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.
AMD made 45 nanometre SRAMs with a .37 square micron cell size, and product design on 45nm has already begun.
That's all about Charlie has to say about the VERY exiting conf call: http://www.theinquirer.net/?article=30757
Regards,
Rink
IBM announces Opteron and Xeon servers: http://www.theinquirer.net/?article=30756
Regards,
Rink
Re: Conroe 2.4GHz Super Pi 1M 21.45
I just looked at the references Mas provided.
Is this indicative for how good Conroe's overall integer performance is at 2.4GHz?
Regards,
Rink
Fab 36 shipments started end March; 65nm test chips are finally being produced; Fab 30 reaching 30K wafers / month.
http://www.investorshub.com/boards/read_msg.asp?message_id=10505348
Regards,
Rink
Kate, Your content free posts are approaching 90%
Regards,
Rink
Sandia-designed Supercomputer Ranked World's Most Efficient In Two Of Six Categories: http://www.sciencedaily.com/releases/2006/03/060331152034.htm
In various ways significantly outperforms Purple and Blue Gene because of it's balance between bandwidth and processing power.
Regards,
Rink
Chipguy, thanks (eom)
Kate, re: I'm many things but girly or gossipy aint them.
You're sure?
Regards,
Rink
Mmoy, Are you indeed a different poster than Michael Moy?
Just to clear up my mind (I think I might have confused you with Michael)...
Regards,
Rink
Energy lab to run petascale computer
By Joab Jackson, GCN Staff
NEWPORT, R.I.—The Energy Department’s Oak Ridge National Laboratory will start to take delivery on a computer at least as three times more powerful than any now in operation.
The system should be operational by 2008, according to Thomas Zacharia, ORNL associate lab director for computing and computational sciences. Zacharia spoke at the High Performance Computing and Communications Conference this week in Newport, R.I.
Cray Inc. of Seattle will supply the system, named Baker. It will run approximately 24,000 2.6 Ghz quad-core Opteron processors made by Advanced Micro Devices Inc. The nodes will be housed in 187 liquid-cooled cabinets. The system, which is still in early stages of design, will have either 187 or 400 terabytes of working memory (depending on the cost of memory modules) and from one to 11 petabytes of storage.
Such a computer, if operational today, would be considered the world's fastest; it may even be the first to break the 1 petaflop limit. A PFLOP is 1 quadrllion floating-point operations per second.
According to the Top500.org list, today's most powerful computer is the Blue Gene/L, run by Lawrence Livermore National Laboratory, another Energy Department outfit. That system currently runs at about 350 teraflops (350 trillion floating-point operations per second), though the lab and IBM Corp. plan to expand its capability to 1 PFLOP, also by 2008.
The laboratory will rely on the Tennessee Valley Authority for the power needed to run this behemoth. The power company is building a 170 megawatt substation to support this and other ORNL projects.
Like current ORNL systems, this one will be used by some of the country's best scientists to further explore the limits of their fields. Researchers must demonstrate the need for using the computer, and the lab will look for those jobs that are too large for other systems.
"We're entering new realm of scientific discovery, where we are looking at things at a much smaller scale. Computing expands upon our understanding of the natural world," Zacharia said. "Greater computer capacity allows researchers tackle more complex problems."
http://www.gcn.com/online/vol1_no1/40250-1.html
This seems to confirm once more that k8l will have a double performance FPU capable of four double precision (64b) floating point operations per clock per core:
- One single core 2.6GHz Opteron now is capable of 5.2 GFlops.
- 5.2GFlops * 24000 processors * 4 cores/processor = 0.5 PFLOPS.
- Above Cray system apparently has a theoretical max performance of 1PFlops
- Hence it should have double performance FPU / core.
One thing is unclear to me: Does NGA support four single (32b) or four double precision flops per core? I think the answer was posted here before but I forgot.
Thanks. Regards,
Rink
MMoy, Pentium M does 2 64b Int calcs per cycle, with a 3 to 4 cycle latency. Merom will 2 128b Int or 4 64b calcs per cycle, with a couple of cycles latency as well.
re: What this means is that your compiler needs to
schedule the operations to achieve the theoretical
throughput. In practice, this is hard to do because
you need to do loads and stores mixed in with your
arithmetic operations. I'm leaving out other operations
that typically need to be tossed in.
Exactly right. When only looking at int heavy programs and ignoring all other changes for now practice will show nicely evolving performance increase of e.g. between 0-25%, instead of the 100% theoretical increase. Some somewhat exotic cases increase will be more, some synthetic benchmarks too.
Other improvements (like bandwidth, latency to main mem, cache, 4 issue, higher frequency at the cost of a pipeline increase, etc..) come on top of this.
Regards,
Rink
Greg, re: Like I said, absolutely delusional.
