chipguy, I don't have a problem with referring back to old news. What Windsock did was present a 5 month old headline as though it were current. That's deceitful, regardless who does it.
Paul

Petz, re: chipguy,Intel and power

I suspect what he means is there is a good bit of room for improvement and that they already have the ways avaiable to do it. Whether or not that's actually the case - who knows?
Paul

Windsock, Why quit at 5 months ago? Why don't you try digging back a whole year for an article like that? Hey, why don't we just go back to how slow Opteron's uptake was before Athlon was introduced? Gimme a break!
Paul

Mysef - hey, we're all human - live and learn! ;)
Paul

subzero - scoff while you can, that opportunity won't last. :)
re:expensive MB - huh?
Newegg has 15 Socket 754 MBs, 4 are under $100, only five are over $150. Try again.
http://www.newegg.com/app/ViewProduct.asp?catalog=22&DEPA=1&submit=property&mfrcode=0&am...
Paul

Keith, re: Nero - I somehow missed the free part when I read your post earlier. That's great news! Nero will help snag a new generation of users for A64 with this support. One more step... :)
Paul

Doug, sure a TA board would be fine, but when it comes down to it, you want to know how you're going to apply it to AMD, so it's a legit subject here.

re: Would you call the rest of the TA posts "reasoned analysis"?
It seems to me that the folks here involved w/ TA are trying to analyze a market that is less than obviously formulaic, by using TA which, if I'm understanding correctly, is based on patterns which are not based on the obvious physical fundamentals of the stock{P/E, revenues, product lineup, etc.), but on the "corporate psychology" of the market(corporate meaning as a unit, not referring to business entities). These patterns exist, I'm quite certain - chaos theory suggests that there are patterns in just about everything, no matter how random they may seem. sgolds, et al seem to me to be making an honest effort to logically analyze based on some TA theories they feel hold some water. Regardless whether I agree or not, I can respect their effort to make sense of it.
Paul

subzero - re: plenum - Wonders never cease! A post of yours I actually agree with!! ;)
Paul

yb- re: Do you have any idea why good boards are not selling good?
Probably ignorance(individuals) and OEMs cutting costs to the bone.
Paul

sgolds - They specify that they are FXs, but from the layout of the server, it looks like they may be clustering(is that the right term here?) two motherboards within one enclosure, or something similar. Very interesting approach.
Paul

Interesting new 4x server using A64FX-51s. Thanks to Jens on SI.
http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=19588681
http://www.v-t.jp/products/hpc/opteron/server/sv_top_main.html

Monarch Computer(they were an Opteron and A64 launch partner) say in their subscription e-mail that 15 Dec is the launch date for A64 3000+(They have a limited offering of them now for their e-mail subscribers). Received the e-mail Thursday night.

Charlie D. believes it as well.
http://www.theinquirer.net/?article=13153
Hmmm... he says that 3000+ is definitely 1.8GHz/1ML2. Wonder what that 2GHz/512K beastie really is.

Doug re: why can't I post my own analysis? Surely one of mine per every ten others isn't unreasonable. And what makes you think my analysis is any less valid than the others?
Because condescending, mocking insults are not what I would call reasoned analysis.

re: avoid them, if you like, instead of feeling the need to play editor-in-chief.
I've been very up front, fair and gracious toward you about this, it's too bad you couldn't handle this without resorting to more condescension.

My final word on this subject is this - although you and I may not agree with sgolds on the validity of TA, it is a valid subject to explore on a board involved with investing in AMD, just like exploring the "fundamentals" of AMD is. Just because you or I may be of the conviction that it's wrong doesn't necessarily make it so. Investing cannot be reduced to any single formula or even set of formulas or laws, because it deals with human nature. Insulting people who are trying to find a way to understand the complex a little better through analyzing patterns is not helpful to anyone on the board.

Doug, re: enough of what?
Enough of the condescending, belittling cracking posts on TA, like the one I was responding to. Once was funny, but after all that has gone on on this subject, it's getting old. Some people clearly have an interest in pursuing TA(I'm not one of them, BTW), and what you're doing is no better for the thread than the wild goose chases some of the Intel crowd instigate here - a waste of bandwidth and people's time. Realize, though, that I only mean in this specific topic; your other input to this thread is valuable and spurs useful discussion.
Sorry to reply so late, but I'd just run out of posts when you posted on this. <:(
Paul

Doug, enough already! Last remaining post - later all!
Paul

EP - actually, he doesn't have any hidden/ignored.

