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.
mas,
AMD probably needs to forget about making dual-core an FX model and have a new range, say the DX range which spans more than one model.
They seem to be going about it differently, if model 4000 is the indication of future trend.
Joe
Keith,
HP made the switch to S939/PCIexpress in their Pavilion desktop line, finally in the US too.
Excellent news.
What nVidia needs here (to not lose the leadership position) is the 6200 TC graphics integrated in the chipset.
<edited>
Joe
Keith,
It looks like AMD has finally accelerated the 90nm wafer starts, better then my optimistic scenario (up the thread).
I have a feeling that we have almost an identical scenario here as we had with Tbred A and B, with Tbred A and B being replaced by steppings Dx and Ex...
Joe
NaS,
I think the 90nm premium will evaporate (well, in 1 quarter the latest, not be overnight) based on the % of 90nm wafer starts Hector was talking about.
Joe
sgolds,
It seems that Microsoft is targeting end of Q1 for the release.
Joe
Athlon 64 3500 (2.2 GHz, not overclocked, Newcastle, 130nm)
Depth = 12
Time = 3:00
Nodes = 52,368,338
Nodes/s = 290,563
Joe
mmoy,
While I am not sure what exactly you are trying to do, one thing I do know is that if you ghost from a smaller drive to a larger drive, Ghost will increase the the size to fit the larger drive (with all default settings). There may be additional customizations available.
Joe
wbmw,
what do you think AMD can do with Turion that would vastly improve it over the current 90nm CPUs? Surely the benefits won't all be from a miraculous process improvement.
Outside of process improvement, AMD can increase L2 to 1 MB, lower voltage by one speed grade and lower voltage a bit, keep prices attractive (not very difficult vs. very high Dothan prices) and of course, it is 64 bit capable.
Overall, Dothan will still probably lead, but AMD can narrow the gap and win some market share as a result.
Joe
wbmw,
Excuse me, but this is retarded. I have never had my computer reboot so often over the space of a few weeks
I was wondering about the same thing. And generally, when there is some problem (and I don't remember I had the last time the computer reboot) it is a problem with software, or potentially with interaction of software with drivers.
Joe
wbmw,
Funny how the upgrades and downgrades come after the event, not before.
Joe
combjelly,
But choosing between the Dx and Ex steppings has been tough for them.
Isn't it mainly a function of one being ready and other not being ready? With Dx being a shrink, Ex having some changes, there is also the issue of more validation, possibly a re-spin, which may delay availability.
Joe
Keith,
It seems like a routine transaction. It says it is for Aizu-Wakamatsu, meaning not Fab-25. It would be curious to know what it is for. It could be to start purchasing of the 300mm equipment for SP1:
http://www.eweek.com/article2/0,1759,1725979,00.asp
Joe
Keith,
re: prior q. vs. seasonal thrend
If revenue overall is up, both flash and CPU can't be simultaneously really bad revenue-wise (unless the other category unexpectedly shot up <g>)
I think the profit picture may be influenced by possible revenue growth on the CPG side offset by increased expenses in both CPG and other categories.
Joe
chipguy,
Itanium failed to deliver what? Intel developed IPF as
a RISC killer and it is well on its way to doing exactly
that after a slow start in a massive IT downturn.
That's a good point. In addition to the reality of x86 eating the RISC market from below, the FUD of another Intel processor attacking from above must have been a factor in the cancellation of RISC projects.
Where it failed is in being a clear (and the only) migration to 64 bit for all of the x86 users.
Joe
chipguy,
The 64 bit Nocona shipped in mid 2004. That means Intel
decided to build a 64 bit x86 chip in 1999 or 2000 at the
latest
Firstly, I think you are exaggerating the timing, secondly, Intel denied existance of Yamhill and never planned to turned it on in absence of a successful AMD 64 bit alternative.
Deal with it.
The whole scheme (of killing x86 by keeping it 32 bit) blew up when Itanium failed to deliver, which is something you are having hard time dealing with.
