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.
Mike,
many thanks for the update.
For a conclusion, I have been burned too often by conclusions I drew from the variety show this industry (and in particular AMD) kindly presents to us. Insofar, looking at just couple of thousand parts in the few channels the public has some insight due to your diligent service and at some accidently or less accidently posted OPCodes of x52 models, I am still reluctant to make changes to the model to account for faster progress than implicated currently on the process side.
Gut feels lil Keithish, if you know what I mean... :)
K.
mmoy OT
FlashBlock is quite radical, it not only prevents the animation from being played but suppressed flash banners from being displayed at all and just shows a symbol instead.
I did find Flash Click to Play too, better said Flash Click to View, linked to Flashblock as well. Seems to be the same extentions.
Good for now. Many thanks.
K.
mmoy OT
(How) can I prevent Flash-scripts to run in Firefox on default?
I am getting annoyed by these, in particular if they play sound.
TIA
K.
Alan
There won't be retail versions of the OS. It comes in an OEM-flavour. Based on the Pro-version.
K.
Keith
in a static view, it does.
in a more dynamic view, going multicore, I am not sure.
Admittedly, this is beyond the scope of the review. If the author could have resisted to draw conclusions in this respect in his final word i would not mention it at all.
K.
doug
We'll see. I'm just trying to find an explanation why it goes so well. A steep learning curve in straining, maybe.
K.
cd
Sounds somewhat convincing, I admit.
The fact that Mike has 90nm FX55 listings is supporting as well for high bins from 90nm.
Hmm. Assuming its 90nm for the moment then, and D-stepping it would imply bins are up some 25% in just half a year.
Which I would consider as pretty impressive.
If it's E-Stepping, I'd expect AMD would account for architectural improvements in the modelratings, it could be 2,4GHz parts then. I'd be impressed of binsplits as high as that for the initial output of a new stepping as well.
Probably safe to conclude 90nm sweetspot is higher than 130nm in terms of bins. And yielding enough parts to allow a top-Opteron launches spells pretty good yields as well in terms of defect-density and cache distribution.
All in all, this would imply driving migration as fast as retooling allows. Very good news, if confirmed.
A lot better than I expected.
K.
tecate
Possibly. We will never know. :)
K.
tecate
Currently no big deal, I agree. (Although apparently big enough Intel released somthing to shift the public focus away from the issue). In fact, the thread in Japan to be slapped publicly, oh boy. BFD. Really. I mean you can do every sort of BS in a company, destroying billions of Yen. Nothing much will happen if you apologize often enough.
But if it becomes public that the company behaves unlawful, someone has to leave. At least the company. Suicide is not uncommon in these occasions. It's frigging real thread what the FTC produces. Been there - but obviously _not_ done that. But learned if you are not willing to adopt the ritual apology-procedure it's better to stay away from Japan. It's just essential for anything you do there, not a question of if but when.
K.
tecate
Actually, a fine would be counterproductive. Intel would pay and continue its usual practice. As I look at it, the real adressees of the warning are rather the japanese companies involved. It's very likely they let it look more competitive going forward if they feel at risk of a public slap.
Japan is different; as Combjelly posted already.
K.
Alan
AMD does not break out the inventory based on location... I.E. WIP, raw goods, and finished goods
It does. In Q10, issued some four weeks after its earnings release. Should be out within a fourtnight for Q4/04.
Btw the 11 metal layers are for AMDs second generation 90nm process. I am not sure this one will be deployed in Fab30 already. What is in the fab currently in 90nm has more process steps than 130nm though. AMD expanded cleanroomspace by 20% to account for this (and to ease migration, i assume. They weren't exactly specific here when i asked this, unsurprilingly). About TPT they said blabla improved scheduling counteracting blabla. I understood TPT up a tad.
K.
Keith
Yeah. I know. It's not shooting the moon anytime soon. Should add 10-20M per quarter though.
Heck, didn't I say stay away with these needles for today?
K.
Keith
I am aware of the blade marketvolume. Half a billion in 2004.
Exptected to quadruple until 2008. (IDC data and projection)
Not exactly chickenfeed. But well, blades have seen to grow fast for couple of years already and the segment just did not behave like the forecasts predicted. But one source which I believe is credible in a tier-one thinks it will come. He sees doubling every year from current year going forward.