Doug, sorry, I should have stated the obvious there, shouldn't I?
Paul

Issue solution revealed.
(Boldfaced below)
Paul
http://www.realworldtech.com/forums/index.cfm?action=detail&PostNum=1901&Thread=27&entry....
Name: David Wang (dwang@realworldtech.com) 12/11/03

mas (mas769@hotmail.com) on 12/11/03 wrote:
---------------------------
>mas (mas769@hotmail.com) on 12/11/03 wrote:
>---------------------------
>>David Wang (dwang@realworldtech.com) on 12/10/03 wrote:
>>---------------------------
>>>mas (mas769@hotmail.com) on 12/10/03 wrote:
>>>---------------------------
>>>>"
>>>>AMD discloses that its yield problem on the 130nm SOI process was due to the integration
>>>>of Low-K materials into the process rather than SOI.
>>>>"
>>>>
>>>>Now that is the killer line that will get you hits and looky sees all around the Web, lol.
>>>
>>
>>http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=19584598
>>
>>told you. . In fact I think yours was a Web exclusive.

I find it surprsing that I didn't run into any from time EE Times. Perhaps budget constraints? There were some reporters, but mostly those interested in the business implication of technology side of things. One reporter was asking why the same conference would trumpet both low-K and high-K material. He apparently didn't realized that they were to be used in different places. :)

>>>AMD also talked about a few things like how they gradually modified the gate structure
>>>for SOI until it finally achieved the performance they were expecting. Apparently
>>>the transplanted gate structure performed rather sub-optimally.
>>>
>>>AMD had some nice slides, and I've asked to see if I can get them to include with
>>>the article. However, the presenter won't be back in Germany until after the 19th.
>>>I'll either have to wait and see if he'll send the slides or just write it up without the slides.
>>>
>>>According to the presenter, AMD was seeing quite a bit of defects from the edges
>>>of wafers, since the adhesion of the metal stack was sub-optimal. They were in fact
>>>flaking off, and causing more problems toward the center of the wafer.
>>>
>
>Forgot to ask how they cured their yield problems in the end and also sounds like
>the problems IBM have been having too with Low-K.

AMd said that it was a combination of three things.

Edge engineering, improved wafer handling, and altered deposition chamber (procedure?).

Basically the low -K material was flaking off, so it seems that the solution was to do some special work at the edges so they won't begin to come off, and handle the wafers rather gingerly. That was the gist of the handwaving I remembered.

The presenter had some nice slides. Hopefully they'll release them.

>Also did they mention when they were having the problem and when it was fixed ?

They didn't exactly say, but I'm guessing from the unlabelled defect reduction chart that it was 2 quarters ago.

>>Fascinating. It just shows that these guys are almost like explorers going into
>>unchartered territory where you never know what is lurking round the corner. You
>>and Paul have produced a fine body of work here, don't get disheartened, we all
>>read it and appreciate it even if we are not always vocal about it.

You should attend IEDM sometimes, it's pretty much of the leading edge of where computer engineering starts (IMHO), so there's a lot of chaotic development going in all sorts of directions. They're all pretty much attacking the same problems, but trying their own independent solutions. So it's all uncharted territory. (who's built a 4 Gbit Flash chip before, in the history of mankind? Who's built electro-mechanical cantilevers for data storage?)

Coupled that with the fact that process generations are now moving at 1 generation every 2 years, you see that multitudes of solutions to different problems are suggested every year, and they'll change rapidly in the year after.

PC Shipments to Reach Record Levels, IDC Reports

Online staff -- Electronic News, 12/11/2003

IDC updated its PC forecast today, projecting that on strong demand from the consumer segment, aggressive pricing and rapid portable adoption, worldwide PC growth will continue in Q4 and into 2004.
Strong Q3 results and projected growth of 15.3 percent in Q4 are expected to boost annual growth of worldwide PC shipments to 11.4 percent, an increase from prior IDC growth projections of 8.4 percent.

