8.16.2007

OEMs Ignoring Barcelona

OEMs have finally validated Barcelona and they don't like it. The rumours have been spreading and now it seems like the cat is out of the bag.


Advanced Micro Devices Inc. (nyse: AMD - news - people ), meanwhile, lost more than 5 percent to hit a 52-week low. Credit Suisse analyst John W. Pitzer, who took over coverage of AMD from another analyst, kept an "Underweight" rating on the stock.

AMD, Pitzer said, is receiving limited initial support for its new processors from large original equipment manufacturers, "due in part to lower than expected performance metrics and the delayed ramp of high-end chips."


If Barcelona's performance appears "lower than expected", it should be pointed out that half of the inflated expectations came from AMD's excessively exuberant Executives while the other half is due to unrealistic goals and poor execution.

The implications of having limited OEM support are massive and will be damaging to AMD in so many levels. The most critical of all is the scaling back of AMD offerings in the server space due to the impending decline of AMD-based server demand. OEMs aren't quite keen on introducing a new CPU with minute volume and sub par performance and yet is capable of "Osbourning" their existing K8-based SKUs.

Moore's law states that the number of transistors in a semiconductor doubles every 18 months. The economic equivalent means that the price of performance is also halved every 18 months. Barcelona's performance at 2.0Ghz severely lags behind Moore's law which is why it shouldn't really surprised anybody when OEMs are finding no value in it.

19 comments:

Anonymous said...

Let’s get the full picture here, for the holiday season, of course.

The ultimate nightmare scenario of a monster chainsaw wielding Intel at the front door, and a crazed Borg like Nvidia at the back door, incessantly beating and grinding away, while the AMD minions hide in the dark corners of the basement, starts in November.

Resistance IS futile!

One plus

http://www.fudzilla.com/index.php?option=com_content&task=view&id=2463&Itemid=1

One

http://www.xbitlabs.com/news/cpu/display/20070815102710.html


Equals TWO

http://www.xbitlabs.com/news/chipsets/display/20070525075506.html


Does this give a new meaning to the term ‘Fusion’? Are there any conspiracy theorists out there?

SPARKS

Anonymous said...

I'm confused, Scientia has already stated that K10 will be released at ~2.3GHz and he's been right so many times before....you're telling me now it is only 2.0GHz and it may not be the be all, end all?

"There are some pretty good indications that AMD could launch 100Mhz higher with 2.4Ghz instead of 2.3 and then go to 2.6Ghz in Q4." (Good indications = Fudzilla and Inquirer speculation)

"It is my guess that AMD will give the expected Q4 and Q1 K10 speeds when it gives the Q2 Earnings report later this month." (Must have missed that during their earnings report)

"A full set of benchmarks should show this more accurately when K10 is released in August" (still have a few weeks on this one)

"This year, it looks like AMD is pushing the meeting back one and a half months and “will hold its 2007 Technology Analyst Day on Thursday July 26th, 2007”. The push to a later date probably has to do with the fact that DTX and mini-DTX motherboards are due in August as much as the time frame for K10. July also makes sense because FAB 30 should be into the 300mm conversion and FAB 36 should be working on 45nm, and AMD will surely want to mention them."

"The most logical assumption remains that the delay is caused by bugs in the die that require BIOS patches to run." (referring to Barcy delay, I think he meant to say "the most logical assumption, AFTER CONSUMING 12 BEERs AND PUTTING ON MY AMD ROSE COLORED GLASSES, remains that...")

"So, AMD is just going to have to bite the bullet until Q3 when things should improve. The chipset and graphic sales should be up by then and AMD should be fully anchored on the desktop with mini-DTX and DTX."
(So has DTX "anchored" AMD in Q3? I hear so much about this and see so many products...I guess the Q3 results should demonstrate this)

You know on second thought maybe I shouldn't put too much weight in Scientia's analysis. Of course AMD DID NOT STUFF THE CHANNEL IN Q1! :) This is pure myth that is now being supported by several financial analysts.

Anonymous said...

Scientia=turd
where you been??!
!50 platforms??!!!

I'm no expert in anything but that sounds kind of low for the second coming of bajeezus

Roborat, Ph.D said...

