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.
Phil, re: What is for certain is that HP celebrated the arrival of new Opterons, and IBM celebrated the arrival of new Xeons. The vendors have picked their sides.
That's bull. HP intro'ed a whole series around the new Intel Xeon cpu's too. It coined the term 'chipagnostic' to describe its stance. That term has been copied in about 4 articles I read. It's underlined by the many news release this week about their new and upgraded Opteron and Xeon lines. Ofcourse they're proud of their Opteron products - they will get them a few more enterprise customers that want the choice (most happily continue to stack their future on Intel chips like they did for a decade or two already).
Regards,
Rink
Re: Only Chipguy sees Itanic through rose tinted spectacles.
Nope for the moment (until IBM comes up with something way better, if that happens that is) I see good future for Itanium too, at least within the high end server market. HP doesn't have the option to let go of Itanium even if they wanted too, because they sunk their ships behind them. SGI's sales are pretty healthy too.
The 64b Opteron and later the 64b Xeon did ensure the lower end of the server market would stay x86. That's all currently.
The future is not exactly entirely carved in stone already.
Regards,
Rink
Keith, re: What I don´t understand btw. is why you only seem to be interested in single-core performance although the point of Yonah is, well, dual-core performance.
Nope obviously I'm interested in that too, but I'm not that sure about it. When compared with Dothan+Sonoma I'm guessing single thread performance will increase somewhat especially with FP. Multiprocess performance will obviously increase very meaningful, like 50-70%(?). Big WAG: Maybe office performance will increase 15%, games 30%, and specifically multithreaded software 50%? How much do you think it performance will increase?? Do you think Yonah will perform better than top binning A64 when it is released? If so in what area's?
Regards,
Rink
sgolds, fwiw, I agree with chipguy that I2 and Montecito are succesful in taking over HP/Compaq high end, plus take some SUN accounts. This trend will only accelerate with Montecito. Secondly x86 continues to take some share from previous UNIX accounts. This long established trend will also continue with dual core Opteron, and later on with a server version derived from Yonah. After Montecito and after the first line of dual core Opterons that's where it gets a bit foggy. It's my opinion that Intel thinks that by 2007 they might have another chance at unifying at least the platforms but besides that I think that they think they have a another shot at unifying the architectures because by then they've estimated that Itanium becomes cheap enough to emulate x86 cost effectively with a significant bit to spare. This might never happen though. I'm not sure which way it will go and probably won't be for another year.
Regards,
Rink
256b bus at current speed I think I heard (eom)
Chipguy, re: That's an advantage? I2s wallop Opterons on the type of workloads that people buy Altix for even without an integrated memory controller. And Altix memory capacity can be configured completely independently of the number of CPUs. Try that with Opteron. :-P
Sure, and I2 systems wallop Opteron systems in price as well. Evidently for a lot of HPC applications Altix is the best choice, but not for all. From what I heard there are applications too where inter cpu latency and load latency do matter, for others x86 compatibility matters, for others still price/performance for a specific application matters, etc... So it's not like the existance of Altix means there is no market for Opteron in the 128-10000 cpu range.
Also you conveniently forget to mention that Xeon would have made a much poorer choice in this market because of what it offers in latency, general performance, power, average bandwidth, cost, etc... So it's pretty conceivable that Opteron was actually the best choice for designing a x86 system of between 128-10000 cpu's.
Seperately I think Cray recently got a bit more tracktion with better availability of both ot its Opteron lines. Add this to the release of dual core Opterons somewhere during the summer which will help both Cray and Opteron Cluster vendors in general and you might find there is some reason why Opterons share in high end HPC might actually increase this year and into 2006 (just like I2's share will increase as well because of Montecito, while Xeon's share might actually decrease a bit because of lack of dual core).
Regards,
Rink
Chipguy, re: "in the Red
Storm/XT3 architecture Cray engineers only use HT as a
glorified FSB to connect the processor to their ASIC which
does all the routing function. "
True but they do use the on die mem contr, 48b addressability, and 64b computing. Intel's high volume server cpu's still don't have the on die mem contr which is an important feature for Cray's 3d communication architecture to function properly. (All cpu's without mem contr on die have a component and most likely a latency disadvantage).
Regards,
Rink
Chipguy, so IBM is on 'welfare' dollars as well?!
