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.
sgolds -
you can rely on a rumor article already discredited.
Considering that he provided no data whatsoever, what makes Keith's post anything more than a rumor too? At least I made the disclaimer "if true". You just bought the rumor hook line and sinker and passed it along as proof.
Chipguy -
To Dan3: Can you point to any reports of data corruption involving the 3.0 GHz P4? Once again you demonstrate your fleeting acquaintance with the truth.
Just figuring that out? I wouldn't be as polite as you. I ignore him because he's a liar.
All we know is that Intel halted shipments. There are a variety of reasons that this may have been done. But at this late stage of the game it is IMO most likely because the ATE guard band on some minor data sheet spec had been set too aggressively and some samples showed up that failed this spec for one of the operating condition corner cases.
Maybe but I doubt this is the actual reason. It is more likely a new speedpath recently discovered that needs to be added to the vector set. Speed path analysis is not an exact science. Speed limiters are discovered empirically and no doubt the new FSB speed introduced a new one not seen before. I'm guessing that the delay is to allow a rescreen of existing material.
YB -
Good plan on the calls but you are really confused on everything else in your post.
AMD did not screwup last year with ramping fast hammers. They didn't have any hammers at all, fast slow or slower. The MTH problem was years ago and as for the 800MHz FSB issue, you guys need to change your shorts. Everybody's talking themselves into a frenzy when this is almost certainly a minor delay. They've had months to validate this configuration and this is almost certainly a very low percentage issue that just needs a new screen. That's my guess. If they say it'll be released shortly then they're probably just rescreening material with added test coverage.
Chill out guys.
Keith -
K.J. Chou, general manager of AMD Taiwan, said that samples of the Opteron processors have been sent to customers, and volume production will start shortly after the official launch.
Here we go again! Volume production won't even start until after launch. If this is true then add 3 months until product starts trickling into the marketplace. And if they haven't started volume production how do they know they have a production worthy product?
Vaporon!!!!
AMD executing themselves flawlessly.
TB -
Most likely the bug is something I won't understand reading the errata report.
I think you're getting a little carried away here. So far all we know is that this is an anomaly. It may turn out to be nothing or it may require a fix or work around unrelated to the processor.
Springdale/Canterwood reviewed.
http://www.xbitlabs.com/articles/cpu/display/p4-800.html
Paul -
doesn't PCI-Express require a licensing fee?
I don't think so. PCI doesn't.
As for it's use, it is an extremely highspeed serial interface and it will get much much faster. One only needs to use a little imagination to see how it could be used.
YB -
I think you are getting a little carried away. Don't bet your life that Prescott will be 64-bit and don't be so sure that Intel has nothing like aHT. PCI-Express offers higher performance and Intel is it's champion. Use your imagination.
EP
Greg -
Not to nitpick but he was famous for "stay tooned"
Spokeshave -
Gee. I hadn't thought of that.
I know. That's why I try to help when I can.
You still did not say why you thought that AMD, near its 6 month high, was a buy at these levels.
Either the worst is behind AMD or they are dead. I posted here a day or 2 ago that I had rolled my April $5 CCs to July and made $0.30. Now I'm sure that looks like a tiny gain but it's a 26% annualized return with downside protection. I also wrote some May $7.5 puts and picked up $0.70. If AMD goes up fine, my puts expire. If AMD stays the same or goes down, I'll roll my puts out and my CCs can be rolled again to make more money. The ideal case is that AMD stays between $5-$7.5
If AMD goes under then my INTC will do just fine.
Spokeshave -
If I didn't know any better, I would guess that *you* have gotten caught up in the "hype".
Maybe I have but I'm making money. AMD is also a hedge for my larger positions. If AMD goes to zero it won't kill me and my other holdings would have essentially no competition. Consider the big picture Spokeshave.
EP
wbmw -
I have no doubt that AMD will temper any bad news with eager exuberance, like they normally do, and quarter after quarter, investors buy the hype. I am simply betting that history repeats itself.
Well put. Additionally, when the current investors become disgusted with AMD and move on, there is always a crop of new investors who stand in line to hear the hype. Just like you, I'm betting that history continues to repeat itself. At these levels this is a great stock to play.
EP
borusa -
If you are an Intel employee, I would appreciate you saying so.
Why do you ask him? I don't recall anyone ever asking anyone on any board if they are an AMD employee.
EP
Intel's XScale Finds a Place With Palm
Tungsten C will be the first Palm device to use Intel's handheld processor.