But well. Could be both predictions are wrong again.
K.
Keith
I came to the very same conclusion.
I went behind the question why the designs are not there. Explaining this in detail would be beyond the frame of a posting. In short, it has to do with a lot more things than just Intels strongarming - although this certainly plays a role. One of the dominant findings killed another myth in my humble mind: Processor performance in most of the low-level server application does not matter at all. Even less than in desktop space for office applications. Typical CPU-loads there are in the lower single digits.
Insofar, i stand corrected on the penetration curve stallwarning buzzer remarks from Q4 results. It's actually not as surprising as I took it. And consequently rather sobering than alarming as soon as you look at the fragments of the serversegment instead of in total. Actually, the steep parts of the penetration-curve are already behind. There is still some potential to be opened up in the niches where performance really matters and K8 shines, but most of it is already done. Dualcores will help to make another couple percent of the total market accessible, Galaxy and Horus will help as well, but this part is unfortunately meager in terms of CPU-revenues and will add not much to AMDs bottomline.
For the high-volume segments of low serverspace, TCO considerations literally forbid dealing with a second architecture. And consequently, the lack of interest in Opteron cited by IBM and DELL is not as unsubstantiated as often claimed.
Well. Puuuf. So far for this bubble in my mind.
Lucky me, I got more in there. Blades. Power-envelope matters.
Workstations. Performance matters.
Anyway, one bubble-burst per day ist enough. So the folks with needles for the others, stay away from me for today. Try tomorrow. Thanks.
K.
Alloycon
Pick it from here:
http://www.sun.com/aboutsun/media/analyst/sas2005/
K.
Keith
The right to be wrong is the only right I claim for me.
Let's put it to a rest with that.
K.
Keith
The reason many shops don't list FX55 anymore is simply that demand for 900 Dollar CPUs is too thin it would make sense. Everybody puts it on the pricelist first and waits for oders. If none come, you take it off.
Check it out and call a dealer of your choice and ask for it. They will happily take your order as soon as they have confirmed your creditcard.
K.
dacaw
The cost of an animator/illustrator is so much more than an FX chip that it only makes sense to buy the fastest kit.
No argument. That's why Opteron systems are used in render farms.
K.
Keith
With all due respect, stock information on the net is beside the point for the availability criterium. Whether _you_ can buy you one or not if and when you want is the crucial point in this respect. As simple as that. What you rightly cite in your post is just to make you pay close to 900 bucks for it and be proud and happy you got one.
K.
Keith
I would not know of anybody desiring an FX who would not have been able to get one. Do you?
I could get one delivered to my door tomorrow. In my humble understanding the part is available.
K.
drbes
In this respect, it is rather a variety show. Entertaining.
Looking at what the audience makes of it is the comedy part. Amusing.
K.
Keith
Could well be an illusion as well. The FX position requires to make these parts look rare.
K.
StocksR4Me
Many thanks for your elaborations. I guess your reasoning is far more probable than a possible throttling; although there are throttling mechanisms not obvious as well. The first tool to detect these i am aware of is Throttlewatch.
Just in case you did not already come across it, here:
http://www.panopsys.com/throttlewatch.php
Good luck for the Opteron-Servers.
K.
StocksR4me
Interesting. Did you track back the high CPU load issue?
From what you say i assume it was Prestonia Cores, right?
Throttling?
K.
Joe
Well. Less radical, less efficient, less integrative, less compelling.
Most important, it's all about perception. Although there are downsides (you can't add memory modules) my gut says it will be perceived very well if communicated right. It will make things easier for OEMs as well. No more bothering about RAM-modules compatibility, evaluating, testing etc. And cheaper mainboards as well, naturally. A layer less (or two?), no Ram-slots, less testing, less trash. Bottomline: Less hazzle for everybody.
K.
CJ
There is no replacement for DRAM cell in the cards afaik for this decade.
What AMD can do (and possibly will do) is to let the second step of integration follow its first step to integrate the MCT: Bring DRAM to the chip by means of a MCP. Will open up pretty much the same potentials as on grafics cards in terms of clockspeed, timings and bandwidth.
K.