Cray got the money because one of the criteria for spending those dollars is scaling systems based on HIGH VOLUME components, besides innovation and plain value ofcourse. SUN got it for wireless approximity communication iirc. Don't recall why IBM got it... Still this might indicate that although SGI obviously wanted the money (why else apply for it!) they didn't get any because their system components are too low volume and its innovations are too evolutionary and there was not enough value for money from a DOD point of view.
Regards,
Rink
CJ, does your 64b ndiswrapper use a 64b windows driver for wireless or a 32b windows driver for wireless (my system is dual boot too, 32b windows xp and 64b linux fc2).
Tx,
Rink
wbmw, FYI, AMD's official aim was 10% of server market end of year 2004 (not 10% averaged for all year). Also, afaik, none of the analysts last year repeated the word 'average' in combination with '10%' within this context. I read the articles you referenced too and came to a very different conclusion and am convinced your math for Q on Q server growth is wrong. That said I too find it odd that those isuppli numbers show negative opteron growth month on month during Q4 for some major west european countries.
Regards,
Rink
At least once (eom)
HP and Packard-Bell systems in The Netherlands, plus Acer laptops.
Regards,
Rink
Keith, re: The differentiators available for AMD are fairly limited. So what´s the strategy? I can think of some cache games, socket games, but that´s basically it, and not convincing to me.
Agree. Strategy seems clear to me though: Maximize ASP per cpu segment largely defined by Intel. Differenciators are limited but useful: 64b/32b, power, single/dual mem. channels, HT bus speed, # of HTc ports, frequency, cache. Each has their limitations compared with Intel having different cores, but they make the most of it for as far as I can see. I'm wondering if there will be any new non A64 technology in Turion.
Regards,
Rink
Keith, it's more like: "I'm reluctant to let go of the general believe that Intel is by far the best cutting edge design and technology company in the world. It has been for decades, and will be for at least another couple of decades. Yes I'm an American, and damn proud of it too."
Regards,
Rink
Joe, they have 64b Win + Linux LAN drivers available on their website for more than half a year already IIRC. It's the wireless drivers that are lacking, especially the 64b ones.
Regards,
Rink
CJ, exactly why it pays to buy NVIDIA chipset+gpu if you want Linux, especially in 64b mode. For wireless I've got the same problem though. Everything incl graphics works in my dual boot laptop setup (WinXP 32b, FC2 64b) both under Windows and under Linux, except in Linux the wireless (broadcom chip) connection can't be set up. I tried compiling the ndiswrapper in 64b using gnu for the heck of it but it ofcourse didn't get the Win 32b broadcom driver to work for Linux.
BTW, I expect the new nforce4 boards to become rather useful for the Linux minded (I've got a nforce3 laptop board).
Rink
RE: I gathered from AMD´s analyst day that 130nm K7 Sempron will continue to be produced for the not-so-far-developed countries in H2/05, and maybe beyond. Probably the most economical choice.
...but produced by UMC for as far as I can tell.
Regards,
Rink
Chipguy, Article: HP, Intel move means SGI running scared.
M++, Tukwila, Project Z et al
http://www.theinquirer.net/?article=20304
Quite an interesting article, even though from the Inquirer. I was wondering if and how far and why you'd disagree with them (as I think you might at least for a bit).
Tx,
Rink
Chipguy, mmm, I somehow thought you'd have a bit more info than that, specifically where it relates to next gen Itanium chip sets. Anyway, if you're saying that Montecito can't go 16-way then that would be an architectical advantage for Opteron.
Regards,
Rink
Chipguy, Couple of questions:
Would you have a ball park idea how a Horus based dual core 16 way Opteron system compared to 16 way Madison and 16 way Montecito?
Secondly would you happen to know how Bechtolheims Kaelia Opteron systems will differentiate themselves from competition (Horus based systems)?
Lastly would you know anything about the Serverwork chipset that Broadcom/AMD/SUN are creating (heard they have working first silicon but do not know how it's supposed to be different from Horus).
Thanks, regards,
Rink
Jules, re: avg. $4.60... Good timing! I invested in AMD the first time from '98 or so to '00. Second time I was less fortunate as I invested at $10 just before it went down to approximately where you bought it. The shares bring some money in now (but not that much because of the declining dollar). My LEAPs are doing much better.
Regards,
Rink
Jules, my boy, don't ever just believe what artii says about chip manufacturing unless he mentions the facts. He doesn't have the background himself to make any such statements, nor the insight in either AMD or Intels processes. Hence he can only repeat what others say. Again, don't get me wrong: I can also only repeat what others say but I hope you will note I try to collaborate it with some reasonable logic.
