Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
http://online.wsj.com/article/SB120792982404908185.html?mod=yahoo_hs&ru=yahoo
Advanced Micro Devices Inc. said Phil Hester has resigned as senior vice president and chief technology officer.
Mr. Hester, a 23-year veteran of International Business Machines Corp., joined the chip maker in 2005. The decision to leave was his decision and was not associated with AMD's recently announced job cuts, said Rob ...
-VBG
AMD Roller Coaster Enough To Make You Sick
http://community.investopedia.com/news/IA/2008/AMD_Roller_Coaster_Enough_To_Make_You_Sick.aspx?partner=YahooSA
-VBG
They don't even mention the massive power gap. :-O
??? in the conclusion you mean.
From the page you linked:
Intel's idle power consumption (or lack thereof) clubs Phenom harder than a baby seal on the first day of hunting season, and things don't get any better under full load.
From Krazit's blog:
There might not be a more costly undertaking in the modern technology industry than the process of designing and manufacturing high-end PC and server processors. Only six or so companies really do it in high volumes (Intel, AMD, IBM, Sun, Fujitsu, and Via).
Wow, I learn something new every day...
-VBG
just heard that george ou has been laid off from cnet/zdnet/techrepublic. never got along with the guy but that's a major bummer
— Ryan Naraine (@ryanaraine) March 26, 2008
How many more times will you make this so-simplistic-it's-meaningless argument?
Ooooh! Ooooh! I know #6!
b) 10^42
I hate when reporters do this:
Knox did say that the 45 nanometer process that AMD has created in conjunction with partner IBM is on track and that the Shanghai Opteron variants, which will have 6 MB of L3 cache, are on target for shipment later this year, despite the delays with Barcelona. Given the shrink from 65 nanometers to 45 nanometers, AMD should be able to push clock speeds well above 3 GHz and the tripling of L3 cache should have a big effect on performance.
Attribute the first sentence to a company employee then follow immediately with his own speculation, leaving it ambiguous as to whether Knox said the second part about clock speeds and performance. I seriously doubt anyone from AMD is going to go on record promising Shanghai at "well above 3GHz", but I'm sure there will be partisans quoting this. Probably the same people who were predicting Athlon 9000+ release in 2006. *SIGH*
-VBG
They might not gain enough of a benefit to make it worth it.
HK/MG (or at least gate-first HK/MG) seems to benefit PMOS much more than NMOS. If they don't re-layout the cells to take advantage of this, they might lose much of the value of the IdSat improvement. I'm just speculating here, but I keep remembering the comment about Nehalem 'taking full advantage of 45nm'.
I do believe AMD could bring this into the process in mid-stream, but I expect there would be an initial yield hit, so the yield/performance trade-offs may not work out.
Just my 2 cents.
-VBG
Just because they show something in research doesn't mean it will become reality. III-V may not pan out. Remember the "TeraHertz transistor"? FD-SOI didn't seem to make it out of the lab, although other techniques like raised source-drain did.
-VBG
I'm confused. I thought cache benefited more from a shrink than logic. So shouldn't we expect a design that is mostly logic and little cache like Barc to shrink poorly relative to a design with more cache %?
Regardless this doesn't seem very promising from a cost standpoint for AMD. I would expect that using immersion @ 45nm would provide more benefits.
Has AMD divulged any details like contacted gate pitch for 45nm? Has AMD announced any specific improvements in Shanghai that might account for the extra die area?
Maybe Hans will do a detailed die plot analysis soon....
The surprise for me was not that they did the demo, but that they are sampling to customers. Shanghai might not be that far behind Nehalem. I still doubt their process is going to allow much speed upside and I don't think they have any prayer of recapturing the performance crown. Maybe it will give them same cost advanatges v.s Barcelona. What is it, ~15% smaller die?
-VBG
Intel warns for Q1 - due to NAND
http://biz.yahoo.com/bw/080303/20080303006511.html?.v=1
Arrgh!!
-VBG
I decline your invitation to wrestle my porcine friend.
I hope others do the same.
-VBG
Pete,
I work with people doing 32nm and 22nm development.
You don't.
End of discussion.
-VBG
Come on wbmw, will you lay off poor Peter.
He has a hard job trying to convince people that AMD is somehow ahead of Intel on process development. (I say job, but I suspect it is more of a strange hobby.)
Allow him his small victory. He found a rumor that Nikon delayed delivery of a EUV tool to Intel, combined that with the announcement that AMD had used EUV to lay some metal lines down on an existing 45nm test chip made in Dresden and it actually worked. 2 + 2 = 79 and Intel is in imminent danger of being overwhelmed with 22nm EUV parts.
Yes, it's a patently absurd conclusion, but it's only convincing to those who are both technically ignorant about EUV (many readers, I suspect) and unfamiliar with Peter's massive credibility gap (very few, I hope).