Joe
chipguy,
The poster I responded claimed that
Intel planned to drop x86 in favour of IPF but was only
stopped by AMD64's success. The public roadmap
evidence I described clearly shows that Intel had long
term x86 development plans in 1997, long before IPF
shipped and long before AMD announced AMD64 let
alone garnered OEM support for it.
You can twist this all you want, Intel did plan to drop x86 by keeping it 32 bit only. Natural progression to 64 bit would kill it.
It didn't happen that way because AMD gave life to 64 bit x86 and Itanium has turned out to be noncompetitive as an overall package (price, timing, execution, software support).
Joe
Tiger64,
there is way too much manipulation going on and the government should step in...
Yawn
wbmw,
You and l_banker believe that x86 server hardware can match or exceed the performance or RAS that IPF can deliver.
I don't believe I said that. I don't have a crystal ball about relative performance levels, but I am fairly confident that x86 will continue to beat Itanium in price / performance in 1-4 way market (compoinded by the price and availability advantage of real world apps), and the performance will grow faster in the short run than historical trends.
If this is true, then you are probably right. However, outside of the processor architecture, which some can argue has no inherent performance or reliability benefits, there are still micro-architectural differences that can put IPF on a tier much higher than x86. So even if we assume the ISA is completely transparent (which may or may not end up proving true), all Intel needs to do to justify Itanium is to continue to use multi-billion transistor budgets to design a micro-architecture that leaves x86 in the dust.
While I am not sure about "much higher" and "leaving in the dust", let's assume that Itanium will have a performance advantage at expense of large die, low volume, expensive chip. How has anything changed? The trend is for x86 to gain, everything above to lose.
This is possible, but also more expensive. Even so, Itanium is meant for the high end, so higher costs can be justified by higher margins.
The high margins (mostly for OEM) is something that customers hate to pay, and will, if at all possible, avoid. This high margin business benefited greatly from some of the limitations of x86. As the limitations disappear (number crunching capability, parallism with multicore, 64 bit addressing, increased memory capacity with FBDIMM.
The effect of this on the still high margin business is similar to sea levels rising 1000 feet. While there is still some ground left to retreat to for this "high end" business, but just about all of the fertile areas will be gone, and it will be anything but high margin. High price, high cost, low volume, no margin.
This is possible, but also more expensive. Even so, Itanium is meant for the high end, so higher costs can be justified by higher margins.
But you are thinking of this in terms of Itanium moving down, but it is really a way for x86 to move up.
Joe
Klaus,
I look at the common platform as part of an exit strategy of IPF
I was thinking along the same lines. If Intel creates a scalable interface (ditching shared FSB) for processors, it will open the doors to Xeon to higher end (which now is problematic with shared FSB and awkward connections between groups of 4 Xeon). Once Xeon can scale to higher performance levels, the need for Itanium end up lessened.
As far as emulation of IA64 on iAMD64 Xeons, it could be a face saving measure and some defense against litigation, it would be just a stop-gap measure for Intel to put the Itanium chapter behind.
The bottlenecks for Xeon were lack of 64 bit linear addressing, shared FSB and difficult (expensive) way to build systems with massive amount of memory.
iAMD64 solves first problem, the rumored "links" architecture will solve the second and FBDIMM will solve the third.
There will be a 4th benefit - increased parallelism through multicore chips. So the x86 server performance is about to make a leap, extending commodity pricing regime to much higher end systems.
The result is no need for Itanium, and a carnage for high-end niche server vendors.
The customers who are willing to spend big bucks are commercial customers desiring fast database performance. x86 vendors will fully satisfy this need with upcoming leaps in performance.
What's left is the HPC (scientific / technical computing, server rendering farm etc). These customers have no money to waste. They are looking for price performance, which again leads to the king of price performance - x86.
Joe
Keith,
Why would the title say that they are manufacturing for AMD? That seems as if they are making products (motherboards) that will be sold with AMD label...
It is probably a misunderstanding / misleading title of the article.
Joe
Keith,
re: MSI to invest US$12mn to start manufacturing for AMD
Do you know what this means?
