Ah, but BITS were probably not flat, so that ASP increase may not have covered increased costs.

$120M in increased headcount over one quarter? That would be a lot of people. Do you expect such an increased drag to carry forward, then?

Yuri/conductor

p.s. last post until midnight

If I may ask a personal question?

Are you long AMD stock currently?

You have seemed very negative over the past several weeks, while AMD stock has risen from the $11s back to today's (A/H) $14.00.

If you are long, cheer up! We are entering the historical "good" part of the year for tech stocks.

Yuri/conductor

Thank you for reiterating my point to wbmw. Perhaps your post will disabuse him of the notion that the doubling of ICG losses was due to non-flash ICG divisions.

Yuri/conductor

Yes, what you see is Henri Richard giving the answer, but Hector attempting to explain why the number is 10-12%, and not the recent historical average of +30%: Because AMD is more of an enterprise company, less pure consumer.

AMD was VERY confident about the CPG division during the call.

Yuri/conductor

You should point some things out to yourself. :)

Why do you think Dell forced Intel to introduce Nocona?

How can Intel not be losing x86 server share in 2004 if AMD has been gaining server share all year, expected to be at 10% in Q4? Magic? Wishful thinking on your part?

Merom, successor to Yonah (which is dual core Pentium M) is supposed to finally get AMD64.

They need dual core sooner, with AMD64, so they need to use, as Elmer from SI suggests, some Prescott 90nm MCM-like "dual-core" product in an attempt to limit x86 server market share loss in H2 05.

Yuri/conductor

Isn't it obvious? Intel gained flash marketshare through pricing cuts.

Yuri/conductor

Merom, successor to Yonah, is rumored to be Dothan64.

Yuri/conductor

No, wbmw said MORE THAN one year from tapeout has been the rule lately, and he considered Q1 06 to be a realistic date. Seeing as Yonah is 65nm, and 65nm won't be a volume process until 2006, it would seem he is right.

Yuri/conductor

Aren't there other rabbits to pull outta the hat?

Not on a short time schedule. Dothan lacks AMD64. They need something to try to hold server share. That's where dual core matters most.

Yuri/conductor

Yes, but the other business groups had almost no change in revenues, so it would seem likely that flash is the culprit, no?

Yuri/conductor

Here is an AMD/Intel CPU marketshare comparison.

This excludes the rest of the market.

 
       Q2                     Q3      
  
 AMD   554M  (8.79%)          673M   (10.20%) 
 INTC 5751M                  5928M 
  
 Tot  6305M                  6601M 
  
 

Yes, 65nm volume in 2006 suggests wbmw's estimate of Q1 06 for Yonah was accurate. But unless Intel plans something on 90nm for dual core, they will be significantly behind AMD.

Yuri/conductor

The only excuse for the bad GMs was the very slightly lowered inventory, which was one of the few positives.

But consider that the market seems to have been expecting even worse, so at this point, unless they say something dreadful in the CC, they should be up tomorrow.

Yuri/conductor

Not only that, ICG losses are terrible!

They nearly doubled from Q2 to Q3. Looks like those flash gains came at a terrible price:

 
 Intel Communications Group 
  Revenue                         1,327  1,271    
  Operating loss                   (251)  (126)    
 

It only revives questions about Ruiz's definition of "seasonal" for those who did not listen to the CC.

AMD defined seasonal growth explicitly for Q4 to be 10 to 12%, and rejected attempts to move this number higher.

Thus, they expect CPG revenues to grow more than 10 to 12% in Q4.
(And flash flat to up.)

Intel is guiding overall Q4 to be up 5% from Q3, which is very modest.

Yuri/conductor

No volume 65nm process from Intel until 2006.

This would fit with wbmw's estimated Q1 06 launch for Yonah.

conductor

Intel: A growth machine no more
By John Shinal
Last Update: 2:46 PM ET Oct. 12, 2004


SAN FRANCISCO (CBS.MW) -- Tech investors who haven't noticed that the growth torch has passed from middle-aged companies like Intel to Yahoo and other upstarts may want to take a close look at the firms' third-quarter earnings reports.

