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.
Paul, chances are low: UMC was prep'ed for K7, IBM or affiliate like Chartered or maybe even Infineon are the most likely candidates for K8 because they do SOI in more or less the same way. Samsung doesn't have (for as far as I know that is) any process tech sharing agreement with IBM.
It might be relevant too that with K8 Semprons becoming more and more available the K7 Semprons as least as we know them are becoming EOL'ed by around H1 05 (I have no idea how successful a cheap 90nm K7 would be; it's certainly not in the roadmaps that I've seen). So for the moment I think that if they are going to find a foundry partner biggest changes are it's for K8 Semprons.
Kind regards,
Rink
New (at least for me) Acer Ferrari & Aspire notebooks
Don't know if you guys saw this already (you might well have as Keith/Buggi tend to have an eye on this stuff too but I didn't see it so...): Ferrari 3400+ / Aspire 1362/63 /Aspire 1522LMi. Different screens. Different graphics. A64's, Mobile A64's, Mobile Sempron's too. Link: http://www.siliconinvestor.com/readmsg.aspx?msgid=20683541
I think some of the Aspire's are good deals like 1363LMi (low price K8 Sempron), the 1522LMi (good Mobile A64), 1513LMi (highest performance A64 notebook at reasonable price) are good deals.
Their lineup is getting reasonably comprehensive now.
Kind regards,
Rink
Kind of scary but our numbers may just be the same even though I think the numbers are only temporarily finite.
Anyways here's some more on Emma tracktion:
AMD's $249 PC
p2pnet.net News:- On the heels of Microsoft ceo Steve Ballmer's statement that a cheapo PC might solve some of the company's piracy problems comes the announcement of a not-quite-so-cheap PIC from AMD - aimed at the same emerging countries.
'Emerging' means the inhabitants are about to come out as consumers - that's to say, people worthy of serious attention from the corporate world.
AMD is thinking Personal Internet Communicator (PIC) at $249 all-in while Microsoft believes a $100 PC would probably do the job.
There's another difference. And it's a big one.
It would appear Bill and the Boyz are still at the ruminative stage while AMD, "will unveil the PIC next week as part of its '50x15' initiative, in which it hopes to bring Internet connectivity to 50 percent of the world's population by 2015," says an InfoWorld story, continuing:
"Local service providers will be responsible for the maintenance of the system, AMD said. Those local telecommunication companies or Internet service providers will actually brand, market, sell and support each PIC."
InfoWorld says the PIC is a small form factor desktop with a customized Microsoft Windows OS and basic application software.
"Most of the software settings are locked in before the system ships in the hopes that users won't break any applications, and service calls can be kept to a minimum," it says, adding:
"AMD's Geode GX500 embedded processor powers the bare-bones system. It also comes with 128M bytes of DDR (double data rate) SDRAM (synchronous dynamic RAM), a 10G-byte hard drive, four USB (Universal Serial Bus) ports for the USB keyboard and mouse and a monitor."
Stay tuned to see who'll be next.
http://p2pnet.net/story/2784
Kind regards,
Rink
Keith, darn I completely overlooked it... (eom)
I think the Apple msg is fabricated. I've seen several stories that had Intel's market share down by .8% while AMD's market share rose .5%. Plus I couldn't find the quote you linked on google.
Kind regards,
Rink
Anyone know if that Sempron 3000+ is a K8 ??
Kind regards,
Rink
Keith, I agree. Still I'm also happy to see they weren't discontinued. That said it's also true that I expected Gateway to be in this situation two months ago already.
Kind regards,
Rink
re: but $14 remains a real obstacle for some reason
maybe not for too long...
Tx mas! (eom)
One remark on AMD 90nm: 1.5V is higher than originally said it would be, and higher than current Newcastle parts I believe as well. So even though power is apparently lower I think AMD still has an issue here that isn't solved. Once that voltage goes down power will go down even more.
Can't be sure but my gut feel is AMD can solve this issue.
Kind regards,
Rink
Blauboad, for as far as performance is concerned BTX is for at least three quarters of a year a non event - Intel isn't going to introduce any 4+GHz part. BTX primary reason d'etre was that it allows for better cooling. Now there isn't going to be a 4+GHz Prescott / Xeon this is only potentially useful for dual core systems and those won't arrive until H2 05.
Kind regards,
Rink
AMD expects that revenue from businesses will account for 10 per cent of its worldwide sales next year. It works with more than 2,000 manufacturers and partners worldwide to create more products that can use 64-bit computer platforms.
I think that's an error.
Rink
Tiger, your figures are well exagurated, and equally wrong if you meant it to be a generic remark. Though I fully agree with your remark about their investments in the future are limiting their current profitability.
Kind regards,
Rink
Huh, are you saying Chipguy can be wrong??
Keith, AMDzone has a shortened transcript that got posted directly after the call.
Kind regards,
Rink
Chipguy, congrats on your dual core Opteron performance projection.
Kind regards,
Rink
Mas, just to ask if you forgot Montecito??
All big tin processors I've seen so far have a different development cycle, getting to the next gen manufacturing process about half to a full year later, with less intermediate speed bumbs (like is so common for desktop processors). Itanium followed this pattern too so far (remember Merced?).