Joe
wbmw,
I've rarely seen Intel launch two microprocessor cores for the same segment in the same year
I think the desktop segment is a more important application of this core.
By the way, do you know what other enhancements are supposed to be in Merom?
According to some rumors in the Inq, Yonah is a direct descendant of Dothan core, Merom / Conroe is a brand new core. As far as the list of features, I think it is way too early and Intel is likely to keep these close to the vest. Most if not all of the technology initiatives are most likely going to be implemented, plus definitely something to make it worthy of a whole new core.
If it is a larger enhancement, Intel would need a second large mobile development team working in parallel. I'm trying to judge the likelihood of this.
I think a better way to think of this is to think performance core, not mobile core, and maybe at power consumption that it will have mobile application. I would guess that many of the Netburst core set developers would be reassign to it.
Joe
wbmw,
Is there *any* possibility that Yonah has EM64T? It seems hard to believe that Intel will have 32-bit mobile cores until Merom in 2007.
I think the possibility exists, but, based on the rumors, it will more likely not have it. The timeframe for Merom / Conroe seems to range from 2005 when first mentioned (will not happen) to 2006 and 2007. My guess is 2H 2006.
Joe
Alan,
I really doubt bin splits is the reason
It does not have to be huge to be an issue, just say one bin. I doubt this is the main reason.
I think I posted my reason for holding it back (independent of software) was to limit liability of a potential problem showing up. Intel was apparently rushing these parts to market, and it is safer to release parts gradually. Enough time has passed for this not to be an issue any more.
It seems the "flavor" of most posts regarding intel and 64b remains "it sucks" so they are not releasing it. This is just wrong.
You are probably right.
Every desktop PC I have checked at retail that comes with SP2 installed, also has an Intel NX enabled CPU. I don't think they shipped any NX /many NX CPU's until SP2 came out. Intel is not advertising this fact.
Because the stepping supporting it barely made it by SP2 release.
Another thing to think about is that all the PC manufacturers are in on this conspiracy.
I think you have to look with a microscope to find the SKU that has the support of 64 bit. Intel could just as easily enable it on all P4s to remove this from being an issue.
Personally, based on conversations I have had with Intel folks, I think there is an element of "let AMD carry the load" on this one. Don't step in until they either fail, or complete the job. It will be interesting to see how it unfolds.
Interesting theory. But I think it is a bit beyond that point. Viability of x86-64 is not an issue. AMD did have to carry the load on this one, but that is also in the past tense. The only issue is slowing it down, or at least not accelerating it, which seems where Intel is right now. So the question is why, and I think the reason is lack of 64 bit support in Dothan and probably Yonah cores, existence of 64 bit support in Prescott, a chip that Intel would rather not even have, given its problems. So that's my theory.
Joe
alan,
My earlier thought was that the power increases, which reduces the bin splits a bit, which is already a problem on prescott. This theory is somewhat in jeopardy due to the release of 3.8Ghz EMT64 product.
Sounds plausible.
This really just leaves us with the whole software infrastructure thing, which is what Intel has been saying.
Lets say about 6 months ago Intel decided to make all desktop Pentium 64b capable... they would have shipped something over 25Mu by now. How many of those folks would decide to download win64 beta? How many of those would have problems due to a lack of drivers? What would the backlash be for all those folks who bought 64b CPU's only to find out they do not currently run 64b software? How would Intel and microsoft handle all the support calls? I doubt it would be very pretty.
Sorry, it makes no sense. There are million beta's under way, and it does not mean Intel should stop selling processors that are able to run this beta. There was a beta of Windows 95, Windows 2000, Windows XP with the exact problems, but Intel did not stop selling processors capable of running those OSs.
Joe
morrowinder,
Lets put this in terms of a real user.
Let me ask you this: If there was a great opportunity for Intel to gain something (revenue, profit, market share, mind share), but it did nothing for a user of the product, do you think Intel would let this opportunity go?
How about something that arguably does nothing for the user, but could hurt Intel?