Driggers is also concerned that AMD has been stingy with releasing the various Barcelona models so that his company can test the processor with a combination of different customer applications. By contrast, Driggers said he has already received five different processor models for Intel's Caneland platform.

Ofcourse AMD is stingy. They have no choice when it's obvious that Barcelona is being forced to launch in it's very unready state. Usually you need another year to fix your problems when you're 4 speed grades down from your target.

"Customers had been very excited about Barcelona and now they are a little nervous, and they are looking to us to tell them more about it," Driggers said, adding that he believes that since Barcelona can plug into existing dual-core sockets, the company might hold off releasing the chip for as long as possible.

Catch-22 there. AMD has no CPUs to give out while the vendors are expected to validate and give feedback to AMD. That's a bit difficult when PowerPoint slides are the only ones shipped so far.

Anonymous said...

"AMD officials have said that as additional Barcelona models roll out, the clock speed will increase with every release."

Am I missing something here? Isn't this stating that the sky is blue and I have a hunch that if the sun comes up it will be from the east?

Here I was thinking as additional Barcelona models role out that the clock speed would go lower, or just stay the same, I'm so glad AMD clarified that. Folks may have been confused and thought that as additional models are rolled out they would have the same clockspeed as the precious and just be given different names...

And in other news AMD stated that as additional models of Barcelona roll out that will mean there will be more models than previous!

Anonymous said...

Let's see AMD touts the some of the advantages of Barcy like the ability to split power planes...

They also tout the socket compatibility and the drop-in capability into existing server systems...

Did you hear that?

Shhh... did you hear that?


That was AMD whispering that if you do choose to drop-in the new chip you may not actually get the some of the benefits they are talking up. Ahh.. details...who needs to be bothered with details...certainly not the press... oooh look! A shiny new chip!

Anonymous said...

It's rather humorous how Scientia attempts to appear unbiased but his assumptions and comments are so ridiculously slanted.

The latest blog:
"AMD will make more money as its margins improve with cost savings from 65nm."

Coming from a man who clearly doesn't have Si background --> 65nm will increase the cost of the wafer by ~10-15% (if done well).

Well this is offset by getting twice as many chips you say no?

Well actually not quite - the dirty little secret is that you do not get double density (although in theory you should) on K8, AMD was able to scale ~30-35%.

Still - that more than offsets the 10-15% processing cost increase no?

Well sort of - assuming the die size wasn't changing... any thoughts on how a 65nm "native" quad core will look vs a 90nm dual? (even a 65nm K10 dual will be somewhat larger due to cache, extra transistors in K10 logic, etc..) And as far as I know all of those K8 Opterons aren't being made on 65nm now or ever!

Oh and yield considerations anyone?

Conclusion - don;t expect miraculous drops in unit cost (especially not on the scale that ASP's have been dropping)

I like also how he cherry picks the desktop to do his analysis - if you look at the relative ASP drops in sever and mobile you will see AMD had a MUCH LARGER drop than Intel.

"Otherwise I would expect Intel to pull its desktop ASP back up to its preferred level of $99 and AMD to increase its to a preferred level of $70. This will likely be dependent on Intel's flash spin-off not being a $300 Million a quarter drain"

This just goes to show what a financial IDIOT Scientia is (don't mean to personally attack but that is the only thing that comes to mind). He is telling his readers that the average selling price of a desktop chip is somehow dependent on the flash spinoff profitability?

Does he mean gross margin? Or does he really think Intel sets the price of it's desktop chips based on how much it think the flash biz will take in? Is he really that dumb? I think he must be thinking about overall gross margins?

"Presumably this will become a genuine battle in 2008 as Intel ramps Penryn while AMD ramps K10."

Funny, wasn't 2nd half of 2007 supposed to be the battle? Can anyone say LOWERED EXPECTATIONS!

"AMD's average volume share in 2006 was 23% and even though Intel has been fighting hard it remains at 23% in Q2"

AMD has INCURRED MASSIVE LOSSES and screwed the channel over to maintain this - also note how he quotes the average 2006 instead of looking at the end of 2006 (if he used that it would show AMD declining) - at least use average 2007 to compare to average 2006. Of course that would include some really low Q1 #'s and he thus would not be able to come to his conclusion. The process here is make conclusion and find # to support it (and compare apples to oranges if necessary)

"AMD has had a desktop ASP drop of 42% since Q1 06"

Funny in the same period AMD went from profitability to loss, even though they were getting 300mm benefits and now some 65nm benefits. Want to know why?

