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.
Re: Does anyone have a single quote from any source that IBM has good yields?
Power doesn't seem to have any trouble out-yielding Itanium. Didn't Intel just have to cut the projected speed of the next Itanium chip by about a third so they'd have parts to ship? Clearly, if the process parameter were set to design specifications, yields dropped to zero or close to it.
Re: Just look at AMD's price sheet, and you'll see that given a competitive product
No. Wrong.
Because AMD does and would face competition.
AMD doesn't cheat and pervert the market with extortion to limit buyer's choices to the overpriced product or nothing.
If AMD raised prices unreasonbly they'd lose too many sales to their competition. That's the way free markets work.
Intel has used threats and bribes to contaminate the market and limit buyers' choice.
Re: Why don't you prove otherwise by showing where Athlon XP-M had strengths that rivaled Centrino
It runs Office nicely and fits easily into the power envelope of a ~4lb wireless notebook.
Which sums up what Pentium M / Centrino does.
AMD had these XP-M systems available (from Twinhead, despite Intel's extortion) IIRC, around the time or even before "centrino" was out.
http://www.tigerdirect.com/applications/searchtools/item-details.asp?EdpNo=1334812&Sku=A501-AV32...
Re: Your complaint is that Intel sold
My complaint (and this issue is only a small part of the AMD's suit AMD) is that Intel used extortion to block ME from buying the AMD systems I needed.
Me, personally, and I'm personally pissed about it. I wanted to be able to buy the best performning (which also happened to less expensive) systems I could get for the projects I support and the sleazy crooks at Intel used extortion to limit my choices.
I'm particularly and specifically angry that I had a very hard time finding Athlon XP PC's. I'm specifically angry about the difficulty I had finding dual processor boards for the Dual Athlon. I'm specifically angry about the universal exclusion of Athlon notebooks from any "government" or "corporate" lines, which meant I couldn't order them or had to spend literally hours having our procurement process an exception to their standard policies. I'm still furious about the way Intel extorted Solectron into refusing to ship the Opteron boards that were going to be in servers I had spent my precious time configuring.
I'm sick and tired of the criminals and thieves at Intel making my work harder so that they can (attempt) to force me to buy their overpriced, underperforming, overheating, crap. The stunts that Intel pulls are the kind of thing you hear about from cheesy third world countries with collapsed economies where what you buy or pay is determined by the calibre of gun pointed at your head rather than the quality and value of products presented for sale in a free and open market.
That's my complaint.
Re: clueless
Still convinced AMD will never ship dual capable processors?
Re: What brief? what reply brief? a complaint gets an "Answer". IT DOES NOT GET "BRIEFED" NOR IS THEIR A "REPLY BRIEF" to a Complaint.
It goes like this - Complaint, motion to dismiss or an answer, discovery, evidentiary motions with briefs.
It's possible, but unlikely, that it goes right to trial. The detailed evidence for someone unfamiliar with the way the industry has been run by Intel for the past 5 years is not in the Complaint or Answer. They're essentially two lists of allegations without evidenciary support.
It's possible there will be discovery not preceeded by motions to dismiss nor motions following discovery - it will just go straight to trial.
But this case certainly will, at minimum, have plenty of discovery disputes leading to motions to compel and the accopanying briefs could give you an idea of the legal theories they're focused on.
After discovery you will get evidence based motions, like a request for summary judgement (entry of judgement without trial). Incidently, motion practice involves a moving brief (request for relief plus grounds and support), an opposition brief from the other side, and the movant's reply brief commenting on the opposition brief. There could be additional briefing (sur-reply and response thereto, but that is somewhat unusual except maybe in a case this size). Until you get to those motions to dismiss (on some points) or for summary judgement, which goes to the merits, little of either side's evidence or theory of the facts and law are revealed. It's those motions and briefs that may let an outside attorney understand the progress of the case and gain a sense of the relative strengths of each side's case.