In both cases Intel will do what is good for Intel, independent of the "terms of a real user".
I and millions like me made the choice to stick with 32 bits(Pentium 4 3ghz purchased 2003). Exactly how is this going to hurt me?
Fine choice. Suppose you bought one (Prescott) today with 64 bit capability, NX non existent or disabled. How is it going to hurt you?
I just help my friend whose machine was brought down by spyware / viruses. Dell Pentium iii 1 GHz (which I recommended, BTW). The machine was in hopeless state, so the solution is loading XP and upgrading memory to 512MB. (He is a light user - web, e-mail, word).
3 to 4 years from now, such a fix / upgrade from today's Prescott would be denied to you.
So I think you are wrong in terms of Intel decision making (which is dependent on the benefit to the user, and you are wrong about what benefits a user.
Joe
wbmw,
some stockholders would argue that it's his responsibility to put the best spin on the situation until he can make things right
That's a good point. It is in investor's best interest to analyse which parts are spin and which parts are not.
FWIW, I agree with you that by now, there is no reason for Intel not to go full speed to 64-bits unless they really doesn't have anything competitive to offer. Then it makes sense to hold things back until they do. Fortunately for them, 64-bit infrastructure has been very slow to ramp. They can thank Microsoft's typical sloth-speed schedules for that.
Nocona and Prescott seem ok from the initial tests and have neutralized some of the AMD advantages, but promoting 64 bit would only result in making Mobile Athlon 64 more desirable compared to Pentium-M, which is something that Intel does not want, considering how well Pentium-M is doing and how profitable it is.
Not that there is a great value in 64 bit (or for that matter dual core) notebook, outside of maybe software developers, but when at some point 64 takes on a life of its own, it would be a serious deficiency, therefore Intel is tryig to slow things down.
As you said, Microsoft does its part, and so does actually AMD with slow transition from K7 to K8 and with introduction of 32 bit Sempron.
Joe
Tenchu,
funny as this might sound, Duke might be more accurate than even he realizes. When PCs transitioned from the 286 to the 386, most software was still 16-bit. It wasn't until Windows 95 was released that 32-bit software finally outsold 16-bit software. And by the time that happened, the choice really was between the Pentium and the PPro.
I am not exactly sure what it has to do with 16->32 bit tranisition. Both chips were 32 bit (as was every chip starting with 386). Pentium Pro was not really a mass market chip the way Pentium was, and its less than optimal way of running 16 bit apps is not much of an issue, since Pentium Pro was targetted to more purely 32 bit environments rather than mixed ones.
Also, there were other popular applications of 32 bit software and virtual 8086 in 386 that were quite popular prior to Windows 95, such as Deskview, Novell 3.X, Windows 3.x (which had virtual x86 as well as a way to run 32 bit programs with special API. Then there was Windows NT 3.1, released in 1994.
Joe
tecate,
You know that just doesn't cut it, by the time 64bits is REALLY needed, most of these people who bought AMD machines will need a new one - for speed, or graphics, or something. If Intel had pushed this on the public there would be daily if not hourly reports on the Inqurer about how Intel is a bully.
I don't know what you mean, and the little I can discern is off base.
Let's look at the 16bit -> 32 bit transition. Did Intel disable 32 bit from 386, 486 chips etc. pending Microsoft or someone else writing OSs and applications for them?
No. Intel released these chips, and over time (long time) software developers developped the entire 32 bit infrastructure. During this long transition, people who wanted to use 32 bit tools used them, others used them later, as the 32 bit support (or features of 386 processor) trickled in to more popular tools such as pre-Win95 versions of Windows.
During this time, processors without 32 bit capability fell off the face of the Earth, The world moved to 386, then to 486 etc. Seeking out a 286 when 486 was available because you don't "REALLY" need it 32 bit would, at the very least, be puzzling.
What I find interesting is the lengths the people on this board will go to defend even the most idiotic statements of Intel executives.
Of course Intel executives say these things not because they are true, but because what they say suits Intel better. Which makes me wonder if you are just such an enthusiastic supporter of Intel party line or if you are so clueless that you buy it.
