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.
yb
AMD had working 90nm-silicon last November already. And yes, the chips spotted now look like production samples.
What makes you think they are not from Fab30?
K.
mas
90nm apparently coming in with a 512KB L2 die initially would bode well for Fab30 capacity next year.
The technicals of your link are beyond my scope, could u elaborate in a simplified way pls?
K.
fpg
From what I see happening on the shelves out there, many people seem to think like you.
K.
confused
Enjoyed your opinion, third paragraph of your posting in particular. Welcome to the board.
K.
Ben Lynch (Deutsche Securities)
downgraded Intc from Buy to Hold.
(He upped AMD April 19 from Hold to buy.)
K.
Thanks Desti (www.planet64bit.de) aka Linux4me @w:o
mas
Some jobs will never probably be multi-threaded.
Definitely... as long as serial problems exist, that is. K.
alan81
Great job, thanks much. K.
Thanks much, Mike. K. eom
fpg
as nearly their entire product line is switching over to it in the 3rd Q
Interesting. Where do you have that from?
K.
mas
Well, looking at this Chart from the cast today it is not only cannibalizing Itanium going forward but replacing some of last year's deployments already...
http://images.visualwebcaster.com/23346/7418/Large/Slide13.jpg
K.
Reseller Mike
Thanks lot for including Celerons.
K.
cg
Right. Agreed, new stepping in validation is at least as probable as the take of my previous posting. Thanks.
K.
Paul
Looking at page 6 of http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/25759.pdf,
AMD did not find it necessary to package the last two K8-steppings for Opteron-products. I assume they would certainly have done so if there were any substantial 90nm delays or 90nm products would not be expected to become available for high-performance parts.
OTOH, a new stepping just weeks before deploying 90nm parts carries the message that 130nm SOI will still be manufactured for considerable time; "great yields" on 90nm probably just means "good enough for ramping wafers for shipments of server-(and FX) products".
K.
CJ
Ok, there are a lot of other things that could be different, but it is a possibility that IBM and AMD are using a very similar, if not identical, process at 90nm. It would make sense if they were as close as possible so that AMD could fab out certain parts even at 90nm.
Toosets of Fab30 (200mm) and East Fishkill (300mm) are significantly different and process different wafers as well, not only in size.
But then, AMD talked of a "second-generation 90nm process recently." Could easily be a 300mm 90nm process to be developed in East Fishkill by the joint team, as a bridge to their common 65nm process if you want.
K.
Thanks for digging this one out, Keith.
Interesting, indeed. K.
buck
As long as Takayasu Sakurai's (and other's) forecasts remain valid for 65nm, there is not only a small chance for power dissipation going up for given clockspeeds.
Parallization allowing to increase performance while reducing clockspeeds is one obvious way to work around the problem for a while, multithreshold transistor design another. A lucky breakthrough in materials or geometrical designs is always possible, but not predictable. Multi-fin-fet geometries will probably ease things, but that is still couple of years out.
K.
wbmw
Do you know which product is manufactured in Leixlip?
K.
p.s: Hillsboro D1C and D1D Prescott, Rio Rancho Dothan, can you confirm?
alan
Text from an Intel Presentation Chart:
Background - Intel’s announced 64-bit extension technology support on workstation & server platforms
EPSD plans to support 64-bit extensions TTM with the processor POR. This currently includes:
Alief, Brandon, Jarrell, Sun Prairie, Grand Prairie, Volcano Peak (Nocona)
Buckner (UP Prescott/Copper River)
Harwich (Potomac/Twincastle)
Eastsound: EPSD Nocona-based blades (Nocona)
Intel RAID Controllers (via new OS device drivers)
K.
Petz
Just noticed that there's open interest of 145,334 contracts on the AMD Jan 06 $5.00 put. Perhaps a big chunk of the convertible holders decided to just insure their holdings against bankruptcy while keeping most of their appreciation potential.
As it is open interest, it would appear somebody seeks insurance, or did I misunderstand?
Also, rather than shorting the stock, some convert holders might sell LEAP calls, and get some cash flow.