Re: In trial, if a witness or a defendant is caught in one lie
If Intel's thugs start torching buildings, that would turn this into a criminal case, but they haven't done that so what does either my statement or your "lie in trial" comment have to do with my Attorney wife saying the complaint looks strong but she can't give an opinion about the case until she's seen the brief and reply brief.
This is a big case and it will be briefed back and forth for through several iterations before it gets anywhere near to a trial.
Given that Intel's already had to pay several hundred million dollars in penalties assessed against them for abusing their monopoly (after lying about doing it), your point that Intel will walk into court with several strikes against them is a good one.
Re: Intel has been governed by adults for a long time and its business practices are almost surely clean.
Well, I can tell you from personal converations I've had with OEMs that Intel's business practices are not legal (regardless of how "clean" they may or may not be). Particularly from the standpoint of civil vs. criminal law.
If two parties (like AMD and Solectron) have engaged in a contract, it's illegal for a third party (like Intel) to interfere with that contract by threatening to withhold critical supplies or services.
Intel knows, that, too. They had to pay hundreds of millions in damages to Intergraph after they used the withholding of product plans as extortion against Intergraph (with a much less obvious impact than withholding cash or critical parts).
Intel Pattern of Conduct Harms Competition
Intergraph brought its private lawsuit against Intel in November 1997, having been forced to seek protection from a year of coercive actions by Intel. At that time, Intergraph viewed its case as an isolated example of Intel abusing its monopoly power. However, some eight months later the FTC presented broad evidence of a pattern of conduct by Intel - a pattern that had threatening implications for competition in general. Intel's intimidating and predatory behavior discourages companies from exercising their patent rights ... which reduces overall innovation ... which lessens competition ... which lessens choice and selection ... which hurts consumers and the U.S. economy.
Intel's actions have certainly hurt Intergraph. Given the clear evidence of such, including evidence subpoenaed from the FTC’s investigation, Intergraph expects to win verdicts and collect damages in all three areas of its broad lawsuit: patent infringement, illegal coercive behavior, and antitrust violations.
http://www.intergraph.com/ip/documents/journal/11204.pdf
They were let off pretty easily there, but you can expect the Judge's attitude towards letting Intel off with wrist slap this time will likely be "fool me once, shame on you, fool me twice, shame on me."
Re: Your point is ??
WBMW wrote: How about you cite some precedence for these levels of legal damages.
Like I said, $10 Billion is peanuts.
Re: Does your wife the attorney
My wife wants to see the full brief after discovery, and the reply brief.
She does think it's a very, very, strong complaint.
We have a lot of XP-M machines from Twinhead, Fujitsu, and compaq and they've been great.
Re: For right now, the story says that Centrino was
That's your story, and you're sticking to it. But you're just blabbering on a message board.
The only story anyone's been willing to swear to in a court of law says something very, very, different.
AMD's been collecting info for this lawsuit for 5 years, and Intel is in a world of hurt over it.
And the lawsuit is building at a time when Intel is a year away from having anything even approaching a competitive product in any segement but notebooks.
Intel's only hope is that AMD takes forever to ramp FAB36 and that they can't get anything out of FAB7, either.
Once AMD changed it's strategy to allow production at FAB36 to start on 90nm, if necessary, I think Intel's chances of taking a serious hit over the next 12 months became a near certainty and taht 12 months may stretch to 18, the to 24, then to - who knows?
$10 Billion is peanuts:
Florida smokers who won a $145 billion class-action lawsuit last summer in Miami-Dade Circuit Court.
http://www.no-smoking.org/may01/05-09-01-2.html
Re: Wells made that announcement
How about a link for that?
Re: just wait for the BatteryMark tests
BatteryMark? How are Intel's engineers doing on modifying the code for batterymark? Are any them in the cube next to yours?
Someone had better remind them that all the publicity surrounding the lawsuit means that this time they've had constructive notice that what they've been doing is illegal. Their latest ("if Genuine AMD then disable half the instruction set" changes will be taken to court.