Tom Krazit, IDG News Service
Wednesday, April 09, 2003
Palm's new Tungsten C handheld will be released around the end of April as the first PDA bearing the Palm brand to use one of Intel's XScale processors, a source familiar with the company's plans said Tuesday.
The Tungsten C showed up in the ordering system at office supply company Staples in February for $500, but Staples employees could not confirm any details at the time.
The Tungsten C will feature 802.11b wireless Internet technology, in contrast to its sister Tungsten W handheld, which uses a GSM/GPRS (Global System for Mobile Communications/General Packet Radio Service) module for wireless data communications.
Chip Competition
The Tungsten C marks a design victory for Intel over Texas Instruments and Motorola. Motorola's Dragonball chip powers the Tungsten W, and the Tungsten T uses TI's OMAP chip.
Intel's XScale processors for handhelds have mostly appeared in personal digital assistants that use Microsoft's Pocket PC 2002 operating system; but a few Palm OS devices, such as Sony's Clié, exist as well.
A Palm representative could not be reached for comment. Intel declined to comment.
Semi -
I like this line -
"Sources close to AMD have reported that first real silicon is expected for its 64-bit microprocessor, codenamed Sledgehammer, as early as next month. Engineering samples of the chip are already with large PC customers. "
So first silicon "real" is expected the next month but samples are already in the hands of customers! I guess that aren't real silicon.
wbmw -
Whatever frequencies they can reach now is just a starting point. If they are holding a hand of aces, you better bet that they'll play it.
I agree. There's no conceivable reason to hold back anything here. They are in for major embarrassment when a 1.8GHz Opteron fails to beat Northwood in AMD's choice of benchmarks to highlite Opteron, SPEC. They promised the highest performance in essentially everything so if they have it, they'll ship it and if they have 2 they'll ship em both<G>
NotaShort
suppose all those times INTC went to comdex and/or other events with preproduction silicon being shown to reporters (often times with special cooling) they should have immediately stopped production of all other Pentiums just to sell the few high speed demo machines they could make at that time?
I realize you're being intentionally silly. Production material has to meet strict quality requirements and specifications. Any company will sell the fastest material they can produce that meets the requirements. If they aren't selling it then it's because either they can't produce it, they haven't qualified it or they haven't got the bugs worked out yet.
INTEL: Intel unveils ultra-thin stacked flash memory technology for cell phones
Intel Developer Forum, Tokyo, Apr 09, 2003 (M2 PRESSWIRE via COMTEX) --
Intel Corporation unveiled flash memory products for cell phones with up to five ultra-thin memory chips stacked for greater memory capacity, lower power consumption and optimal space savings. The features of the Intel Ultra-Thin Stacked Chip-Scale Packaging (CSP), featuring 1.8 Volt Intel StrataFlash Wireless Memory, are designed to ease manufacturers' provision of such features as camera capabilities, games and email in relatively thin cell phones.
Intel also announced that it has shipped its 2 billionth flash memory unit, a milestone underscoring the increasing role that flash memory plays in the cellular and wireless market segments. The announcements were made in a keynote address at the Intel Developer Forum in Tokyo by Darin Billerbeck, vice president of Intel's Flash Products Group.
"Stacking is quickly going mainstream in the cellular wireless market segment," Billerbeck said. "By combining our highly dense Intel StrataFlash Wireless Memory in an ultra-thin stacked-CSP, our wireless customers receive the flash density needed for their feature-rich phones, while at the same time saving space for small design footprints."
It took Intel 12 years to ship its first billion discrete flash units and only three more years to ship the next billion discrete flash units.
Additionally, Intel to date has shipped more than 150 million Intel Stacked Chip-Scale Packaging (Intel Stacked-CSP) devices.
Early on, Intel Stacked-CSP products were designed and manufactured on a custom basis for manufacturers to offer high-end functionality on their cell phones. Today's announcement marks a new manufacturing approach to bring more stacked products to the market faster using a standard, high-volume manufacturing process. The new standard stacked flash products will enable wireless device makers to easily upgrade their designs with higher flash densities to meet additional memory and performance requirements for their wireless devices.
World Class Silicon & Packaging Intel Stacked-CSP products use a new Intel Ultra-Thin Stacked-CSP package, an advanced wafer thinning and package technology, to obtain lower-profile package heights and more stacking flexibility. By combining 1.8 Volt Intel StrataFlash Wireless Memory with the new Intel Ultra-Thin Stacked-CSP package, Intel can offer extremely small and compelling stacked-product solutions.
