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.
Petz
I dont think there is any inventory buildup from Q3 production. Manufacturing a CPU takes a quarter, as you know.
With regards to XP-inventories, I hope AMD shipped lower speedgrades (from inventories) this quarter than they put in from Q3-waferstarts. This implies little or no more Thoroughbred waferouts - which leaves AMD with two maturing processes for Barton and K8.
Therefore I doubt FAB30 currently is able to churn out more marketable parts than AMD ships in the strongest seasonal quarter.
K.
Petz
I didn't dream that, did I?
No, that was real. Real rumors, though. K.
Petz
So Q3 was not operating at max capacity
You bet FAB30 was running full steam.
COGs are calculated on the basis of products sold in the quarter, so yes, cost increase in the ballpark you mentioned.
Unless currency-impacts drive costs this quarter. K.
Petz
Sounds like real life.
However, from the announcement of an AMD-64 OS, using M$s usual development timeframes pointed to a launch by mid (or even fall) 2004 realistically (better said, even optimistically). K.
elmer
c'm on, don't be narrow-minded here, we did not say which country's fortune 500, did we?
I mean we already called every single american company as we did not get any response on our mailing - and we have a program and budget to spend.
Unless you come up with a brilliant idea to get this piece down to american throats, let us just continue.
P.O.
p.s: I am pretty confident we can announce this chicken-farm thing in Usbekistan before Christmas, being the first ever solution worldwide counting eggs REALTIME.
chipguy
LOL, even IT people in developing countries are passing on AMD's el cheapo 64 bit solution.
Well, it seems Intel had a long run to find the first candidate for its free evaluation program.
Interesting to follow where the next one will be. Kongo?
Chris
While you still could be right, Mike Dell seems not really convinced things happening would be good for his company
http://biz.yahoo.com/rc/031216/tech_dell_shares_3.html
Andy Bryant seems as well not completely confident in Intels (short term) prospects, as he did not buy back shares for the company last quarter for the first time since I follow the company - and sold the majority of his own Intel-position as well couple of weeks ago.
http://finance.yahoo.com/q/it?s=INTC
K.
Keith
The price war in the INTEL chipset market and the end of the christmas selling season are factors, but maybe there´s more to it?
Yes. A lot more. What do you think Intel will use capacities for being replaced by Prescott-Ramp? I would expect these numbers to be only the tip of a really huge iceberg every chipsetmaker is heading towards. Except one, that is. This one could easily be a target for an acquisition. Because grafics is maybe the next intended usage for capacities to become obsolete for CPUs next year, and GFX-2 could be not enough to fill enough of it. It executed, side effect is an amputation of some annoying "virtual gorilla" out there.
K.
wbmw
Hmmm. Ok.
As for: "It may be a very good time to invest."
12.17.03 18:48:03 13.98 13.97 13.98 627500
That yours?
K.
wbmw
I'll take that as a complement
It was meant so. Few people are able to give up convictions.
I do completely agree in the rest of your posting.
However, why did you omit any word about IPF in this context??
K.
wbmw
What they really need, though, is a next generation architecture
Well, they communicate they have one already.
Frankly, I'm stunned it is you making this statement. :-0
Welcome to the club!
K.
Keith
Yes i did. And I agree, impressive indeed. Lessons to learn.
K.
wmbw
This tidbit in the review
On the Pentium-M, a mere 1/32 of the L2 cache is permanently active while the entire L2 cache on the Mobile Athlon64 has to be permanently powered.
is new to me. Can you (or anybody) confirm it?
K.
Edit: I found confirmation for that - and more:
"Fine Grain Hardware Gating". Pretty impressive.
yourbancruptcy
I bet that is what they did with SSE3 now
That would perfectly fit my gut feelings about Hans de Vries' discoveries of Prescotts 64-bit capabilities.
K.
jules
Thanks - did not find this particular piece.
I just wonder if the latency-penalty of DDR-2 will counteract AMDs low-latency solution of integrated memory-controller.
Maybe someone can stick in?
K.
keith thanks
I read this piece (and others) long ago, thats why I am somewhat sceptical about inhowfar DDR-2 is really beneficial for the majority of applications in the specification of 2004. (Maybe the lower Voltage benefits mobile Computers by a Watt, i dont know). I understand there is potential in it that DDR-1 does not have, so from 2005 DDR-2 should be superior to what DDR-1 could deliver then. Therefore, i am not expecting this one to develop like RDRAM, but AMD to follow the DDR-2-way.
One thing is for sure: As soon as Intel in on DDR-2 there will be plenty of applications in the benchmark-suites that will show increases in bandwidth and barely none which would show the downsides of higher latencies of this type of DRAM-bars.
Im not expecting DDR-2-capability for the 939-intro. But for the 90nm-intro in automn, yes i think these parts will be capable of handling DDR-2. However the transition from DDR-1 to DDR-2 is completely independent after 90nm launch, as i would expect 90nm Athlon64s to be able to deal with DDR-1 as well. So the MB-Makers can differentiate their products as market requires (as for demand and Price/availability of DDR-2).
K.
sgolds
Ok - thanks. eom
Sgolds
Ok got it.
I see that another 33%-increase in some numbers is served.