Could be. But iirc the terms of the convertibles allow AMD to buy back from Feb 05 already. Writing anything for dates beyond that would be uncovered. Plus, you can only write strikes for calls or leap calls above conversion price without adding risk to the game.
I tried to find historical short interest figures for AMD, but the NYSE only has monthly on their website. SI did rise from 47.3 to 53.7M shares from 4/10 to 5/10.
Dilution from full conversion of senior notes is about 54M shares. As I said, I look at this figure as being the shares sold short to be converted later from senior notes.
K.
Joe,
AMD management is allowed to buy back shares, so they could do so. But not "these" from the senior notes: Looking at short interest figures it is obvious that all or close to all holders of the senior notes will convert a day before AMD is allowed to buy them back.
=> Short interest figures have to be understood as covered empty sales. One reason these guys have not converted yet is that they like the interest on the notes for another 18 months. As you consider it as a hypothetical discussion I might add another consideration would be that AMD could run into trouble again next year and they could cover their empty sales even below conversion price. Hypothetically, I hope
K.
p.s: Besides Senior notes, there is dilution for the Convertibles of some 20M shares. But this only comes into the equation when the common will be priced at or above the conversion price (24 USD and change).
joe
... and except the terms of the senior notes do not allow it before end of next year.
K.
buckwheat: Dilution of some 50M shares is already in the stockprice today (qua short interest).
rupert
Choosing a brand name is nothing really emotional. Define target groups, gather data and process it to make a decision. If you do the first steps with excellence, you can be pretty sure to pick the best alternative from what you have to choose from.
I dont consider myself as being in a target group for the product, so it is completely irrelevant what associations I have.
As for perpetuation of the Duron-brand, it is not really an option unless you intend to sell Paris and Palermo for twenty bucks and change.
On a side note, i wonder what the asian markets make out Sempron. I frequently noticed Celerons being adressed as "Pentium 2" there.
K.
Keith
Thanks much.
I pretty much agree in your assessment of current AMD-based mobile-offerings.
Low-Power notebooks are a platform issue, so we are looking at a handicapped virtual gorilla here.
Put it that way: Competitive mobile offerings are a power issue. (pun intended)
K.
sgolds
Fab 30. Not announced, but no reason to expect anything different from when Fab 25 transitioned from CPUs to flash. Once Fab 36 is up and running with 300mm/65nm CPUs then expect such an announcement (2006 with conversion in 2007). Just history repeating itself.
Maybe later. For this decade, Fab30's toolset allows better use. Think Geode: 90nm-K8 for succession of NX. And GX-III.
K.
alan
The problem Spansion has is they have some older factories that it appears they have no plan, or method, to upgrade to more modern facilities. They will run them until they die at the older generations. They will be a fairly big drag on margins until they finally shut them down. The last time AMD shut down a factory they took good sized hit to earnings.
In fact, it appears that upgrading JV1 or JV2 just would not make sense, so I dont expect such plans. Anyway, I would assume as well JV1 will reach EOL next year or so. However, charges for Fab-closures are mainly related to severance payments for employees. Looking at the facilities in Aizu-Wakamatsu there is an obvious plan/method to avoid this; i.e by gradually moving employees from JV1 to JV3's expanding capacities, which is just next door. JV2s life is extended by means of mirrorbit.
Essence: No big charges looming, let alone drag on margins.
K.
fpg
"Did Intel write the entire article/advertisement for C-net?"
The other way round: C-net writes (and publishes) for its owner.
K.
Could be.
I don't know which issues RC2 still has. However, one I know: Validation of Intels implementation of NX.
K.
clspie
There was a RC2 in May, so i would not hold my breath for RTM in June.
K.
rupert
I do not have any doubt at all Intel indeed gains Flash-mss this quarter. Andy said clearly that most of the upside-surprise comes from Flash. Which implies a lot:
1. We talk about a nine digit figure in Dollars.
2. We talk about at least +25% Flash-revenues QoQ.
3. I cannot neither imagine 25% capacity expansions QoQ nor significant price increases.
4. ==> Intel is selling its Flash-inventories off this Q.