Total shipments in 2003 are now expected to outdo 152 million with a value of just more than $175 billion. That represents an 8.8 percent shipment growth over 2000, the previous highest year for PC shipments, but a 22 percent decline in total value will emerge due to aggressive pricing and a shift toward lower-end configurations as overall computing power increases.

http://www.reed-electronics.com/electronicnews/article/CA341234?spacedesc=news

EP - LOL! Why, you been sticking wires in electrical outlets again? It couldn't be shock at AMD's yields! ;)
Paul

CJ - re: At least they got it to work. Else they wouldn't be presenting a paper anywhere on the problems...
Good point.
PauL

130nm issues revealed, not yet fully defined; issues with low K, not SOI. IEDM report @ RWT. More to come Tuesday. Thanks to ixse.
http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=19584598
http://www.realworldtech.com/page.cfm?ArticleID=RWT120903021115
http://www.realworldtech.com/forums/index.cfm?action=detail&PostNum=1901&Thread=23&entry...

More Sun to come soon?
http://www.theinquirer.net/?article=13138
AMD64 is an Itanium killer, Scott McNealy says

Colin Barker does his Christmas shopping

By Mike Magee: Thursday 11 December 2003, 11:52
WE NOTICE THAT our old mucker Colin Barker is getting some Christmas shopping done in Berlin, and stopped buying those nice biscuits to have a chat with Scott McNealy, of Fun Microsystems fame.

Scott chatted to Col about a few things including AMD. Scott reckons that the Opteron is for two to eight way systems, and he's about to strike another chip deal early next year.

He describes the decision to strike a deal with AMD as a slam dunk, Sun took the Xeon Solaris binary and "it immediately ran on Opteron".

The Itanic, he claims, is "last millenium" architecture, and AMD is "an Itanium killer".
http://www.computing.co.uk/News/1151480

FASL ramping...
http://www.xbitlabs.com/news/memory/display/20031210172900.html
FASL Increases Output and Employs New Fabrication Process
AMD and Fujitsu See Growth in Demand for Flash

by Anton Shilov
12/10/2003 / 05:30 PM

FASL LLC – a joint-venture of AMD and Fujitsu – plans to ramp its manufacturing capacity for Spansion Flash memory products in 2004 as part of its long-term strategy to meet wireless market demand from AMD and Fujitsu customers.

By the end of 2004, 110nm 128Mb equivalent capacities is planned to exceed 80% of the current output of Fab 25 and JV3, the company’s two leading-edge Flash memory fabs. The 110nm 128Mbit equivalent capacity is planned to be over 60% of total fab capacity output.

“Wireless solutions from our leading-edge customers demand high-fab capacity, with leading-edge Flash technologies, supporting state-of-the-art performance,” explained Amir Mashkoori, Group Vice President and General Manager of the Wireless Business Unit for FASL LLC.

110nm MirrorBit technology is designed to support a major extension to the Spansion Flash memory portfolio with the addition of a high-performance, fully-featured, 256Mb 1.8V product for the wireless market. The planned feature set includes greater than 80MHz burst mode, fast access times and low energy consumption. Early silicon on this product, the Spansion S29WS256N device, is already in-house, with planned sampling in the first half of 2004.

“The far-sighted decision taken back in 2001 to convert Fab 25 from a logic fab to a dedicated Flash memory Fab is validated by the ramp in output from this Fab and the solid execution supporting 170nm in 2002, 130nm today and 110nm technology in 2004,” said Bertrand Cambou, CEO and President of FASL LLC.

As a Flash facility, Fab 25 is on track to deliver greater than 30% higher wafer output in 2004 compared to its output as a logic fab, at greater than 40% lower wafer cost. The 128Mb equivalent output on 110nm and 130nm from Fab 25 is planned to grow 8-fold by end of year 2004 compared to the last quarter.

“The inclusion of Fab 25 in FASL LLC together with JV1, JV2 and JV3 doubles our total capacity through 2004, establishing a manufacturing powerhouse in Flash memory in support of our customers’ business growth,” Cambou added.

In general, the move means there is growing demand for flash memory used in numerous consumer devices, including cell phones and PDAs. Additionally, bolstering manufacturing capacities is likely to positively affect FASL's market share.

Just in case anyone is snickering at emachines and the T6000...

http://www.businessweek.com/magazine/content/03_46/b3858106_mz063.htm
eMachines
How did Wayne Inouye turn a reviled, money-losing PC maker into a winner?

When Wayne R. Inouye took over as CEO of faltering consumer-PC maker eMachines Inc. in March, 2001, he figured that to understand how to fix the company, he needed to answer calls from irate customers personally. One of his first calls came from a volunteer at a church that had purchased six eMachines. The desktop PCs -- which sold for under $800 apiece -- were fantastic bargains, but one didn't work at all. When Inouye explained who he was, the woman's pique turned to pity. "She started praying for me on the phone," he recalls. "I was raised Buddhist, but I appreciated it anyway."