Can't blame you from saying that. I can see why you think it might not make sense. It's not that I don't think other reasons have better chances among which is the fact MS is prone to delay. Still I can't rule it out either.
Regards,
Rink
Snowrider, re: I don't think the problem is that you consider it a possibility in a list of posibilities. The problem is that it is your default position and you don't really want to hear about the other posibilities because they don't align with your agenda.
Nope. I consider it high on the list of low probabilities.
Regards,
Rink
Greg, pretty clear that we disagree. I stand by my suggestion that MS could possibly have agreed Intel on a introduction term for 64b Windows Home Edition, however unlikely you may find that.
Regards,
Rink
Tenchu, That's nice for Charlie Green.
Just to be clear: I haven't claimed whether I found it likely or not. Just that I find it a possibility that I am not willing to rule out.
Regards,
Rink
Greg, Duke, Kate, Just fyi, the three situations I put to Kate have been subject of WSJ online at one time or another. Either as a serious possibility (Bush having completely reasons to liberate Iraq, and Bush's first election being possibly crucial for MS's antritrust case at that time), or as a reality (DRAM price fixes).
Regards,
Rink
Kate,
Why do you think MS got off the antitrust hook after Bush Jr was elected the first time?
You still believe Bush when he says there WMDs were the reason to liberate Iraq?
Have you never heard of any debacle in which DRAM makers agreed to fix prices?
Whatever area of society the list of insane political agreements is just unbelievably long which means it can't be ruled out that MS made some kind of deal with Intel in relation to the release of 64b Home Edition version of Windows in some kind of quid pro quo.
Just to be clear (again): I do not know if this is the case. But if it happened I think around the time that MS told Intel that they'd only do one 64b version of Windows for them, either for Itanium or for Yamhill, is the most likely moment it could have happened.
Regards,
Rink
Kate, well, when you say so..
Regards,
Rink
Chipguy, as I said you may well be right. I just can't rule out the other explanation either, not even on the basis of your logical arguments.
Regards,
Rink
Chipguy,
We'll see. You may be right for as far as Vista is concerned. You may be right too in thinking we'll find any Merom systems on the shelf before Vista is released. Then again at this moment we don't know that for sure. This is in line with my statement that I "wouldn't be surprised" to see Merom systems for sale approximately at the time Vista is available for the reason of inter-company politics. Do you know what MS did or did not agree to specifically at the time they told Intel they'd only do one 64b version for Intel either for Itanium or for Yamhill? I'm not saying I think any special agreement was made at that time or any other time; I am saying it can't be ruled out.
Maybe you pride youself in being a good at politics, but the tone in some of your posts makes me doubt that seriously. In politics the most insane agreements are not that uncommon at all.
Regards,
Rink
Kate, that article from Ed Stroligo. If you put a lamp at one side of his head you'll see the light cone coming out at the other.
That said I'm not presuming any 65nm volume meaningful enough to influence the bottom line either this year. I'm expecting that to happen in Q1 '07.
Regards,
Rink
I see you understand politics.
Regards,
Rink
Wouter, wbmw, Turion X2 is a different design than Opteron/A64. It uses a much higher amount of low power transistors that trade a bit of performance for a lot of power savings. The result is that - compared to the A64/Opteron design - that the top frequency is down a bit and VCC is down significantly. This again means optimum performance/watt.
Regards,
Rink
Smooth, I've been thinking the same for years now. So far delays in a consumer (Home Edition) version of 64b Windows have lived up to my assumption that MS won't release it until Intel is one hundred percent ready. Quite possibly this time might be different but still I wouldn't exactly be surprised by a Merom delay to a post xmas date either.
Regards,
Rink
Snowrider, re: I'm assuming you're saying this [If the frequencies are available at launch it'll impact Intel's mobile ASP's until Merom in a significant way.] because you believe that 64 bit support is a significant factor in the purchasing decision. If my assumption is correct, then we'll have to agree to disagree because 64 bits has done little for the single core Turion over the single core Pentium M processors and Centrino Platform.
Windows Home edition still doesn't support 64b. Vista will support it across all releases (incl. Home). So the major difference now is that Vista is imminent and will make 64b the sole standard (32b Vista will work but with less features than 64b). Consumers - stupid as they are - are starting to know this. And a bit after May they'll be offered the choice (dual core 32b Centrino now, 64b DC Turion X2 now). Turion X2 will be popular because it has the features the consumers and businesses want. Consequently Centrino (Yonah version) will become slightly less popular. Only 'slightly' because after all it is an Intel product. Also only 'slightly' because AMD is locked out of Dell, Sony, and the like. Still I think it'll be enough to affect Intel mobile ASPs in low single digits. I know: Just *one* opinion based on logic.
re: Correct me if I'm wrong, doesn't Core Duo have higher IPC than Turion? If true, then a 2.16GHz Centrino Duo platform should have no problem outperforming a Turion X2. Personnally, I'm curious to see the battery life of an Turion X2 based system.