You refute every crazy post he makes and I appreciate your dilligence, but it doesn't stop him from coming back. Maybe if we ignore him he will go away. Why not have a trial period, say one month, and we can count his posts before and after?
What do you say?
The above all my own opinion of course.
-VBG
AMD Strategy to be Watched Closely
I don't agree with the analysis, but...
http://biz.yahoo.com/zacks/080225/11638.html?.v=1
Advanced Micro Devices (NYSE: AMD - News) is the second largest producer of microprocessors and chipsets in the world. December quarter revenue missed the consensus by a sliver, although the EPS exceeded handsomely. Forward guidance is for revenue to be down in-line with seasonality (we estimate around - 7% sequentially). The company has launched the first triple-core processor families, and these products are expected to gain traction in 2007.
Additionally, margin improvements shall continue as new products gain momentum and 45nm production ramps up. We are extremely bullish about this stock, and our Hold rating is based on uncertainties in the macro environment. Shares of AMD are currently trading at 0.6x our 2008 sales estimate (P/S).
While Intel Corporation (NasdaqGS: INTC - News) remains unparalleled at the high end of the microprocessor and chipset market, AMD has been steadily taking share at the low end. Management's strategy of providing lower cost solutions has enabled the company to gain share in the greater China region.
We think we will see the advantages of this strategy unfold over the next two years. As AMD churns out faster and more efficient products at the low-to-mid-range, a large percentage of users, with smaller functional needs are likely to switch to the lower-priced product. We are reiterating our Hold rating and $11 price target, which corresponds to a 0.94x multiple of our 2008 sales estimate.
-VBG
Anyone up for a prediction contest?
I'm thinking
When will AMD 45nm be avilable for retail sale?
and
What will be the highest frequency (available immediately, not announced) at said time?
My guesses: Feb '09 and 2.6 GHz
-VBG
Intel Unveils Healthcare PC Solutions
http://www.koreatimes.co.kr/www/news/nation/2008/01/123_18133.html
-VBG
Yup, looks like about a 4 month lead to me ;-P
-VBG
X2 3600 sells for $45.When will Silverthorne-like products catch it in performance ? 2010 ?
It won't and that's not the point. The point is Silverthorne will (hopefully) sell in the 10-100s of millions at ~$20 and at reasonable margins.
Edit - A dual core version of Silverthorne might well be reasonably competitive with low end X2s, but no timetable there. Also I'll note that if MIDs take off, I'm sure AMD will respond with a product, but unless it is already in the pipe, they will not be competitive for a couple year window.
But are MIDs going to rock the world or will they have the fate of the ViiV ?
That is the 10 billion dollar question. Wish I knew. Hope PSO is right. Have my doubts. Let me ask you, where should Intel focus their efforts? Continue to drive performance in their niche only or look for adjacent opportunities? Where would you have them try to grow?
-VBG
Disclaimer: The $20 price number was pulled straight from my butt and I have no actual knowledge concerning any product pricing or strategy.
I think your questions and beliefs mirror much of the sentiment about Intel stock in the market right now.
As another poster mentioned, Intel is not doind a good job of selling their story. Wall Street was heard Intel CEOs promise to break out of the PC space for at least 8 years now and are rightly taking a wait-and-see attitude. Is the MID a viable product? Will it take off? Will Intel be the prefered silicon provider? Will Intel cannabalize their existing high margin business? Can they break into the CE space? These are all valid questions hanging around the stock.
I am dissapointed by the reaction to Q4 results but as a employee I am more concerned that management focus on executing their plans rather than selling their plans to Wall St. If they execute, the stock price will follow.
Having said that, in hindsight I think Paul, or better yet Stacey should lay out some numbers instead of all the dancing around questions about margins. Here's what we think the market will be in 2-3 years. Here's what we think the margins will be like. Yes, they are lower than traditional CPU, but here's the top line growth, here's the bottom line growth.
-VBG
And Intel can't thrive in the sub $100 market.
I'll have to disagree strongly here.
Intel is much better prepared to compete in these markets than AMD. Intel is designing specific products for low cost market segements. Has AMD announced anything comparable?
It is AMD that can't thrive in the sub $100 market. They can't thrive in the markets they are in now. And they have no chance competing with 65nm X2s against Silverthorn.
-VBG
Based on die size and yields QC might not be a laughable % of wafer starts.
-VBG
we believe that the technological gap is only 4 months
Agreed, truly unbelievable. Granted I'd not choose to be in Mr. Hester's shoes at the moment.
I'd be willing to bet the transistor performance gap is a lot wider than the lithography gap. And the lithography gap is probably close to a year depending on how you measure.
Interesting comment at the end wrt "asset-smart" or "asset-lite" or whatever they are calling it these days.
-VBG
The claim on the SI AMD thread is that AMD can hoard opteron dice in a die bank, and process them as demand requires. It is hard to know how much excess production they have been able to squirrel away in the bank.