I think APM is 'only' a major differentiator for as far as it concerns competing with Via or the like. I also think that APM is not nearly that big a differenciator when compared with the chip manufacturing control suite Intel uses. The reason for why I think that is that Intel often manages quick ramps, quicker than AMD that is. Remember:
- 130nm,
- 130nm SOI,
- 90nm SOI to a lesser extent but you can't claim it's going that fast just that it is ONLY half a year behind Intel which is an improvement in itself, and
- although the problem was probably a bit different in nature mirrorbit scaled extremely slow so far too but I think this last problem might be solved as AMD in fact claims.
Just a couple of examples.
Yes, AMD talks about APM as being great, and yes it is in fact likely to be great, but what exactly makes you think Intel is that far behind in process control and doesn't have anything even close to being as efficient as AMD's suite?? Just the fact that they are not as vocal about their APM suite as AMD is? Can you please state your reasons for why you think AMD's APM results are so much better than Intel's, or a link, or anything else than just blankly believe artii that is?
Regards,
Rink
Reference 1: Here is what artii said: With APM, AMD has been able to bring new changes and enhancements to market at a blistering rate. [Any evidence, or even anacdotes??] They can identify a problem well before a wafer is completed rather than having to wait till it pops out at the end of the line to test it [very generic statement to say the least] . Did you know it take 2-3 months to complete a wafer? 2-3 months till you find out if a change worked or not! [Note here is where artii implies Intel never does in-line testing or never close to the really superior way AMD does in-line testing] AMD has found a way around a lot of that wait and the results are earth shaking.
Reference 2: http://www.reed-electronics.com/semiconductor/article/CA420774?pubdate=6%2F1%2F2004&industryid=3...
Just take this marketing like the "we are better than them" and "APM is 68% better than no APMe" with just one tiny grain of salt too. Please note the chart doesn't mention a timeline (high volume product vs. low volume; completely different time lines). No data on either horizontal or vertical axes. Lastly note it's compared to previous generation results: Intel also ramps better all the time; I've seen charts of that too. And normally we don't get to see any hard data either... This typical example should show that the information that is available isn't exactly telling, and CERTAINLY not enough to state that when talking about both AMD and Intel that AMD's APM results are "earth shaking", implying Intels are far behindt.
Jules, FYI: artii does not have any meaningful in depth knowledge or insight (my opinion), and this post is an example of that. Please do note I'm not claiming I'm that much more insightful, but you might want to consider the below points that I hope balance artii's post a bit.
Yes AMD is ahead in SOI, HT, AMD64 a bit, Mirrorbit, etc..., but it should at least be balanced with the fact that Intel is at least 6 months faster with next generation processes and ramps them fast, has 300mm, redundant and geographically spread fab capacity, has better strained Si, has way better economy of scale, has Montecito and Yonah coming, and has the money and cloud that comes with being THE industry leader for decades. Plus not all of Intel's CPU's run hotter on 90nm (just the Prescott/Nocona design).
Itanium might actually still work good enough magic for Intel. McKinley (now Madison) is the successful big tin cpu ever for them. Partially because its performance is good, partially because HP burned all their old ships, because SGI has 512 cpu single image systems, because SUN does not compete performance wise, and because a lot of verdors and customers alike like following the top dog. Plus the Itanium roadmap is strong with Montecito coming next (really excellent performance at rather reasonable power consumption rate). Furthermore Intel is likely to have something close to APM (remember they typically ramp new products faster than AMD). This is just to show it's not logical to imply the fight is over now based on current fundamentals. Even the opposite: With Madison now, and Montecito coming Intel has again space that AMD can't reach into as effectively as Intel is doing. And as the design of Dothan/Yonah/etc.. isn't standing still they have a good competititve advantage in the important notebook segment that will be extented to the desktop too. So currently I think it is normal to assume that Intel does have good means to gain back some of what they are loosing now, starting 2006 at 65nm.
Pls don't get me wrong: I'm long AMD because the market is clearly shifting bit by bit in their direction (now and likely all of 2005, may extent into 2006 and beyond but that's really far from certain).
Regards,
Rink
Hellraiser, FYI Buggi is one of THE most valuable resources on this board (you will find this is not only my opinion). He links analyst reports on his site (that you reference without mentioning the source) and provides good insight. Could you please cool it? He asked you a normal question, that is in line with normal behavior here.