Maybe we'll see some Intel software engineers in orange jumpsuits and not just the exectives.
Re: But you're wrong. Yonah is 31W and that has been disclosed numerous times
There's no such thing as a Yonah (unless you want to claim AMD has a quad core 65nm chip).
Yonah is a future chip on an Intel roadmap. They've produced some prototypes, and are as happy with them as they were with the "soon to be released power saving 5ghz Prescott" at this stage.
What's worse, it's a 32-bit chip that won't even first ship until next year - when 64-bit windows will be the hot news of the day.
Re: You are obviously confused. Banias was a 130nm part that was the first Pentium M
You're right, I should have said Dothan, not Banias.
Thanks for the correction.
Re: Pentium M does the same thing - right down to 5.5W for the ULV parts, or 10W for parts that go up to 1.6GHz. AMD just barely managed to catch up to Pentium using voltage scaling tricks
AMD had the 32-bit Pentium M matched 3 years ago when they introduced PowerNow in the 32-bit XP-M.
Intel has caught up to AMD's 32-bit XP-M, but now faces the 64-bit Turion.
And all of these families of CPUs use roughly the same amount of power, particularly in normal use, which is editing documents (do you really think most mobile business use is ripping CD's while unplugged on the airplane? - that's the only case in which Pentium M doesn't lose.)
http://www.laptoplogic.com/resources/articles/42/13/1/
Re: Everyone else here already knows the answer to this
Like the 5ghz Prescott was going to be next year's answer last year?
Like 105w Xeon was going to be next year's answer last year?
Like Rambus systems werew going to be next year's answer the year before that?
You have a great track record for "already knowing" next year.
Re: Athlon XP-M was gaining share until Intel launched Centrino
LOL !! You mean, the Centrino extortion program.
You can read about what really happened in the lawsuit:
http://www.amd.com/us-en/assets/content_type/DownloadableAssets/AMD-Intel_Full_Complaint.pdf
And this isn't just two guys arguing on a posting site, that lawsuit has been sworn to and filed in court - it's for real.
Re: AMD won't get a dime, just like all the other lawsuits
And I think the final settlement will be around $10 Billion.
We'll see who's right.
I've been posting for a long time, I expect the initial damages to be $10 Billion (revene difference seen by AMD due to ASP difference) which will be automatically tripled (by law, in cases like this) resulting in a jury award of $30 Billion.
Intel will appeal and I estimate that award will be cut to around $10 Billion. There have been other estimates as low as $3 Billion, but (other than a few Intel fanatics like yourself) that's about the least anyone expects to see Intel paying AMD.
Re: What is the MTBF for Opteron?
Hmmm. Good question.
We started getting Opteron boxes about two years ago. We now have more than 50 Opteron servers in our server room that use (mostly duals, a few quads, a few singles) about 100 parts.
So far zero failures. Infinite MTBF in the real world (thank goodness for RAID 5 - some drives have gone, but haven't brought down any servers, so far, anyway).
Re: And that obviously is never going to happen since they are now running at 60% reliability (while the system is up) since the other 40% if off-line.
Looks like it was a big F-Up to go with Dell and Noconas.
A very good point - thanks, Joe!
Dell / Xeon reliability is running at a consistent 40% failure rate in a least one major datacenter installation due to Intel's horrible power problems. (the ones they're going to magic away with fairy dust in a year or two)
Re: Otherwise, both cores may be operating simultaneously and still fit within the 31W power budget. That's 15.5W per core, max
AMD is shipping 64-bit server parts with dual high speed on die memory controllers now that are speced for 55w under server loads.
At this point in banias's life it was being touted as a 21w part - then it came out as a 27w part.
6 months from now, configured as a notebook chip, AMD could probably get that down to 35w or 40w.
Meanwhile Intel (if past history is any guide) will end up shipping Yonah as a 32-bit, 35w to 40w, limited floating point part that requires another 5w or so on top of that for the memory controller. And if it can't turn off one of its cores to save power when there's no need to keep the other core running - I'd say it's an even worse design than I'd thought. Turning off the sections of the chip not needed at the time is supposed to be Intel's strength.