I mean that is perfectly fine with me as long as it helps selling boatloads of magazines, CPUs and memory modules.
Now, apart from that, any real benefit in here? What about latency DDR-2 vs DDR? Timings and stuff?
Iaw, any improvement to be expected at all except to be seen on some synthetic benchmarks?
K.
Doug
Sorry, i cant answer that. If anybody, Buggi probably would know. K.
sgold Thanks for the DDR-2 primer
If I get it right your point is that K8 (939 pin) is likely to work with 133MHz DDR-2 which would probably be called
DDR-2-533 and the modules could be called PC-4267 bandwidthwise?
Basically the same modules that Intel intends to use from next spring with their chipset for LGA-socket, right?
K.
Chipguy
Not my turf. There are folks around here far more capable to solve the riddle for you - if there is a solution at all, that is.
K.
sgolds
I remember. Actually what I understood when they said "future parts" was 90nm. Could be wrong.
But taking availability, JEDEC-specification, price and benefits of DDR-2 into consideration automn 2004 (90nm) would be fast enough for this (in my books).
OTOH, if its really trivial (and doable en passant with a new stepping), why not earlier - it could help to sell a couple more FX next spring.
Memory is really not my field of expertise - what performance benefits are likely to be delivered with DDR-2 over DDR(-1) at the same clockrate?
And, it requires other memory-slots on the boards, right?
K.
Paul
But the main point is that with dual channel, 939 should be an even more substantial bump over 754.
What else would you expect going from 754 to 939pin, other than allowing for 128bit-wide Memorybus?
K.
sgolds
I sure hope SUN not only survives but prospers again. But i clearly see Sparc to become obsolete - only if Scooter will insist on building SUNs existence on it, it is in danger.
K.
Keith - thanks.
Thats even a lil less than I thought it is.
K.
chr p
I haven't seen yet any good reason why SPARC should be thrashed
From a pure architectural view, I dont see one either.
The simple reason Sparc will go the Alpha-Way (and the DEC-way, for the matter, there is alot more examples) is economies of scale (in many respects). Unless there is something what Sparc can do that X86 cannot do. (Which I would not be able to see).
From my point of view it is not a question if that happens but when. Today im fancy enough to lean far out of the window and predict that SPARCs will be thrashed at about the time IPF will be canned. To prevent from misunderstandings: Neither architecture will be given up officially, but factually.
K.
Keith
those numbers are derived from
some place between my ears.
K.
paul
FX is already on 128bit Channel, so it will be only improved by the reg-factor on 939 pins. As for 128 vs 64bit Channel, could be i underestimate the effect. Lessee, we will know more in two months time.
K.
Keith,
well then, from 0,6 to 1 Million would be an increase in the dimension of what IDC has. Are you assuming these numbers? K.
paul
128-bit MemoryChannel is not a holy grail. We have seen in the K7 architecture an intelligent Chipset for SingeChannel memory that overcompensated any DualChannel-advantages.
Maybe it is good enough to use the odd numbers (3100+, 3300+,....). Nothing much more, i guess.
As for "754 relegated to Celeron status", you are kidding, right?
K.
Keith
Thus, the 72% number I brought up is indeed correct.
Well, if the IDC number is indeed anywhere close to reality, Q2 must have been a desastrous mobile-quarter for AMD, would you agree?
K.
doug
At $215, they are going to generate a TON of demand for these parts, even cutting into sales of the 3200+ parts
Well, from what is perceived out there it could well be that XP-3000+ systems and maybe XP-3200+ systems as well will be sold out at Christmas upon adding Athlon64-3000+ to AMDs price list.
K.
doug Athlon64 CAAMC
http://www.watch.impress.co.jp/akiba/hotline/20031213/image/na634.jpg
K.
sgolds
Don't expect to see a writedown for excess returns
Sure I dont. With a lil bit of a stretch, they can book in on a development budget.
K.
chr p
I neither dont "know" what Hector and Scooter negociated in detail - but thats not really important imo. However, i share your empathic perception of Scooters moves.
As far as SUNs future is concerned, I dont think its too late, SUN still enjoys excellent reputation in the markets. As soon as AMD-64 has enough traction (which will take two years or so), it still will be a painful step to trash the Sparc-Architecture (will probably called "transition" then, but its gonna be unavoidable for SUN to survive as an independent company.
In the meantime, SUN will open many previously closed doors for AMD. This collaboration is a good thing for both companies - as you pointed out.
K.
elmer
I don't expect many to be returned.
Absolutely d'accord: Transportation and disposal of electronic waste is costly. I know, you know, Intel knows, Evaluation Customer knows. That's why the vast majority of these systems will remain with Customers.
Again, call this whatever is convenient for you.
K.
chr p
Well.. from what I absorbed from your thoughts I guess I would agree in general, although your posting is about as "organized" as SUNs strategy.
K.
elmer
Even if this is not your field, maybe you have some idea of the value of system Intel takes it back after "evaluation".
Feel free to call it whatever is convenient for you.
K.
gb
For those not familiar with how it can be done perfectly legal I gave an example in brackets.
As restrictions of your valuable time obviously did not allow to read all of my posting - thank you for taking the minute to teach me business law.
K.
elmer
Hmm. How long does an evaluation of a server system take?
K.