These advanced stacked-CSP products will allow up to five stacked die with package heights as low as 1.0 mm. The products feature 16 bit and 32 bit buses as well as SRAM, PSRAM and LP-SDRAM options. The products will offer up to 512 Megabits (Mbits) of flash memory this year and 1 Gigabit (Gbit) of flash memory by next year, to facilitate both code execution and big data applications. They will also feature Intel's latest multi-level cell flash memory.
First introduced by Intel in 1997, multi-level cell flash memory (MLC) doubles the amount of data that can be stored in a single memory cell and provides higher density and lower cost. The 1.8 Volt Intel StrataFlash Wireless Memory is Intel's fourth generation of MLC flash technology built on leading-edge 0.13 micron process lithography. It is the world's first low-power MLC flash memory operating at 1.8 volts and significantly improves performance, battery life and space savings for wireless devices.
Pricing and Availability
Intel Ultra-Thin Stacked-CSP products featuring 1.8 Volt Intel StrataFlash Wireless Memory are currently sampling, with production volumes starting in third quarter of 2003. Pricing will vary by specific flash and RAM memory combinations.
wbmw -
Don't you think SUN suffers a net loss on each US sold? They make money (or I should say they used to make money) by selling an integrated system complete with their own software. Switching to Opteron eliminates the silicon loss. I think it's a pretty good idea. To completely replace US, Opteron will need to scale beyond 4/8 way, otherwise SUN still has to spend the r&d on silicon development or go Intel. No other way around it. They are truly stuck between a Iraq and a hard place.
EP
YB -
I continue to believe that AMD offers an excellent opportunity for profit in the short to mid term, and possibly longer term if they can actually produce Opteron and see continued success from Flash. In my mind, the key is to not try to make a killing here but to be content with slow steady profits from a range bound stock.
Wrote some May $7.50 puts today and also rolled out my April $5.00 CCs to July. Got $0.30 for the call rollout. That's 6% for 3 months or 26.25% annualized.
Chipguy -
Yet you still advanced the theory that the on-chip peripherals could be affecting Hammer's clock rate on several occasions IIRC.
You don't recall clearly. I think exactly the opposite. The peripherals shouldn't affect it at all. That's the point.
Your acknowledgement that this idea is nonsense seems like a clear admission you were deliberately and knowingly spreading FUD about Hammer. No surprise considering where
your loyalties obvious lay but it is nice to get it on record for posterity.
A silly statement. Please try to direct your comments to the correct person. It makes you look foolish.
sgolds -
I'll choose door (c): New architectures are always introduced at conservative clock speeds. You don't have a basis for any conclusions on the capabilities of the manufacturing or any other part of the design for a full year after introduction.
I think we've already seen that year... This is what happened after a year of trying to get it to work.
Chipguy -
Why would you ever expect speed degradation? What critical processor timing path would run through a peripheral?
I wouldn't. That's the point.
Chipguy -
LOL, do you deny the fundamental physics that SOI eliminates the PN junction capacitance associated with the source and drain active areas of a MOSFET? Look at the capacitance breakdown on a typical net and do the math.
Please... Don't point me to a textbook, customers don't buy textbooks. Show me a product on both bulk silicon and SOI where SOI shows an advantage. The only one I know of is Hammer with their Athlon core and it shows a disadvantage.
Have you considered something as simple as this is the first
implementation of the K8 microarchitecture while AMD has re-implemented the K7 microarchitecture around half a dozen times? Maybe they've learned something along the way. Engineers tend to do that.
Hammer is a K7 with 2 more pipeline stages, a memory controller and aHT ports. The additional pipeline stages should have added to the frequency. Couple that with SOI and now we have a 15% reduction is speed, if expectations hold. I gave you an example of Intel adding a memory controller to their CPU core plus a graphics controller and seeing no speed degradation whatsoever compared to their mainstream cpu. All indications are that SOI does not deliver on it's promises, textbooks excepted.
Chipguy -
If you redesign all your circuits from the ground up, and you know what the hell you are doing, then you can realize 10 to 15% higher clock rates with SOI
Where is the evidence to support this? This statement is problematical for AMD. Considering that AMD's choice of SOI seems to have resulted in a 10-20% reduction in frequency, you seem to be pushing AMD into a corner here. Does this mean that a) AMD didn't redesign their circuits? or b)AMD didn't know what the hell they were doing?
How else do we explain a ~15% drop in frequency when we should have seen a ~15% increase? When I look around our industry I just don't see anything more than talk to support the claims that SOI can yield higher frequency.
For IBM, its Power4 chips are manufactured in such small numbers that even if SOI doubles the silicon cost of each uP that it is still far less expensive than an extra $100m
in design effort and speed path tuning respins (especially for a server chip that has an long and expensive verification and requal).