FWIW, I know you both from here and from Yahoo, and value your input too. Don't think you will but there's actually a slim chance that you remember me from yahoo as I jumped that ship about a year and a half ago for SI and this board.
Regards,
Rink (Ixse)
Ref: http://www.investorshub.com/boards/read_msg.asp?message_id=4567756
Elmer, it's more like bzzzt ouch! bzzzt ouch! bzzzt .....
Though history often repeats itself it is in fact created by mere men. Could it be that you forgot to mention that you think there is actually a 5-10% chance this time that there is a group of people that changed the rules possibly enough this time to make the last part of that sentence read: bzzzt BINGO!
Provided you do not in fact know any current inside Dell info very high up the management chain your remark lacks any balance, but then you know that too.
Remember this could be one of those times you're just wrong. FWIW, I like your information on Montecito and big tin cpu's in general much better because it's more based on current facts.
Regards,
Rink
PS, I think that there is a 25-33% chance (significant but not very likely just yet) that Opteron might make it to Dells business line before end Q3 05.
(deleted; same ? as Joe)
Weeelllllll, I'm ok that you're right :)
I've got the dollar problem too. Today I sold most of my other US stocks as the dollar is projected to go further down, and I wanted to reduce my exposure to it. Sucks. Still made good money on that other account too. AMD made me more money though and I just will continue to ride it until at least end spring next year.
Regards,
Rink
Keith, yes he does give his own reasoning. That reasoning is however the same as what I think that most of us on SI thought was Ruiz reasoning last spring. Secondly I'm quite sure that I've seen Stahlman repeating this Ruiz remark before at least one quarter ago. This time he is more definitive though; I like that.
Regards,
Rink
Keith, I saw that Dell comment too, but believe Mark Stahman is essentially just repeating Ruiz, just like he did before with regard to this remark. I believe Ruiz last spring comment was related to AMD being first with dual core x86 server chips.
Regards,
Rink
Chipguy, do you know what technical element causes the TDP of Montecito to be lower than both the 6 and 9 MB Madisons?
Thanks,
Rink
Yes (eom)
Athlon 64 Revision E adds SSE3 Support. This is next revision of the 90nm A64 core.
Also:
"On the Intel side of things we heard the name Cedar Mill thrown around a bit more. In the past some have referred to Intel's dual core Pentium 4 as Cedar Mill, but we now know that is Smithfield. This time we heard Cedar Mill referred to as the first 1066MHz FSB Pentium 4s that aren't Extreme Edition chips. Cedar Mill is supposed to be out in the 2nd half of 2005, which supports a claim we made in our recent review of the Pentium 4 3.46EE: "the 3.46EE will be followed up by the 3.73EE as the only two chips to support the faster FSB for almost a year." ...
When Intel talks about aggressive ramps of new products there's usually little attention paid to exactly how they're going to achieve such an aggressive ramp in product shipments. For example, it turns out that the market for the 915 chipsets is very soft, so motherboard manufacturers are finding that selling LGA-775 motherboards equipped with the 865PE chipset is much more in tune with their customers' desires. The result of this is that shortages of 865PE chipsets will occur (such as the one going on right now) and the only way to get more 865PE stock is to not only purchase more 865PE chipsets, but also more 915 chipsets. It has to work this way because otherwise the motherboard makers would take much longer to transition to new platforms and that wouldn't do so well for chipset or even CPU sales, and since Intel holds the power in the business relationship things like this can happen.
The problem is that right now, no one wants 915 motherboards - they simply aren't selling well at all ...
Soon there will be no more Socket-478 Intel CPUs left on the market, with the only remaining chips being LGA-775 based. Here's one of those aggressive ramping situations. By getting rid of Socket-478 Pentium 4 CPUs, Intel ensures that the only Pentium 4s you'll be able to buy are LGA-775 chips. When you buy a LGA-775 CPU you'll most definitely need a new motherboard, and with most LGA-775 motherboards being 915 based, there's a high likelihood that you'll find yourself buying a 915 motherboard as well. Then of course you'll need DDR2 and a PCI Express graphics card, so the memory and the graphics card makers benefit as well. But the chain reaction will take place tomorrow, as far as a solution today goes, unfortunately for the motherboard manufacturers - it's going to be a tough few months.