Both will report after the close of trading Tuesday, and the growth comparison will be stark.

Yahoo's (YHOO: news, chart, profile) quarterly revenue is expected to rise more than 80 percent from a year ago, while Intel will likely post annual growth in the high single digits.

Looking past this quarter is even more instructive.

Analysts expect Intel's 2004 sales to be $33.7 million. In 1999, near the height of the last growth cycle, the company's annual revenue was $29.4 billion.

So if Intel (INTC: news, chart, profile)hits its number this year, it will have posted a five-year average annual sales growth rate of - drum roll, please -- 3 percent.

Yet many on Wall Street - especially those who made big money on tech stalwarts like Intel, Cisco Systems (CSCO: news, chart, profile) and Sun Microsystems (SUNW: news, chart, profile) during the last bull market -still believe those companies will lead the next leg of the tech growth cycle.

But the long history of U.S. markets, like the recent history of Intel, doesn't support that thesis, according to one fund manager.

"In tech, it's all about the new supplanting the old," said Kenneth Broad, who co-manages $1.1 billion in the TransAmerica Premier Growth Opportunities Fund.

The key word in that quote is "old."

Although Intel is arguably one of the most successful innovators in U.S. corporate history, it began making semiconductors the same year Richard Nixon first won the White House.

So while Intel's business depends on being innovative, it is nevertheless an old - and mature -- business.

Some industries move slowly enough so that one firm can dominate for decades - think Coca Cola in retail. Yet tech firms tend to get put out to pasture at a relatively young age, Broad said.

One example he cited: Silicon Graphics (SGI: news, chart, profile), whose work stations were all the rage among graphic designers in the early 1990's, but is now a bit player. "If you look at the SGI stock chart, it's a 20-year round trip," Broad said.

Not that Intel is the same as Silicon Graphics. The firm has consistently used its emphasis on engineering innovation to stay on top of the market for computer chips.

For example, the company's push to make chips for laptops - which are more profitable than those used in desktop PC's - gave it a boost in recent years.

Yet its efforts to branch out into other markets - for communication and networking chips, for example, haven't met with nearly as much success.

So the company still gets the vast majority of its revenue from its core microprocessor business.

And that business has matured.

For that reason, growth investors will have a tough time finding a reason to own the stock.

Those who do may be thinking with their hearts, rather than their wallets.

"People tend to cling to the past," Broad said. With Intel, "they're looking in the rear view mirror."

http://www.marketwatch.com/news/print_story.asp?print=1&guid={66184585-771F-4133-82AD-00D123658C...

Perhaps he means the Accelerated Depreciation for captial equipment that expires on Dec 31, 2004. That is not also being extended, is it?

Yuri/conductor

Actually, Mobile Athlon is much stronger than Dothan in FP, so I think it is fair to say that on average Sempron has a similar IPC to Dothan.

Yuri/conductor

The idea is to compute about how many dies "sit" on the border of the wafer. Then, on average, half of the area of each border-sitter is inside the wafer, and so should have been subtracted from wafer area before the division of that wafer area by die area. Or the short-cut I used: subtract half the number of border-sitters from the die count.

Yuri/conductor

Smooth,

Please refrain from personal attacks.

Thank you,
Yuri

Mobile Low Power Sempron has a K8 core, and similar IPC to Dothan (better in floating point). Yes, Dothan 2MB cache will help in some applications more than others. But broadly speaking, they are quite comparable at the same clock. Also, Mobile Sempron has the NX feature.

Yuri/conductor

You may "feel" that way, but it is not so.

Yuri

If you simply count up the number of anti-AMD screeds and anti-Intel screeds authored by him over the past year, it will become very obvious to you.

Yuri/conductor

His point was that recent processors have taken LONGER THAN the 1 year (which put it at Q3/Q4 boundary), and that Q1 06 is more likely, especially as Intel is launching the Napa platform at that time.