Well currently you're right ofcourse. But as of May it's different because current Core Duo (Yonah) IPC is in rather general terms on par with current Athlon X2, and will be rather close to Turion X2 as well. They'll be more or less on par. Battery life differences between the two will be much more determined by components like screen, video, harddisk, etc... than by the cpu's, meaning Turion X2 systems might still have slightly lower battery life. A bit further down the road Merom (Core Duo as of Q4 this year) will outperform both its predecessor and Turion X2.
Regards,
Rink
Kate, sorry but I'm not going to react with real arguments to your well thought out thesis supported by so many indisputable facts.
Regards,
Rink
Kate, that Richard Suttmeier does know TA but is imo likely to know next to nothing about FA of semi's. I may well be wrong, but earlier today I bought another chunk of AMD Jan '07 $20s. I think AMD might go up a few the weeks preceeding earnings (or possibly the day after). I'm looking to sell again at that time. Also earlier today I said I think Intel hasn't reached bottom and that I think that is more likely to happen in Q2 (AMD thread on SI). Just saw that Rick Whittington agrees: http://www.investorshub.com/boards/read_msg.asp?message_id=10250388
Regards,
Rink
PRU downgrades INTC. Sell. EPS of 86c. Target: $15
NTC: THREE GROSS MARGIN HEADWINDS - LOWERING ESTIMATES AGAIN
Current: Underweight
Target: $15.00
HIGHLIGHTS
• We think that Intel will face gross margin headwinds from three areas for the foreseeable future: 1) share loss to AMD, 2) startup costs associated with 45nm and Micron JV NAND memory ramp, and 3) potential inventory build. Our analyses indicate that:
• 1) further share loss to AMD could conservatively translate to 350 bps of gross margin pressure in 2006.
• 2) startup costs could create another 275 bps of gross margin pressure between Dec-05 and Dec-06.
• 3) a potential inventory build of $500-$900 million in the Mar-06q could result in 800 bps of gross margin pressure in 2006, similar to what the company saw in 2004.
• We are lowering our GAAP EPS estimates by $0.06 to $0.86 for 2006, and by $0.05 to $0.99 for 2007. We are materially below consensus, and think the Street forecasts are too high. We recommend investors sell INTC with a $15 price target.
________
Bad Q1 followed by a bad Q2 is why I'm waiting at least till somewhere in Q2 to get into Intel.
Regards,
Rink
Michael, In Holland a big American multinational I worked for got India telephone support. They standardized on Dell for Windows and HP and SUN for UNIX systems. Maybe it US telephone support is reserved for the US?
Regards,
Rink
Morowinder, Turion is specifically designed for mobile in that it uses much more low power transistors than A64. Apart from this you're right in that it is a retrofitted desktop chip. Maybe retrofitted server chip is more precise.
Regards,
Rink
Morrowinder, re: I highly doubt AMD will have the volume [for Turion X2] to impact Intel's [mobile] ASP's.
I don't. Fab 30 doesn't need to be converted and can focus on simply producing at max level. Fab 36 has been added, and estimates by knowledgeable but not directly related people are that it started at around 1500W/W, which will ramp higher during the year. This allows AMD to keep max units close to Q4 production while aggressively ramping % of DCs. Fab 7 (Chartered) is due to be added to the mix in Q4. This allows AMD to ramp both volume and % of DC. After this fab 36 ramps 65nm to high volume. Let's say for discussion purposes that I think Turion X2 will be 15-20% of Q3 AMD mobile unit sales, and that it - even when it pales to Yonah volume - it will suffice to affect Intel's mobile ASPs by say a low single digit percentage.
re: Remember the original Turion took forever to go high volume.
For the original Turion OEM's had to design new notebooks. Turion X2 is fafaik drop in replacement. I think it's safe to assume it'll ramp a heck of a lot faster than original Turion.
re: And in fact Turion is competing on price in US retail since is an unknown brand and relatively worse than Centrino at the same frequency(perfomance and battery life) but Centrino's asps have held up fine so far. So you think an x2 will reverse this? I find it hard to believe that the x2s will be better on either performance or battery life than Yonah...