The next problem is that once 915 sales begin picking up next year, 915's replacement will be just around the corner - adding 1066MHz FSB and DDR2-667 support. I wouldn't be too surprised if Glenwood and Lakeport get pushed back to late in Q3 of 2005 rather than towards the middle of the year to at least somewhat better accommodate the motherboard manufacturers. Now you can begin to see why introducing a 1066MHz FSB variant of the 915 wasn't an option for Intel; with so much unsold 915 inventory, the motherboard manufacturers would be in a very difficult situation if they were given a 915E to sell as well.
On the AMD side things are much simpler; just about every single motherboard manufacturer has a nForce4 solution for AMD as their high end Athlon 64 platform. In fact, NVIDIA is quickly turning into the Intel of AMD chipset manufacturers, which is something we've been asking for ever since the introduction of the Athlon.
Although there is a lot of support for ATI's upcoming chipsets (you'll read about them here next week), almost all the manufacturers were saying that their ATI products will be Intel-only. The worry is that with such a strong competitor in the Athlon 64 realm that their ATI products won't sell; ...
See page 3 and 4, Keith's link.
Regards,
Rink
Tx mas (eom)
SUN Solaris 10 incl AMD64 support launches Nov 15 (Keith / SI): http://www.siliconinvestor.com/readmsg.aspx?msgid=20705918
Regards,
Rink
U&D, re: I take your post as meaning that Intel's new roadmap shows they've responded to the AMD threat.
Somewhat right yes. I see it likely that Intel's roadmap for the next really fully new generation of cpu's (not Smithfield or some other kludge like that; I'm talking beyond dual core which is rather likely an advantage for AMD; 2006/2007 timeframe) will have at least some real answers to AMD's competition. None of us (for as far as I know that is) knows exactly what that answer will be. There is a reasonable chance that it isn't fully complete as they might find it hard to catch up with the evolving hypertransport interconnect technology from AMD, but it'll (I presume that's all I freely admit) likely be at least a reasonably good answer, as Intel will continue to have their own strengths (SSE4/compiler/cloud/manufacturing capacity/manufacturing process strenghts/channels/etc...), and if everything really gets together for them and they do catch up with hypertransport 3.0 by then then AMD will have a really difficult time again. I think it's likely that Intel will come up with a good desktop answer. Hypertransport makes a good answer to Opteron successors a bit more difficult. Anyways, chances are the reality will turn out to form itself somewhere in the middle, which is reasonably good for AMD.
Regards,
Rink
Bobs, I think there is actually some chance that Intel doesn't get it's act together good enough in 2006/2007 but I only rate it at 10% for now. I'm really hoping you're right especially because I think it's a possibility but am not counting on it. Intel's current desktop / mobile strategy is almost completely based on Dothan successors. Itanium is a different story. So Intel did give the Israeli team the prime lead for future desktop and mobile cpu generations. Their weak point is they're still trying to consolidate Itanium and desktop lines onto a common architecture and they might find that hard to accomplish because of their competition. So I'm just a bit with you and hope Intel will really strangle themselves but don't dare count on it too much (I'm long until Intel shows something real for first 2006 and then 2007). Besides if Intel doesn't screw up, it's equally possible AMD might not screw up either (I've got reasonably high hopes for their joint efforts with SUN, especially if that would include capacitor based wireless chip interconnects).
Kind regards,
Rink
(deleted; duplicate)
Come on Bobs, you're not going to say you don't agree that Centrino is a tasty bone in the eyes of just about every dog are you? (just one question). Plus wouldn't you like AMD to have half the cloud Intel has? And manufacturing capacity? Market share? Margin? Channels? Just to name a couple of also relevant points... Remember the equation is rather hard to change; it's like going against a wild water river (possible, but only for the lucky salmon even when smart and in excellent condition).
Kind regards,
Rink
Yeah, depends if they hit the iceberg at above or below the waterline, right? Guess that for both of our investment strategies it's safer to assume the former (while knowingly hoping for the latter).
Kind regards,
Rink
Bobs, don't you think management is improving now they acknowledge their mistakes and put in place some focus (right from the moment Tejas got cancelled and they accepted AMD64 in sort of the same spring / end spring kind of timeframe)? They've got catching up to do. But they're reasonably likely technically capable of doing just that with the first real next generation of processors (i.e. 65nm ones that are based on an entirely new core), and management seems to push exactly that. So I agree management has made some rather bad decisions, especially in hind sight, but they're taking some very right actions (imo) now. That's at least my stance until some significant indication shows they screw up with that too. Rereading your post I think that despite the strong words you might actually kind of agree with this as you mention Intel has to get it's act together in 2006 at the latest.
Kind regards,
Rink
PS