Intel still has a problem with 64 bitness. They now have a chip that supports it, but it is a turd. The gem of a chip that Intel has does not support AMD64 yet, and most likely will not support it for between 1 and 2 years.
So in the meantime, Intel strategy is to polish the turd and to confuse everyone about 64 bitness.
Anyway, there is a big difference in how Intel approached 16->32 bit transition that invalidates everything Intel and Intel supporters say about 32->64 bit transition.
One might at this point wonder: Why is Intel acting differently this time around, during this transition?
The reason is that during the past transitions, Intel was in charge of x86 instruction set. The blunder was to seek to completely monopolize computing with Itanium instruction set, and to sacrifice the leadership of x86 to achieve the monopoly.
Looking at the bleeding of market cap year after year of the duration of this Itanium venture, I would say that it was a mistake. Maybe by now, Otellini also thinks it was a mistake. But he is about to take helm while the Queen Mary 2 is still moving in the wrong direction. In order not to cause panic, he is not going to say to the passengers they are going wrong way. He is going to (or already is) genly turn the ship to the right direction. He will hedge, lie, mislead until Intel has either 64 bit in Yonah or Merom / Conroe. Only then he will say "full steam ahead".
That's where it's at. When you or your son "REALLY" need 64 bit is irrelevant. If you REALLY needed it yesterday (and it was true), Intel would just lie and say that you did not need it.
Joe
Nurl (Newb of URL),
Wrong chips. The choice was between the pentium and the pentium pro.
Maybe you should do some googling...
Joe
Elmer,
Remember the Tylenol scare of years back? The company pulled all product. Now AMD processors aren't poison but they can be deadly to whatever system they go into.
AMD should do the right thing and replace all product that's at risk.
Thanks for some humor on a dreary day like today.
Joe
chipguy,
Wow, the transition to the 64 bit desktop is turning out
much harder and will take longer than any droid would
have accepted a year ago.
Sure, and a smart during this transition is to buy a computer incapable of 64 bit operation, just as it was smart to stick with 286 during the 16 -> 32 bit transition, while 386, 468, Pentium became available.
Joe
CompUSA to Sell Gateway PCs
August 16, 2004
http://www.eweek.com/article2/0,1759,1636266,00.asp
Joe
Keith,
I was at CompUSA within last 7 days, and the bad news is that Gateway PCs are being sold there as well (in addition to Best Buy), but I did see some AMD based models. These may be older models. We will see how it plays out...
Joe
Chipdesigner,
Well, whatever was supposed to happen on December 28 obviously did not happen, or was not subject of a press release.
BTW, the news of the dual core Opteron samples and info on dual core target frequencies did come out in that timeframe.
Joe
Keith,
I wonder if these are Newcastle or Winchester die with 1/2 cache disabled or if this is new 90nm 256K die.
Joe
Durl,
What happened to the URL from that article?
Joe
Golfbum,
have you tried to calibrate it for color profiles?
No, I have not tried to calibrate it yet.
Joe
"Intel only", blade pioneer RLX exits the hardware market, after introducing Xeon ET-64 blade:
http://www.eweek.com/article2/0,1759,1745252,00.asp
Joe
Golfbum,
Here is the link:
http://h18004.www1.hp.com/products/monitors/tft/l2335.html
All the info is there. It use the same LCD panel as the 23" Apple display, but is much cheaper and more capable as far as various inputs are concerned (it makes a good HDTV monitor as well, but I have not yet ordered a second cable box to try it).
The challenge for me right now is to make kids games (that for some reason are designed for hardware of a decade ago) running normally at 640x480 to run them at 1280x960 (1 pixel = 4 pixels on display) centered. The graphics card can hapily scale the image to 1600x1200, which is 2.5x scaling, but it looks a little grainy and uneven. 2x scaling would look better.
Some games are so dumb that I have to write scripts to change resolution on he fly (attached to an icon that the kid double clicks to start the game)
Joe