Yes Turion is competing on price because performance is worse. That's current situation. I'm pretty sure that Turion X2 ML will slightly outperform Yonah in most benchmarks provided AMD can launch them at stated frequencies. Let's say they'll be on par for this discussion. Much more importantly however is that Turion X2 is 64b which is a feature customers want in order not to miss out on full Vista performance. Even when not that well informed customers buy features and Turion X2 will have them all, while Centrino simply won't until Merom. Turion X2 will provide somewhat unprecedented competition to Centrino because of: 1) on par performance and 2) better features. This will allow AMD to take a bit more market share in Q3/4 when compared to Q2. Not too much because AMD hasn't been able to attracked Dell, Sony, and the like.
re: Never heard of that german site? Are they reliable or are they like digitimes...
PC Welt is a major German IT newsmagazine. It's ok. Definitely better than Digitimes. I think it's unlikely that they published this without *some* form of confirmation. Still they're obviously not exactly 100% reliable... I'd say they're ok.
Regards,
Rink
An indication of Turion X2 - May 9.
1.8 GHz Turion X2 @ 25W
2.2 GHz Turion X2 @ 35W
http://www.siliconinvestor.com/readmsg.aspx?msgid=22272108
If the frequencies are available at launch it'll impact Intel's mobile ASP's until Merom in a significant way.
Regards,
Rink
Kate, whether or not Fleck lost money on his Intel puts depends on when he bought them. I think he made money on them. I also think I read he bought leap-puts (is that a good word?). http://finance.yahoo.com/q/bc?s=INTC&t=6m&l=on&z=m&q=l&c=
If you forget about that 15% company discount for just a second: Did you still never loose money on Intel shares?
Anyways, as you know I worked for an American IT company as well. They had the same package, but because of dutch law we had to pay 52% tax on the 15% company discount. We could sell them anytime (even the day we received them) without paying any additional taxes. All profit from equity trading is tax free. We do have to pay tax over net assets (1.3% over all assets minus all debts) and in Holland it's unusual to have debt beside mortgage, some pay off a loan for a car or washing machine, but virtually noone really uses credit cards for anything but business (the habit is to use debit cards instead).
Regards,
Rink
Wbmw, just a couple of details. This small group of enthousiasts has a large audience. Well I can see your point too. I'm not expecting very big change either; it'll still be better than average seasonal decline of course. I think mobile rev might have been mid single digits (yeah wag but just to give you an idea of what I'm thinking) better than it'll be now.
FYI: Osborne is written without a 'u': http://en.wikipedia.org/wiki/Osborne_effect
re: Given the number of Centrino Duo machines that already sport a "Ready for Windows Vista" logo, I find this comment highly speculative bullsh!t. Consider the source (Digitimes).
K, good point.
Regards,
Rink
Joey, I'm waiting till at least after Q1 earnings, and possibly till bad Q2 results are fully accounted for in the stock price (as far as I can determine that ofcourse). Yonah simply can't have made up for market share loss to Opteron in this timeframe. My take is it might be at the low end of Intel reduced guidance. Q3 is the first quarter that starts benefitting from Merom derivatives. I think Q2 is early enough to pick up Intel; it might even be too early, but what the hey. Similarly although I reduced AMD in my portfolio to just 25% I'll might sell it in the same timeframe for the same reasons as I'd pick up Intel. Depending on more information I might change my tactics. For now I'm feeling comfortable with this. Just my 2c.
Regards,
Rink
Kate, I read Fleck has owned Intel puts for quite a while now. Pretty sure you're dead wrong about him loosing money on Intel as those must have been at least reasonably good to him.
Now if your husband continued to own company shares like I think you once said...
Regards,
Rink
Wbmw, Well Osborne didn't believe in Osborning either. And as for your argument that the news about Merom and siblings might not be widespread enough outside the US to enable significant Osborning I disagree as through the internet and every local IT magazine on earth they're touted as the coolest news the biggest chip company on earth has had for years. On top of that Vista is big news too and it has apparently began to be a reason to buy a cheap 64b alternative in notebooks instead of an expensive 32b dual core (just to be sure you can run tomorrows software) which implies that people know 32b is outdated because of what is due to come soon.
Regards,
Rink
Gordon, that article is from Jan; it doesn't state Feckensteins current opinion.
Regards,
Rink
Kate, 'fanfils'? Is the 'fils' part of that from the french word 'fille(s)' / girl(s) - i.e. fanfille? Or is it perhaps slang (not my native language)?
Anyways, in case it's indeed derived from french: 'fille' is pronounced something like 'fEy'.
Regards,
Rink
Duke, Yep I did. I seldom game but like beautiful game scenery when I do. I only buy notebooks for personal usage, and they're priced below minimum price for quad SLI alone. So, alas, quad SLI is out of my world.
From Intel's point of view it's obviously in the game of winning from AMD and has put out just that notion with the Conroe vs FX60 demo setup. And for that purpose it has seldomly uses nVidia products for it's demo's as ATI has shown it more loyalty previously in various ways. Just my opinion ofcourse.
Regards,
Rinik