It is not a "math mistake" that leads to 1-2 extra quarters. That was wbmw's opinion.

Nice try.

Intel has a way to pull things in when they want to.

LOL! You mean like Dothan, Prescott, and the 1.7GHz I2? Alviso? 4GHz Prescott?

You have a way of pulling our leg when you want to!

Yuri/conductor

Here is what the announcement said:

Acer, Medion, Twinhead, and many more OEMs and system builders are also expected to offer AMD Sempron processor-based notebook and desktop systems in the second half of this year.

Visiting Acer, we find this:

http://global.acer.com/products/notebook/as1360.htm

This however, appears to be the Mobile Sempron, not the Low Power Mobile Sempron.

Yuri/conductor

Ed is agnostic. He is into overclocking and at the best
price. I don't think he knocks AMD much, he's just not
a fanboy of either conmpany.

Nice try. :)

Yuri/conductor

Coppermine was troublesome, but not for as long a period, wouldn't you agree?

Yuri/conductor

A small lead in power consumption. The 25W TDP Mobile Semprons have been in production for a couple months now.

Yuri/conductor

Would not this portion apply? group = Germans, ethnicity = German

2. victimizes, harasses, degrades, or intimidates an individual or group of individuals on the basis of religion, gender, sexual orientation, race, ethnicity, age, or disability;

Yuri/conductor

It looks (to my eye, I admit) larger than that from wafer photos, but if you say so, then that is a correction one would make.

Yuri

Dothan has a small lead right now, although AMD is now producing 25W TDP mobile parts at 130nm, and is apparently working on second generation mobile designs (here I mean mostly process changes) at 90nm, beyond the low power Athlon 64s that have been recently launched on 90nm.

Yuri/conductor

While I am not a German national, I would think that such slurs are likely against the TOS. If not, they should be. They are certainly in very poor taste, and do not contribute to the quality of the forum.

Yuri/conductor

Let us do this from first principles, leaving in the yield factor.

1. 100mm radius wafer gives 31415mm2.
2. 90nm Sempron is 69mm2, but must account for inter-die spacing, so increase to something like 79mm2.
3. Divide, gives 398 die.
4. Edge effects (partial die are no good). (628/sqrt(79))*0.5 = about 35 bad die to subtract: 363 die.

Good 90nm 69mm2 Sempron die from 200mm diam wafer = 363 * yield.

Choose your overall yield factor, and you have an estimate.

With overall yield of 70%, you get close to your number. This is reasonable. Pete's number is way too high. It exceeds even 100% yield. He forgot inter-die spacing and wafer edge partial die loss.

Yuri/conductor

The 65nm technology node has the potential for making life difficult once again for AMD.

Given that AMD has managed with its 130nm node to make life difficult for Intel with its 90nm node, shouldn't we assume AMD's 90nm node will hold its own against Intel's 65nm node?

If not, why not?

Yuri/conductor

Semi, if you could note the message to which this is a reply... I would think this is an example of the sort of thing that you would find unacceptable on this board. If I could PM I would. Please delete this message as well.

Yuri/conductor

Ed is very biased against AMD, and has spent the last period of a year or so inventing excuses why AMD will not succeed. It can be said that his reasons have grown more creative as time has gone on!

It is probably not worth your time to consider anything he says about either AMD or Intel, unless you are seeking amusement.

He is widely mocked on all forums I follow.

Yuri/conductor

Low Power Mobile Semprons use considerably less power. 25W TDP.

http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_11600_11613,00.html

Worth seeking out. Or, a drop-in replacement for ~$107-$134, assuming you are comfortable with that, and that the BIOS (or a utility program) can correctly set the lower voltages at each speed-step.

Yuri/conductor

wbmw, 'livelock' is one of the ways a processor or process can hang. It's technically different from a 'deadlock', but from a user perspective, just as bad.

Yuri/conductor

I believe it said the bugs "will" be fixed with a BIOS update. Is that update available now? How does that update fix the bug? One way would be to disable NX functionality.

Yuri