By next year, when Yonah first rears it obsolete little head, 32-bit is going to be moving more and more to the forefront of what differentiates entry level from the mainstream.
Re: giving end-users what they REALLY want
That's the big question, isn't it, what will be the driver of sales next year and the year after?
Intel has low power, long battery life, and that's it (for now).
AMD has low power, longer battery life, and 64-bits now.
Intel will be adding a somewhat higher power, dual core, shorter battery life part, still 32-bits, early next year.
Intel hopes to have a new core by the end of next year that will let have 64-bits at some power level.
AMD will be doing the same - but AMD's has a lot less further to go to get there.
Re: and Intel rightly took their market share away with the introduction of Banias
Athlon XP-M was gaining share until Intel started using extortion to deny AMD access to what would otherwise have been a free market.
We'll find out for sure about the "rightly" part of your statement at the verdict, and just how much their final loss was after the damages phase.
AMD may end having done OK, after all.
Re: AMD is doing very well, but they could have done even better had evil Intel not broken the law.
Actually, they're only now starting to do OK (financially) but it looks like they're going to be doing a lot better going forward.
Then, once their business is really doing well and the new FABs are letting them gain market share at a rapid rate, the lawsuit will give them $10 Billion on top of their other profits, ripped out of Intel's hide as retribution...
Now that's what I'll call doing "very well."
Re: Yonah is reported to dissipate 15W in LV mode with dual core
Yes, "reported to" is all you can say about a chip that consists of a few low speed engineering samples so far.
Just remember how certain you all were (at this stage of the game) that Prescott would be hitting 5ghz within a couple months of its release (don't feel bad, AMD was convinced of it too, and planned to make only 1mb cache Athlon 64s in order to would stay competitive against P4 Prescott that shipped only as 4.2ghz, 4.6ghz, and 5ghz).
Yonah in low power mode shuts down one of the cores - keep in mind that you're now talking a single core part. Nothing wrong with that, it's a great power saving strategy. But the Turion core uses 15w (or quite a bit less) under some PowerNow settings, too.
And Turion is an actual, shipping, production chip - it isn't a (possibly) forlorn hope of a chip.
Re: They'll have an aging and obsolete Fab 30, a partnership with Chartered that may or may not work out, and one more 300mm fab that will be their only means of competing with Intel
Hmmm.... that "obsolete" FAB is putting out low power SOI workstation and server chips that beat anything Intel will have for 1 to 2 years. The partnership with Chartered, as you say, may or may not work out. You're focused on the "may not", and I'm focused on the "may" - we just have to see what happens. But FAB 36 is as sure a thing as anything is in this business, and AMD will be, as you guys keep reminding us, under pressure to pay its loans.
FABs 30 and 36 will be cranking out desktop, mobile, workstation, blade, and conventional server parts at full blast. AMD may or may not be minting money, but they'll be taking share from Intel. It'd be ironic if AMD wound up doing to Intel in the CPU market what Intel did to AMD in the flash market, wouldn't it? Now that the law is limiting Intel's illegal monopoly abuse, it can't use extortion to block AMD from what would otherwise have been open markets. This time Intel's products will have to compete not just in retail desktop but in the business, notebook, workstation, server and blade server segments. What if "desperate to pay its loans" AMD crashes Intel's share by down to 70% and it ASPs by 50%...
Another thing to consider is Intel's "attachment rate." When AMD loses a CPU sale, that's pretty much it (except in servers, where they sell some chipset parts.
When Intel loses a server socket, they also lose a chipset, the motherboard that chipset would have been on, and a couple of network phy chips. With notebooks they usually lose a chipset sale, as well as a CPU sale, every time an AMD notebook heads out the door at Office Depot or from HP's online web site.
Re: Which they'll keep for another couple quarters, before the erosion starts.