Inouye needed a little divine intervention. When he took over, the Irvine (Calif.) company had just ended a year in which it had lost $219 million on $684 million in sales. It was reviled by customers for its buggy PCs and horrendous service. And just after Inouye joined, eMachines' stock was delisted from NASDAQ. Analysts assumed that Inouye, who had become wealthy as head of the computer-retailing division at Best Buy Co., would liquidate the company and fulfill his dream of retiring to play golf.

But the 51-year-old Inouye wasn't ready to put himself -- or eMachines -- out to pasture. After eMachines Chairman John Hui took the company private in a $161 million buyout, Inouye delivered a turnaround that's just short of miraculous. In the midst of the tech industry's worst-ever slump, he overhauled everything from PC design to demand forecasting. He cut expenses by outsourcing manufacturing and relying on retail partners to market his products. The payoff: The company is on track to pull in $1 billion in sales this year, up 33% from last year, and has been profitable for eight quarters, Inouye says. This summer, eMachines elbowed aside Gateway to take the No. 3 position in IDC's rankings of desktop PC makers, behind Dell and Hewlett-Packard.

With the company functioning smoothly at last, Inouye's goal is to expand out of the ultra-low end of the market. This spring, eMachines reentered the notebook PC business with a $1,250 model. Now, Inouye plans to debut a family of notebooks -- the PC industry's fastest growth segment(my comment - maybe the first "tier 1" A64 notebook at BestBuy?). His goal is to double revenues to $2 billion by 2005, partly by convincing consumers that the brand isn't just for penny-pinchers anymore. Making the transition won't be easy. "It's very hard to move from the value side to the premium side. People will assume there's a trade-off in quality," says IDC analyst Roger Kay...

Keith, re: Very interesting. And the HPC compiler for Opteron...
Who's doing this compiler?
Paul

Keith - re: Fujitsu/Siemens - one more step in the long journey to success...
Paul

avatar, the way things are going, that rumor should take us down to what - $10? :P
Paul

Cisco beefs up high-end routers
http://www.theregister.co.uk/content/5/34452.html


Cisco Systems has refreshed its carrier-class 12000 series routers by doubling core network capacity to 40 Gbps and adding revamped supporting products, including line cards, software, and hardware modules.

Among the most prominent product additions are the 40 (Gbps) per slot Cisco 12800 router and accompanying higher density OC-192/STM-64 and OC-48/STM-16 line cards.

According to Cisco, these offerings, being field upgradeable, will enable service providers to increase core capacity to 40 Gbps, without requiring major rip and replace hardware upgrades.

The 40 Gbps upgrade provides "dramatic cost-savings" for service providers while doubling network capacity, according to Mike Volpi, senior vice president and general manager of the routing technology group at Cisco.

In addition the company introduced software-upgradeable Cisco 12010 and 12006 edge services routers, designed to provide carriers with the a migration from 2.5 Gbps to 10 Gbps.

Correct me if I'm wrong, but this should mean more flash. Isn't networking equipment where AMD was strong in NOR before the bubble burst?
Paul

Sun sets Solaris x86 free
http://www.theregister.co.uk/content/61/34450.html

Almost two years after it went on hiatus, Sun Microsystems' Solaris x86 has come back in full form with free downloads of the operating system becoming available once again, The Register has discovered.

Some users call Sun's decision to give Solaris x86 away again a small step, but without doubt the move does bring some degree of closure to part of a messy affair. Starting this week, Sun has removed the $20 price-tag for the OS - versions 8 and 9, replacing it with a red FREE sign instead. A fair number of Sun users have kept a close eye on this saga and will remember that it was way back in Jan. of 2002 when Sun first said it would halt development on Solaris x86 only to buckle and bring the OS back at cost.

Big Solaris x86 fan Bruce Riddle wrote, "This is really great news and shows that Sun is committed to Solaris x86," on a message board dedicated to the subject.

And why not pat Sun on the back?

