockets it motors along in a single socket and dual socket environment ... almost oblivious to the restrictions of the fsb because of its superior prefetch / cache design.

AMD's Barcelona architecture is designed from the ground up as a multisocket server cpu ... It scales well but does not have the IPC of core2 ... even against the lower end models with smaller cache.

Therein lies the rub.

Add Intel's superior process technology (transistor design) that gives it the edge in speed and now power ...

Take AMD's die complexity ... 4 core real estate problems coupled with inferior process and manufacturing at 65nm ...

End result is where we are now.

AMD should not have thrown so much energy at one project ... too much is riding on it.

They should have done a dual core first ...

If they wanted to compete better in the single socket enthusiast market.

So I ask ... given the above ... did they intend too ??

Well their bragging prior to the launch suggests yes ... they did.

So either they were deluded or something went horribly wrong between tape out and final silicon ... the process most likely failed them.

But I could be dribbling ... it is late ... and I have been known to dribble after too much koolaid ... er ambrosia.

It will be all over bar the shouting if Nehalem is as good as the Koolaid AMD fed us before.

Unless some AMD engineer pulls a rabbit out of a rather worn hat ... or AMD corners the market on processed Hafnium ... heh heh.

Reynod, those are pretty fair statements.


Unfortunatly anytime AMD develops a new arch it is "all-in". If they don't have a winning hand, they're out of the game.

Intel is in a position that they don't have to go "all-in". Hell, look at the later days of P4. They made a mistake with those Preshott spaceheaters, but they have the capital that they were able to survive (and actually profit) during those times.

I just am so concerned about AMD's financial situation. I hope AMD shareholders understand what preferred stock VS. common stock is.

Up 15%? Thats what I heard, while everything else is tanking, maybe theres hope yet? maybe.... OK Ill wait til tommorrow when it tanks too

http://www.anandtech.com/weblog/showpost.aspx?i=443

Anand's been mulling over K10's performance beyond single socket ... blogging away.

Its a server chip ... thanks ... I think we figured that out reading the other review months ago.

nuf said.

I found this to be a very interesting graph. Not a roadmap that we typically see or discuss:

Yes, these defects likely still draw power causing the X3 chips to be less efficient and reducing the thermal headroom they have to work with. It is certainly possible that AMD will use more aggressive clock gating, rail shutdown techniques, etc. to get this under control. It could even happen in new stepping. Though, I would suspect AMD will be more focused on the newer cores coming out later this year as well as making a quicker transition to 45nm.

Actually, it is the professional market that brings in the most revenues.

If AMD had to sacrifice Phenom to get ATI, then that is their fault. So what they are really saying when they blame ATI is that they made a mistake. Whether that mistake was sheer incompetence or in part a bad gamble, I can't say. However, it is clear that AMD didn't anticipate or even plan for their current situation when they bought ATI. That, at the very least, can be blamed on management.

Precisely. AMD had a good technology and a good roadmap to the future. Unfortunately, they haven't done a very good job executing it (at least in the desktop market).

Deneb will probably be better. After all, Phenom is better than the X2's overall. Will it put AMD back on top in the desktop market? Not likely. Many of the changes seem geared, once again, towards the server/workstation market. It is possible that Deneb will make AMD more competitive in the Mid-High end desktop market, but I doubt they'll get to the High end unless Intel botches Nahalem.

Their main goal after the acquisition of ATI was Bulldozer. Everything between here and there are stepping stones. If they land on a bad stone, they'll just move on to the next one as quickly as possible. In other words, they'll try to fix the mistakes before the bulldozer launch, but I wouldn't expect them to spend any more time than necessary fixing the current chips (I.E. fewer steppings in favor of design progression). If they can execute bulldozer properly, though, then the gamble will have paid off.

Actually, it is the professional market that brings in the most revenues.

edit: I miss typed. The professional market has higher margins and thus higher net profit. Revenues are actually much higher in the desktop/notebook space.

All silicon wafers have defects; when a core doesn't work, it's generally because some transistors on that core were built on top of a silicon defect. So long as it's properly disabled and the rest of the chip is properly tested, there's really zero impact on the reliability of other cores on the chip; I suppose you could have some thermal issues if one quarter of the CPU isn't producing any heat while the rest is producting a lot, but I'm sure they've spent time evaluating those.

As mentioned above, some mid-range and low-end GPUs are actually faster GPUs which failed the production tests and had the defective units disabled; similarly, cheap RAM is often built from defective RAM chips with the defective parts disabled. Doesn't seem to have caused any significant reliability problems there.

We have only a few days before Intel has to come in with their email . Will there be anything incriminating? Will there be gaps? And if so, will those gaps in themselves be incriminating?

Or will the spaces between the gaps ... oh ... never mind.