On the other hand I do agree it's remarkable that the desktop processors are currently so close to the big tin ones. And even though Montecito will change the performance equation rather significantly I think Opteron will remain a serious consideration even for floating point because of die size, economy of scale, power, and prospects of industry standard. Industry standard not only with AMD64 instruction set, but also because its direct connect architecture that enables a good variety of scalable solutions (latest example is Cray's XD1 further encroaching on Itanium's turf).
Kind regards,
Rink
BTW, just being curious: do you post on Aces and RWT too?
Chipguy + Alan TX (eom)
137-140mm^2 (eom)
Chipguy, would you be willing to share similar MDR numbers for Nocona and Prescott too?
Kind regards,
Rink
From the same link: Solaris 10 on Nocona.
Not remarkable, but just shows they are agnostic enough not to be loyal to AMD when that's what their customers want.
Kind regards,
Rink
INQ quotes SUN saying "Itanium server shipments declined quarter on quarter"
http://www.theinquirer.net/?article=18102
Wonder if it's true, but if it is...
Kind regards,
Rink
Buggi, from the C't article: Der erste Prozessor namens „Oakville“ soll nur halb so groß sein wie seine 130-nm-Vorgänger - was bei einer linearen Skalierung von 130 auf 90 nm ja auch zu erwarten ist - und er soll nach Insider-Informationen bei gleichem Takt kaum mehr als die Hälfte an Energie verbrauchen. Statt 62 Watt wie beim Mobile Athlon 64 3400+ (mit 2,2 GHz) ist von 35 Watt die Rede. Und mit Powernow! lässt sich der Verbrauch im realen Betrieb noch deutlich reduzieren....
Quick & dirty:
The first cpu called Oakville should only be half the size of its 130nm predecessor - like what was to be expected with linear scaling from 130 to 90nm - and it will according to inside information only consume half the amount of energy at the same frequency. Instead of the 62W of the A64-M 3400+ (2.2GHz) it's now said to be about 35W. And with Powernow it'll consume significantly less under normal usage conditions....
Another one: Dem eigentlichen Engpass, dem geteilten Speicherzugriff, will Intel jedoch erst ab 2006/2007 auf allen Plattformen mit integrierten Speichercontrollern begegnen - hier hat AMD also drei bis vier Jahre Vorsprung.
Quick and dirty:
Intel will only solve the actual bottleneck - shared memory access - starting in 2006/2007 with integrated memory controllers on all platforms; here AMD has a 3 to 4y lead.
Both tidbits (half power at same frequency, and Intel only projected to have integrated mem contr somewhere halfway down 65nm) are rather good news.
Kind regards,
Rink
Paul, yep, it was too much in front of my eyes, so I didn't see it as I was focussed on longer distance. Tx anyways.
Rink
That's not news: I posted one a couple of months back from Walmart (also AMD based).
Kind regards,
Rink
CJ, re: Very undocumented.
I agree. Indications are positive enough though to have good hopes for better (lower power equal performance) mobile parts soon, and better (lower power equal performance) desktop parts soon after, while better cutting edge (same power higher performance) parts might be due only at year end / q1 05 though. Good enough for me for now. All these items will need solid confirmation for me not to get very disappointed.
Kind regards,
Rink
Buggi, tx, 2nd link in this post works for me (eom)
Paul, I hadn't read Buggi's info on XPM's that are K8's in post #42523. I can't read it (behind a firewall that doesn't like the server it's posted on) so I'm still not sure.
Kind regards,
Rink
Paul, per request here's the original email, so you can decide for yourself:
Dougs email to Mike Haase:
------------------------
Dear Mike,
A number of us have been wondering about AMD's last two posted roadmaps, wherein the "Mobile Athlon XP-M" is indicated to be on the "130nm SOI" process.
Is this a typo?
We were under the impression that the "Mobile Athlon XP-M" parts (at least the K7 variants... I realize there was a mobile "Dublin" XP-M used by HP in March based on the K8) were on 130nm bulk (non-SOI) process.
Any chance you or someone else at AMD could comment?
thanks again,
Doug
-------------------------
Ref: http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=20423361
Paul a.o., you already know the answer to this mail as that was posted earlier. I agree with Doug that by far the most logical conclusion is that K7 is on 130nm SOI, although I also think it's not entirely conclusive just yet. I agree with you that this is at least unexpected.
Kind regards,
Rink
sgolds, Did you see this newer AMD roadmap? (from pc.watch.impress.co.jp so not quite official but still...):
It shows Palermo pulled in half a year.
(Also I missed it completely before on AMD's june 14 roadmap - the one you posted - it shows K7 core has been moved to 130nm SOI in the form of AXPM)
Kind regards,
Rink
Duke, I think you miss the point here. You said Solaris isn't available on x86. I said it is available on x86 and shipping with Opterons (and Xeons too for that matter).
Please note that this is the 32b version of Solaris for x86 that was resurrected last year. The 64b version is expected around the end of the year, and your post about it being 6 months out might well refer to this 64b extention of x86 exclusively.