Since giving Solaris x86 users the finger two year ago, Sun has worked rather hard to put the OS for Intel and AMD processors on equal footing with Solaris/Sparc and Linux. Sun has vowed to have its entire enterprise software stack - known as the Java Enterprise System - run on Solaris x86. In addition, Sun now has Athlon, Xeon and, yes, Opteron hardware to run its OS. The Opteron bit is a particular point of interest, as Sun is quickly establishing itself as the premier enterprise Unix supplier for the new chip.

Guns blazing, right?

Well, a rather feisty chap from Penn State University named John Groenveld doesn't think so...

Camera phone shipments jump 29%
Unfortunately, something tells me Samsung will get a larger share of it than AMD.
http://www.theregister.co.uk/content/68/34436.html

Q3 cameraphone shipments were up 29 per cent on Q2, Canalys said this week, taking the nine-month total to ten million units.

Both Sony Ericsson and Samsung saw shipments rise ahead of the average, with shipments growing sequentially by 48 per cent and 34 per cent, respectively.

By contrast, Nokia's shipments were down 3.2 per cent, but it still look the lion's share of the market - some 34.9 per cent of cameraphones shipped during Q3 had the Nokia name stamped upon them.

Doug and Keith, Nah, there's a simple answer - developing is not the same thing as offering for sale.
Paul

Haven't seen this posted. If it was, my apologies!
Paul

http://sanjose.bizjournals.com/sanjose/stories/2003/12/08/daily23.html
7:00 AM PST Wednesday
AMD to open China development lab

Advanced Micro Devices, Inc., the Sunnyvale-based maker of computer chips, says it's teaming with Beijing Peking University Founder Group Corp. to establish a new joint platform development lab in Beijing, China for integrating and developing products targeted at information appliances beyond the PC market.

Projects will initially focus on the home digital media centers. The lab plans to develop products based on what consumers value in their computing experience involving relevance, accessibility and usability, AMD says.

Founder will provide supporting infrastructure, technical expertise and engineering staff among other things to the lab, while AMD will focus its efforts on providing technology and resources, including development tools, products, dedicated engineering support, and reference design kits.

AMD says the deal gives it growth opportunities for its Alchemy and AMD Geode products in China.

Founder is not a small part of a big university. According to AMD's description, it's a hi-tech enterprise founded by Peking University in 1986 and now owns four major companies, 17 wholly owned subsidiaries and joint-ventures, and employs about 5,000 workers.

TechNewsWorld interview with nVidia. No big news, just more positive exposure...
http://www.ecommercetimes.com/perl/story/32355.html
TNW: What's on the horizon for Nvidia?
Daly: Wow, there's a lot of cool stuff going on. As I mentioned earlier, the explosive adoption of programmable shading is changing the look of graphics applications. You can expect Nvidia to continue to fuel this in future GPU products. We're also excited about the future of 64-bit computing. Our new nForce3 products coupled with the AMD 64-bit processors are just now pulling us into this world. And with our new team coming on board from MediaQ, you can expect to see some exciting developments in the handheld space with our GoForce products.

Keith, guess I should have searched. I heard about the T6000 several days ago, and assumed folks here already knew. Maybe I saw it on SI - bouncing between boards, sometimes I forget where I've seen things. <:(
Paul

Can someone explain what a "boxed" transaction is? On Lycos LiveCharts I see a boxed transaction for 252400 shares @$14.80 (16:14:03).
Thanks,
Paul

doug - right you are, I'd forgotten the official part.
Paul
http://www.theregister.co.uk/content/3/33852.html
AMD officials yesterday admitted that Newcastle's die-size is 150sq. mm, less than the Athlon 64's 193sq. mm size, suggesting that it does indeed feature less cache memory.

I think the chip has to be A64's "Thorton", if anything. It makes sense not to throw out 193MM^2 a pop for bad cache units. Unfortunately, if that's what Newcastle is, wouldn't that seem to either indicate cache problems if they're including it in the upcoming Socket939 version, or solved cache problems now, with enough bad cache die around to populate Socket939s for a while? I hope I'm way off base there.

On the other hand, maybe they've decided there's more than one way to get to 3000+ and although they have cache issues, the binning is good enough that they don't want to downbin 3200+s/3400+s to get 3000+s.

Again, this is pure speculation; I have no info other than this one chip + speculation to indicate there are any cache problems at all.
Paul

Doug, that's really odd, the stepping info is identical to the 1Meg A64. Wonder if they got a defective cache wannabe 3200 sample or something? In any case, with the same stepping, I don't buy that it's Newcastle.
Paul

Keith, Glad to hear it!
Paul