The big 300mm MISREPRESNTATION - while you can get 2X (or 2.25X) the production volume, you are not cutting the costs by that much. 300mm has generally been modeled to give an ~30% cost reduction. So while ASP went down by ~40%, costs did not go down that much even with 300mm and some 65nm in the mix. Also take a look at MOBILE and SERVER ASP (which he conveniently omits) which dropped EVEN MORE.

"Today, AMD has previewed native quad core 3.0Ghz K10 which looks great and once again it is a matter of waiting for AMD to get them out the door in volume."

'looks great' - sorry I must have failed to see the "great" benchmarks AMD showed when they demo'd this - oh that's right no performance was given!

"In 2006, Intel was recovering from revenue shocks caused by AMD's K8. Today AMD is recovering from revenue shocks caused by Intel's C2D."

To expand this simalarity - in 2006, Intel was profitable during these revenue shocks and AMD is... uh...oh they are issuing ~3.7Bil in notes...oops... scratch that similarity.

What a biased, wishful thinking article... If AMD loses money the response is things will get better or they lost less than expected. Don't like the market share #'s - find the right time period. Or cost will drop by that much 0 forget the fact that chips have more transistors so aggregate die size for new chips is larger on the same node.

Don't like the competitive situation - assume a non existent part (3GHz, K10) "looks great and compare it to a part that already has been demo'd (Penryn) and of course assume no performance gain in that over the next 12-18 months?

It's actually starting to get a little sad watching him play AMD cheerleader / apologist... next thing you know he'll be linking to Newegg reviews and claim AMD is selling like hotcakes (oops - that's the other guy

Anonymous said...

Feb'07:
"So, 2007 is now shaping up to be pretty much even in terms of performance between Intel and AMD. And, 2008 shouldn't be much different since IBM/AMD also have high-K and will use it on 45nm just as Intel will. The talk about what Intel would offer in 2007 looked like giants in the mist. And, now that the mist is clearing we can see that Intel's processors don't stand any taller than AMD's."

Aug 07:
"Mostly this means that Q4 07 will be more of a skirmish than the major battle that was expected."


Feb07:
"Intel's amazing 45nm speed has completely vanished. The highest clocking Yorkfield is now only 2.4Ghz and Wolfdale tops out at 3.0Ghz. Worse still, even with these reduced speeds, 45nm has been pushed back to Q1 08. And, adding insult to injury, even Anandtech now agrees that AMD's Barcelona quad core will begin production in Q3 07 at 2.5Ghz."

Well, no need to comment on this one... Q1'08 - oops, 2.5GHz at initial production - oops

Feb07:
"Since K8 is currently hitting 3.0Ghz on 90nm one would assume that 3.2Ghz would be possible on 65nm. So, why is K10 only hitting 2.9Ghz?"

Well he was right about the 3.2GHz K8...oh that was 90nm not the "really good" 65nm process? I;m sorry K10 is hitting 2.9GHz (this is what happens when you rely on INQ and FUD for good new)

JAN 07: "FAB 36 is still ramping and as it ramps a greater volume of the production is on 300mm which does reduce cost. Likewise FAB 36 is converting to 65nm and this too reduces cost. So, by continuing with the same strategy AMD could expect to have better profits in Q1 07 although they may not quite match the Q3 06 numbers."

I pull these quotes out to show that Scientia often looks at a single data point in time, doesn't understand the fundamentals behind the data and then tries to extrapolate from that single data point. Costs are a classic example (300mm is cheaper but he doesn't know by how much, 65nm is cheaper but again no understanding of magnitude)

Process is the same - he though AMD would have high K in 2008, because IBM announced they would be using it on 45m. He had no background on how AMD would do this and keep the current 45nm plans - answer they can;t (which is why the VP of AMD manufacturing has said tail end of 45nm, meaning 2010! or 32nm). If he understood Si he would understand IBM's debelopment promises vs their ability to implement it in production (SiLK anyone?)

So the point here is anyone can take these random data points out of context and not understanding the fundamentals and just about come to any conclusion they want. This does not make you data based or objective, just dangerous and an amateur blogger who doesn't respond well to real (and often times valid) criticism.