Ofcourse once the 64b version is available for Opteron it is '15 minutes' away from being available on 64b enabled Xeons too.
I think the market is and will continue to prove you wrong though on customers not buying the 'expensive' sun servers to run it. Fundamental reasons for this are partly because of attractive bundling, partly because of service, because of very good software availability on their top notch version of unix, and also because Sun is a well regarded first tier company.
Kind regards,
Rink
Duke, re: Sun knows the DAY sunw ports Solaris to X86 is the day they are out of the hardware business. It is the Apple business model.
Sun's Opteron systems are available now with Solaris for x86.
Kind regards,
Rink
Yes, but we have the finest diet colas.
If that would matter then why are americans more obese on average?
Chipguy, re: How is Dell going to keep up with HP's 2% growth in x86 server sales unless it matches it with Opteron offerings
too?
Good point. I thought that was about the only weakness though in the HP/Dell earnings + GS report news yesterday.
What do you think the reason is why they're doing so well? I think it's ultimately their pricing which is based on their low cost direct model (plus their untarnished image as low cost enterprise supplier of everything from notebooks to low end servers). If so HP is't able to match the reason why Dell is gaining. Hence they offer Opteron in order to have some differentiation compared with Dell in the standard server market.
Even considering this I think the HP results for standard servers was very disappointing, while DELL's result is beyond good. I don't think it is anywhere likely but I'm neither dismissing the possibility that HP rethinks a lot of it's options. One indication of what they're thinking will be what units they reorganize as result of the losses.
Anyway you're right in that these results will not be likely to pressure Dell materially into offering Opterons. Even worse it even seems to vindicate their Intel only strategy too. And this is unlikely to change untill AMD's dual core product plus an always unlikely coincidental Intel snaffu like e.g. not having a dual core server chip as dual core Opteron competitor in more or less the same timeframe (could happen though; I give it a 40% probability). I also think this was the basis for Ruiz' remark about thinking that Dell would be on board within a year (i.e. start of summer '05).
Kind regards,
Rink
Chipguy, HP announced they had moved already (among other things management changes) to fix the problem. My point is the problem is hard to fix, especially not with a management shuffle. I think their only way out of losses for now is plain old cost cutting. We'll see if that's true within 2 months or so I think.
Kind regards,
Rink
chipguy, tx for the presentation + the mention that IPF is included with PA RISC in HP's Unix category. It confirms what I thought when I read the HP results article: Itanium showing good growth by itself, while the rest (Alpha, Non Stop, and PA RISC) is declining faster than Itanium growth can make up for (similar to sgolds point of view). The problem itself though is hard to fix. Basically it's currently kind of hard offer Itanium in a broader market than they're currently doing as the lower end of the market is efficiently commoditized by x86. They can however offer the updated alpha at a price the customers can't refuse as an eol gift (though it would influence their Itanium sales it will stop the drain there), and sort of do the same thing with Non Stop (which would have the same effect unfortunately). They can also stop growing their Opteron empire but then their customers for AMD64 will partly move elsewhere (don't completely dismiss it as a possibility though). In other words it's gonna be hard for them to come up with a solution other then just plain old cost cutting (probably in products that are going to be eol'ed anyway), right?
Kind regards,
Rink
@Buggi, re: $200M payment
Thanks for your thoughts. I think we'll need to ask AMD how the payment is structured and if it would be appropriate to expect $20M / q payment on average. Does anyone have an email adres for a question like that? I'll be happy to ask the question and see if we can get a reply.
Kind regards,
Rink
CJ, re: That one works well, but only in 32 bit mode. I gather it isn't completely 64 bit clean. In addition, it needs to use a Windows driver, and it's a 32 bit only driver. Sigh. I am generally stuck running 32 bit Linux because of this...
I'm thinking along slightly different lines. If you've got XP64 beta installed with beta broadcom wlan driver, do your own gcc 64b linux compile of ndiswrapper from the source code, it just might fly. EDIT: I've now checked for 64b broadcom wlan drivers and haven't found them, hence I don't have a solution either - sigh... No chance I think that the win32 drivers will do the trick, but one of these days I'm gonna try anyway.
But then I knew I'd be stuck to dual boot for quite a while. Except from Win XP home all software is free (for now).
Kind regards,
Rink
CJ, maybe I sounded too enthousiastic, but as I just had read about this ndiswrapper and was thinking it had some potential to help me with getting my wireless to work (haven't tried it out; nor do I understand it fully yet I must admit; just think it may turn out to work). Have a look for yourself. You probably are able to make sense of it faster than I can...
http://ndiswrapper.sourceforge.net/
Kind regards,
Rink
Gollum, re: I saw that acer also has an A64 model, the Aspire 1510.
That's the one I bought the first day it became available (1511LMi). Good machine too. Have it running dual boot. Everything works under 64b linux (Fedora), incl great graphics (DX9.0b) and now I might just have found a driver for broadcom wireless too. Good value DTR, incl DVD+-RW. If you want better than XGA screen this is not for you though...
Take a look at the Ferrari 3200 too (it includes the mobile A64, better SXGA screen, slightly better graphics but from less linux minded ATI, but is more expensive).
Kind regards,
Rink