How stupid is that? Let's bank something that continuously decrease in value as it sits on a shelf and increases inventories to boot. When exactly were they banking these die in all the quarters they have been claiming to be production constrained?
Typical logic over there I suppose. I really should visit more for the entertainment value.
-VBG
Memory latency improvements with Nehalem
The thing I found confusing was the system diagrams on the first page that show the memory directly connected to the processor for 1P/2P/4P systems contrasted with the "Tylersburg HEDT" system diagram shown on the second page which have the memory connected to the chipset.
Can we infer that all Nehalem flavors have a IMC?
Is there a redundant IMC in Tylersburg that is only used in certain configurations? Won't this high-end desktop configuration suffer from a memory latency penalty v.s other configurations?
-VBG
Edit: According to HKEPC slides at least, Bloomfield will support 3 channels of DDR3.
Thanks for the link to the IEDM papers.
The thing that surprised me most was the lack of scaling of gate length. Apparently gate length has hit a wall like oxide thickness did at 65nm. HK/MG is a way around the oxide barrier for several generations, but what is the way around S->D leakage at small Lgate? FD-SOI helps, but I don't believe it eliminates the leakage, just reduces it somewhat. And FD-SOI is expensive. Maybe tri-gate will allow performance improvements to continue, but if HK/MG is the biggest change in the transistor in 40 years, tri-gate is even bigger.
-VBG
Has Intel crushed AMD?
http://bigtech.blogs.fortune.cnn.com/2007/12/12/has-intel-crushed-amd/?source=yahoo_quote
Love this line: For starters, it’s clear that AMD’s top bosses drank too much of their own Kool-Aid.
Think the author Jon Fortt lurks here?
-VBG
Anyone else having trouble with this link?
http://library.corporate-ir.net/library/10/101/101302/items/272603/20071207_Lehman_Krzanich.pdf
TIA, VBG
I think I see a pattern...
They also voted the RIM BlackBerry Curve 8310 "Smartphone of the Year".
And the "Notebook of the Year" is a Toshiba with a LV Yonah processor.
And WinXP as "OS of the Year"
Maybe they switched all the also-rans with the first place ones (Core 2 Quad, something Santa Rosa class and Leopard) by mistake?
-VBG
Also keep in mind that AMD is figuring 8W difference per DIMM, which is pessimistic, and 16 DIMMs per server, which is unlikely for two socket servers.
At the same power envelope, Harpertown will probably out perform Barcelona by something in the 30% range. So the Intel power cost goes down by that amount.
Then figure the cost of those ~150 servers you don't buy, provision, support and license and see how many years you need to run them for the power to make a difference.
-VBG
Coddling Intel
Opinion piece from International Herald Tribune
http://www.iht.com/articles/2007/10/30/opinion/edintel.php
No author anywhere on the article that I could see. I'd suspect some AMD hack, but its better written that most of the stuff you see from them.
-VBG
Only thing that matters is real world performance.
I'm sorry but if you want to talk about performance, or manufacturability, or availability you should post here:
http://investorshub.advfn.com/boards/board.asp?board_id=369
On this board architectural elegance is pre-eminent (and FLUENT 6.4.3 beta subtest sedan_4m)
-VBG
Nehalem is a family of chips and MP is 8-core (Beckton).
The demo was a 4 core (Gainestown IIRC). Then there are the DC and DC w/GPU flavors.
-VBG
Interesting article on what makes a good handset experience.
http://seekingalpha.com/article/49988-why-apple-s-iphone-doesn-t-need-3g
My takeaway is its more important for Intel to get better low-power processors into the marketplace than WiMax silicon, FWIW.
-VBG
Hector Interview, Part 2
ROTFLMAO, that was precious! (and perhaps prescient)
Especially liked the part about the 1 billion transistors.
Can you add a question from Mark Lipacis asking about customers double, triple and maybe even quadruple ordering Barcelona because they doubt AMD can fill the orders?
-VBG
Any correlation between a supposed inventory correction and the layoffs is pure coincedence and asserting otherwise is frankly just foolish. The layoffs are due to longer range workforce reductions to make the flash spinoff profitable. Getting rid of people in Europe is expensive. Intel would not do this for a temporary change in output for a couple quarters. More likely they would institute a hiring freeze instead and let normal attrition bring headcount down.
As for the stock price dropping to 23-24, who knows?
As for management being on top of containing costs, you can bet they are.
-VBG
Standard Disclaimer: Any statement above is my own opinion and not the opinion of my employer.
Is AMD preping a new stepping of Brisbane, or have they made 65nm process improvements that are benefiting both X2 and Barcelona allowing higher bins of both?
Maybe AMD has finally taken the Si-Ge plunge?
If so is it in desperation, or is it yielding *maturely*?
-VBG
I like, many thanks. Found this right off the bat.
http://www.eetimes.com/news/latest/showArticle.jhtml?articleID=202400860
-VBG