Based on what? You have a long position and you hope that's the way it works out?
Re: It took them 2.5 years to catch up to Centrino, and once Yonah ships, they will once again be far behind the power curve.
AMD has had a 15w to 25w 32-bit part for years (the XP-M). Turion is 12 to 18 months ahead of anything Intel has, because it's a fast floating point 64-bit part - as Merom will be, when (if - we're talking Intel roadmap here) it ships in 12 to 18 monhts.
Re: Intel's Centrino brand
Intel's Centrino brand is a decent but uncompelling part that's been boosted by a combination of a huge marketing campaign and substantial and organized program of extortion. OEMs were faced with substantil penalties if they didn't drop AMD CPUs from their notebooks.
Have you noticed how rapidly AMD's market share has been coming back since it files its lawsuit? Check out this week's ad from Circuit City, or Office Depot.
The old Athlon XP-M had performance, power use, and capabilities very similar to Banias, but Intel's abuse of its monopoly extorted the XP-M out of the market. Now that Intel (and its OEM victims) are under the bright light a lawsuit's discovery process, the OEMs are offered some protection from Intel's abuse and AMD's mobile share is growing fast.
It doesn't hurt that AMD's 64-bit, gaming quality mobile parts are demonstrably and dramatically superior to Intel's (as has often been pointed out, for many notebook uses, a low power 32-bit chip like AMD's old XP-M can make a fine notebook chip), but the big difference is allowing free markets.
Re: Intel's "equivalent" core is Dothan
No, Intel's equivalent core is Merom, and it's a year or more away.
Yonah is a pair of Pentium Ms - a 32-bit core with mediocre floating point, but good integer and a power miser.
Like Via's c3.
http://www.via.com.tw/en/products/processors/specs.jsp
Yonah isn't anywhere being near an equivalent to Turion or X2. Intel is hoping to come up with a chip like that in a year or so. Wish them luck, they'll need it.
Look at the graphs carefully, some tests are in seconds (shorter is better) and some are scores or frame rates (longer is better).
Keep in mind that AMD is shipping 55w dual core Opteron server parts TODAY, and will be shipping faster parts with lower power long before Yonah, much less Merom.
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
Re: SOI was a one-time gain... Intel will be wiping the floor with AMD by the time 65nm ramps
LOL!!
Intel will be wiping the floor with AMD once they ship Itanium
- oops! Scratch that.
Intel will be wiping the floor with AMD once they ship Pentium 4
- oops! Scratch that.
Intel will be wiping the floor with AMD once they ship Rambus systems
- oops! Scratch that.
Intel will be wiping the floor with AMD once they ship the 5ghz 90nm P4, and it's only two quarters away!
- oops! Scratch that.
Intel will be wiping the floor with AMD once they ship strained silicon
- oops! Scratch that.
Intel will be wiping the floor with AMD once they ship 65nm Yonah
- You just never learn, do you?
Meanwhile AMD holds a big chunk of the server market, and is growing that highly profitable segment at a high rate.
AMD is now shipping a 64-bit notebook chip that uses less power than Intel's 32-bit chip (25w vs. 27w).
AMD's workstation chips outperform Intel's while using barely half the power.
And, most important by far, AMD is tripling the number of FABs they have next year.
And all you can do is whine liar, liar.
Pathetic.
RE: Link please.
It was in the post -
You're slipping.
Re: It's a problem that won't go away, and AMD can't hide from it
I don't know about hiding from it - they seem to have solved it by mastering the SOI process, instead. AMD's 64-bit server cores with competitive floating point presently use 27.5W each (Opteron 265HE). Meanwhile Intel's equivalent core, on the same process node, but without SOI uses 2 to 3 times as much power.
http://www.theinquirer.net/?article=25756
Apple seems to be having second thoughts about moving to Yonah/Merom. Between no 64-bit until Merom, and crappy floating point in Yonah (which can be important for many Apple users) it may be that Intel parts just can't do the job for Jobs.
http://www.eet.com/news/latest/showArticle.jhtml?articleID=170100975
Re: amid budget cuts this upgrade occurred, great of UB, a great institution btw.