Buggi
Hmm. I see the platform issue as well - more precisely not really a shortage of socket 939 boards but rather of such with new chipsets. Is it that what you see as well?
Anyway, i am reluctant to look at this as the dominant factor in the 130 versus 90nm supply dimension; in this respect, what your assumption would imply is that 90nm went into inventory last quarter. Hard to imagine for me.
K.
Keith
Thanks for the comment. Indeed, Hectors statement implies an acceleration of migration from what was said earlier.
HR: Right now we're, within some first order of approximation, we're getting pretty close to over half of our starts being 90-nanometers, and our plan is to be, for all practical purposes, converted totally by the second quarter.
But then, he left himself not only one, but several backdoors open to escape from it.
Considering AMD is investing a ninedigit figure this year for toolchanges in Fab30, maybe we should look at it on a basis of reduced overall waferstarts; insofar it looks more feasible, although still ambitious imo. From this viewpoint, and counting in a stepping and a new dual-core process to be deployed, folks in Dresden are not at risk of getting bored - carefully said.
K.
Petz
We have seen products spotted in the wild by August, so obviously waferstarts happened last May. :) Just not as many as many thought.
They said inflection point of 130/90 waferouts in summer 05 last year. A stepping to come before. They'll make that. Summer in semiconductor language is flexible. And they said migration to be almost completed by year end. They'll make that as well. Almost in semiconductor language is flexible enough as well.
K.
Joe
Lol. Now that I tried a "think pink" posting once again - for a change
Btw i don't see any "90nm mystery". What I see in terms of volume, characteristics and binsplits coming is typical for pilot mode. Volume is still to come. Retail is used as a showroom to demonstrate 90nm is there. Everybody can see it. Most forget it's only a small part of AMDs distribution. In a nutshell: 90nm is not yet as far as many want to believe. Doesn't matter much thoug, as long as 130nm is competitive and there is no manufacturing constraints.
K.
Mike
Thanks again.
90nm 3500+ volume indicates improving binsplits from this process. Good to see.
K.
tecate
the 17 year old hates change.
Huh? Times must have changed a lot. I used to fight for all sort of changes at this age.
K.
mmoy
Sure. Fear sells. :)
K.
mmoy
That sounds like a promising approach. Fear sells.
K.
mmoy
Nevermind. It's good enough. I just click as often as I need to have it big enough. Think along the line everything beyond clicking is beyond my level of sophistication - at least when using a browser.
K.
mmoy
Thanks - works. Helps for now. Don't need zoomout. It's so that I use Ctr+ anyway if small fonts are used. Prevents me from using reading glasses, dont know if I use if for convenience of vanity.
On occasions, I am interesting in some details in pictures. That's what your script is helpful for.
K.
mmoy
Actually my handset is capable to act as a modem for my notebook via infrared at no fixed costs. The carrier says at 56Kbit. Never got such a connection, usually like around 28Kbit. Usage is silly expensive, if I use it abroad (being in Europe almost every trip is considered abroad) it costs double-digit Euro amounts everytime you use it for couple of minutes to get your fix of funny pps and the like apart from spam - plus the one or two messages you really waited for if you are lucky . While you have to take care the mobile phone remains in the right position to the IR-receiver to the notebook - in a train or a taxi, you look really like a fool.
Stoneage. I tried automatic forwarding of messages straight to the handset - works good for inbox, but not good for anwering.
K.
Btw: Any idea why image-zoom still is unavailable for FF 1.0?
I find this one particularly helpful.
Comjelly
Right on the spot. Anecdotically, I had a phone call from an old friend these days. He told me that he recently bought one of these "radio - laptops" to have access to his emails when he travels. He said he could get the thing to work, if I could go through some settings with him?
I found out his understanding was wifi works just as his mobile phone. Now, this chap holds a masters degree in economies and is bright in other respects as well. Somewhat telling where people really are out there - and where you have to pick them up respectively.
K.
tecate
Applause. Girls always have much better comprehension of how boys play than the boys themselves. :)
K.
Keith
Hmm. While I could make another approach on Hector to see if I can get a commitment for a June 30 launch, i know what answer I would get if I were cheeky enough to ask for disclosure of the information you desire: "Tell him this a homework question."
K.