Anonymous said...

"Perhaps because of the lack of gains by Intel there has been a flurry of negative rumors about AMD lately. This includes ideas that AMD is behind in 65nm, is having problems with 65nm, and won't deliver K8L on the desktop until Q4 07"

Yeah glad that rumor is being proven to be completely... accurate?

"AMD has contracts with one supercomputer that will be running by end of Q2 07 so K8L production has to start in Q2"

"Desktop versions of K8L will come out in Q3."

"The earliest that it looks like Intel could release CSI is early 2009."

"Further, Intel's gains have mostly been on the desktop which is the least profitable area." (server gains anyone?)

"Intel will have to be on its toes in 2008 to stay competitive and 45nm by itself won't be enough. Looking further ahead Intel proponents are quick to point to CSI but Intel will then be facing an AMD with two functional 300mm FABs"

(2nd functional 300mm fab = Chartered? 'cause it certainly isn't F30/38)

"It is likely to be faced with the prospect of either spending down its cash even more or watching its stock price plunge. From all of these disadvantages it is clear that Intel is not playing its own game."

Funny if you just replaced Intel with AMD, this statement might actually make sense now.

Anonymous said...

Well actually not quite - the dirty little secret is that you do not get double density (although in theory you should) on K8, AMD was able to scale ~30-35%.

It if was unable to scale explain this:
Atlhon X2 Approximate Die Size: 126 mm2 1MB L2
Pentium E2xxx Die size: 111 mm² 1MB L2

Both have 1MB L2 but the AMD have the IMC Intel doesn’t. Now explain poor scaling again.

Anonymous said...

What horrible blog.

Do you get at least paid by Intel to do this?

Anonymous said...

“What horrible blog.”

Isn’t though? Doesn’t it really suck that all these things are actually being said about AMD? Obviously, the industry insiders on this site have something to say, don’t they?

Further, you might be right, maybe Intel IS paying me! (If you call stock dividends payment, if you call share price increase, payment)

Here is what I think you should do, if you have any balls at all, or money for that matter, go out and invest heavily in AMD stock. Keep those blinders on and stay on the AMD websites. Light a few candles, say a few magical incantations chant, perhaps “Hector, Richard, Dirk”. Maybe you should try, “AMD is not the imitator”

Go Short on INTC. Take a second mortgage, a $100,000 should do.

DO NOT UNDER ANY CIRCUMSTANCES READ AMD’S FINANCIALS!

Come back to this site in 6 months, PLEASE, and tells us how wrong we were.

SPARKS

Anonymous said...

great blog roborat, keep up the good work

Roborat, Ph.D said...

anonymous poster said: What horrible blog.

AMD makes the news. I just provide my opinion. Anyway, thank you for your feedback.

I thought my recent lessons in Photoshop would make a positive impact. :"(

Anonymous said...

5 new Fabs--------$25 billion

Tic Tock Strategy----- 5.4 Billion Annually

Price War---------------47% margins


Startup Frequencies, 45nM @ 4 GHz -----Priceless

http://www.xbitlabs.com/news/cpu/display/20070817173238.html


SPARKS

Anonymous said...

"Both have 1MB L2 but the AMD have the IMC Intel doesn’t. Now explain poor scaling again."

OK... poor scaling is when you take THE SAME ARCHITECTURE and move it from one node to another node.

If you actually compare a 90nm Athlon to a 65nm Athlon, it doesn't shrink in die size by 50%.

Let me know if you need some more explanation.
Of course if you take two different architecture, with different transistor counts and try to draw a bogus comparison than I guess you can say silly things like you do!

Anonymous said...

Was reading Shari-kooks blog (it had been a while and needed a good laugh).

Oneexpert = Sharikou, correct? I noticed Sharikou stop commenting in the blogs and the crap tha oneexpert spews sound reminiscent (sp?) of Sharikou.

I'm just wondering when Sharikou will merge with Scientia...as they both seem to be degenerating into the same thing.

Anonymous said...

It has begun:

http://www.theinquirer.net/?article=41782

With the MS loan taken care of, expect to see things being sold off piece by piece.

Anonymous said...

It's very effortless to find out any topic on net as compared to books, as I found this post at this website.

Feel free to visit my web site :: diets that work