Too bad they were stupid enought to select Intel Xeon from Dell.
UB's Dell Supercomputer Running at Only 60%
Wednesday, Aug 24 @ 09:36 PDT
...its newest and most powerful supercomputer, a 1,668-processor Dell supercomputing cluster that will be used to support university research ranging from genomics, to groundwater modeling to the monitoring of human-rights abuses. According to a report in the Buffalo News, a shortage of electricity has left the new supercomputer running at only 60 percent of its capacity.
The $2.3 million computer was announced in July. The addition of the federally funded computer was said to nearly double UB's number-crunching power. However, they haven't been able to fully start the machine without causing power outages at the computer center at UB's North Campus in Amherst. The shortage is the result of an initial underestimation of the power needed to run the machine.
The chum heads at UB must have believed Intel's TDP "Typical Design Power" numbers - that was rather naive of Russ Miller at UB, wasn't it? I mean, if he hadn't been so utterly igorant of the technology he's responsibe for, he'd have picked Opteron boxes, which use about 40% less power at the wall, and have been OK.
The rest is at: http://www.supercomputingonline.com/article.php?sid=9197
Re: You're really funny. Like Opertons wouldn't have blown the breakers, either....
Opteron CPUs use about 1/2 the power of Xeons (under full load). Opteron CPUs also incorporate the most power hungry (RAM controller) aspects of the chipset, as well. So there are additional reductions in power. Less power use means less use of fan motors for cooling, less use of the HVAC and lower parasitic losses from the UPS.
So knocking total power use down by 40% would have been close, but they probably would have made it.
Dammit! Who's the moron that bought Dell Xeon - He's Fired!
Somebody find me Sun's number so we can get some Opterons in here!
UB's Dell Supercomputer Running at Only 60%
Wednesday, Aug 24 @ 09:36 PDT
...its newest and most powerful supercomputer, a 1,668-processor Dell supercomputing cluster that will be used to support university research ranging from genomics, to groundwater modeling to the monitoring of human-rights abuses. According to a report in the Buffalo News, a shortage of electricity has left the new supercomputer running at only 60 percent of its capacity.
The $2.3 million computer was announced in July. The addition of the federally funded computer was said to nearly double UB's number-crunching power. However, they haven't been able to fully start the machine without causing power outages at the computer center at UB's North Campus in Amherst. The shortage is the result of an initial underestimation of the power needed to run the machine.
The rest is at: http://www.supercomputingonline.com/article.php?sid=9197
Looks like total power IS quite a bit higher than "Typical" power, doesn't it?
Re: As one top analyst whispered to us, "Looks like AMD is heading back to where it started - a little cheaper and a little faster."
LOL !!!
Intel hasn't even the hope of anyhing competitive for at least a YEAR !
And where will AMD have moved on to by then?
Intel has never, ever, been anywhere near this far behind AMD before.
And AMD has never been able to advantage of being ahead in the server and workstation business before.
Did you see that using AMD CPUs has let Sun take back the number 3 server positon from Dell?
http://biz.yahoo.com/prnews/050824/sfw083.html?.v=54
Sun's Opteron server revenue is up 378% over last year.
Dell's Xeon revenue was what? Up 6%? DOWN?
Re: Opteron has a superior per watt performance, making the 5.0 Liter Mustang less perfect analogy
The Mustang analogy is fine, you just have to switch motors....
My 99 Mustang SVT convertible came with a quad valve per cylinder, quad cam, 4.6 liter motor.
4.9 seconds 0 to 60.
On road trips (pure highway driving), much to my amazement, it gets 28 to 31 mpg (at 70 to 80 mph).
I can't go below 50 in 5th gear, because it's at 1400 rpm at that speed.
We needn't go into city driving mileage here, that's where the Mustang/Opteron analogy breaks down....