Andy said inventories would be apprx. flat QoQ
5. Flash inventories down means CPU inventories up then.
On the FASL side, I dont see significant inventories to be sold this quarter. And I dont see a capacity expansion of 25% either.
In essence: Intel gained share. Not for long, though.
But they manage to make it looking good this quarter.
K.
p.s: Dont confuse "Media-Analysts" with Analysts. Listening to the Q&A my impression was that couple of them have a firm grip on what is going on, just from the way they asked into this.
June 3, 2004 -- Intel Corporation expects revenue for the second quarter to be between $8.0 billion and $8.2 billion, as compared to the previous range of $7.6 billion to $8.2 billion.
K.
Reseller Mike
I see. Thanks lot in advance for your efforts.
K.
Mike
Thanks for posting. For a better overlook, could you include Celerons in the future?
K.
@buggi
While I think your figures for North-and Southbridge are fairly precise, the chipset i referred to includes grafics in the northbridge. To determine what power handicap RAMDAC and ADC really mean (which would be heavily dependent on quality of grafics btw in terms of resolution, Colours and refresh rate)
is way beyond my technical capabilities. Thats why I called my estimates uneducated. However, in the lid of my AMD-64-book there is a component producing heat which by the touch does not feel like milliwatts, but like Watts.
Then, although it is fairly old, i think this on is still valid:
http://common.ziffdavisinternet.com/util_get_image/1/0,1311,sz=1&i=10030,00.jpg
K.
paul
Answering my own question, here we are:
http://www.intel.co.jp/design/chipsets/mobile/pix/855gme_diagram_760.jpg?
And I stand corrected: Without chipsets providing digital video out of integrated grafics, there is no way to come even close to Centrino battery runtimes. For a rough and uneducated estimate, AMD has a 10W handicap here for DAC and ADC. (Corrections welcome for this figure)
K.
Paul
Thanks. Yup, I see an arrow from the Northbridge to CRT.
Now, for a low-power mobile design it seems to be not exactly ideal to convert digital video output to analog signals in the Northbridge (DAC); using LCD would require another transformation back to digital signals (ADC), which all not only consumes significant battery power but degrades performance and quality as well. (Or what do I miss here?)
K.
btw, while we are at it: Somebody knows what Video-output Centrino-Chipsets supply?
rupert
Interesting. Grafics seems to be the part missing, unless I overlooked it.
K.
p.s.: Fixed link: http://www.amd.com/us-en/assets/content_type/DownloadableAssets/31760a-nxdb1500_devbrd.pdf
p.p.s: My web-acrobat-version crashes trying to display the pdf. Acrobat opens the downloaded file.
paul
Hmm. I think Keith was already right on the spot: Its takes many puzzle-pieces fitting together to make a product competitive. Chipset is an important piece, but probably display is more important as it accounts for the biggest single portion of battery drain.
K.
p.s: I hope your reconvalescence is positively correlated to your posting frequency...
chipguy
I have no doubt you have these numbers right. However, they have no relevant meaning for my mobile user profile.
K.
Keith - and Andy Grave
The school project could be used to get access to Chinese ressources for making a platform for it in mainland china.
Which could be the basis for a people's notebook for the China domestic market as well to achieve economies of scale. Adding appropriate components I would not be surprised by a product for the world markets undercutting Transmeta-prices by half. An ULV-Barton would be a nice drop-in to come even closer to Centrino. Probably not up to par, though. Therefore it would need K8s. And another platform. This one could be something for the school project - and for the other things as well.
Given we see something from mainland china based on Geode NX which cuts something on the markets, there is a good chance chipset makers will bite and make ULV-K8 platforms. Which is probably what AMD really intends with its recent move.
K.
Keith
Many pieces to that puzzle.
Lol, yes. Frankly, I dont see the whole picture of the Geode-brand introduction of the ULV-Athlons. But I am sure the folks have a plan with it. I knock on wood that it unfolds as intended.
K.