This kind of increase die cost due to poor yields may be absorbeable for Opteron but it will kill Athlon64 which already looks like it will need much more cache to be competitive.
SZ -
Conclusion - AMD's SOI is terrible for speed.
I posted about a week ago another example of a cpu with memory controller, Intel's Timna. It also has a graphics controller on die so it was at least equivalent in complexity to Hammer. Timna was able to run at the full speed that Intel was marketing their mainstream CPU, although Intel held the clockspeed down so as to keep it a lowend product. Unfortunately Timna was ultimately cancelled, but not for any problem of it's own. The point here is that if AMD had simply produced Hammer on bulk silicon they would have been in production long ago and at much higher frequencies than it looks like they will ship on SOI.
Conclusion: SOI was a horrible mistake and probably the only thing that could have kept hammer from being a great success. The time advantage is now gone and the frequencies are now greatly constrained.
neye -
I'll assume that you're saying this for the benefit of others, because I don't believe I've ever scoffed at the 10% - 15% typically seen with iHT speed-up.
Sorry neye - it wasn't targeted at you..
neye_eve
I realize that virtually nothing is linear, but a 10% gain is relatively large in the world of cpu's
It's very small when it's gained by Intel's hyperthreading. Huge if from Opteron's ondie memory controller.
Unisys Has High Hopes by Aiming Low
Apr 07, 2003 (Internet.com via COMTEX) --
Three years ago, Unisys broke free of its heritage as a stodgy mainframe company by redefining itself in the high-end computing market with the introduction of the ES7000 family of Intel-based servers running on Windows.
Capitalizing then on the greater stability of Windows 2000, Unisys was able to make some inroads in the high-end server market due in part to the so-called "WinTel" architecture. With the power of 8- to 32-way processing, Unisys signed on some customers that wanted to move away from proprietary RISC/Unix-based systems.
This time around, Unisys is taking aim at the low-end market. On Monday, the Blue Bell, Pa.-based company will introduce the first expansion to its ES7000 line of Windows-based servers. Timed to capitalize on the marketing push behind the April 24th launch of Windows Server 2003, Unisys is making available the ES7000/500 models featuring the same Unisys Cellular MultiProcessing (CMP) server technology powering mainframes. And while the servers have the capable of TPC-C benchmarks of 118,381.38 transactions per minute (tpmC) at $5.56/tpmC, making it faster than the 8-way server offerings from HP and IBM, the 500 line starts at a base of just 4-processors at a price of $35,000.
"For some people, it didn't work out that well. It's not economically attractive," said Mark Feverston, vice president, Unisys Enterprise Server Marketing, referring to the previous ES7000 line.
The ES7000/500 series is Unisys's answer to IBM's x440 eSeries line , which was introduced a year ago. IBM's so-called "building block" servers also features the "pay-as-you-grow" modular concept with prices starting at $18,000 for a smaller two-way server.
To be sure, Unisys officials contended the low-end initiative isn't being launched because of a lack of demand for the more complex machines. During a telephone interview, Feverston explained that 75 percent of all ES7000 systems that have shipped go out the door with 16 or more processors. But analysts believe that statistic is more of a function of the large enterprise-class customer base that Unisys has traditionally courted.
"The reason that people were only buying big systems is if they weren't buying big systems, frankly they had better options," said Gordon Haff, senior analyst at Illuminata.
And to that extend, Unisys may have resolved one of the biggest issues with the traditional ES7000, which is the high upfront investment costs. "One of the attractive things with the x440 is it is very modular and you can buy as you go. And particularly with something like a high-end Windows system ... where Windows is still thought of as more of a distributed OS ... there is a comfort level for customers to take it slow and buy it gradually," Haff told internetnews.com.
Unisys officials believe they can succeed by targeting their sales efforts in the right directions. Feverston said the three areas where they are concentrating on is in the consolidation market, database server market and in the growing area of business intelligence -- which has long been a stronghold of Unix-based systems.
And in this arena, Feverston believes that, with its WinTel arsenal, Unisys will have success luring customers migrating away from Unix. Even that other Unix alternative (namely Linux) can't compare with a high-end Windows environment for Unisys's customer base, he said.
"The kind of customers we attract are people that need sophisticated, mission-critical enterprise applications," the Unisys VP explained. "We believe Linux does not have the enterprise capabilities. It does not have the ecosystem around it."
That ecosystem Feverston is referring to are the independent software vendors, systems integrators, integration skill sets and utility vendors that support the WinTel platform.
"When you see all those kinds of organizations come together around Windows, that's the biggest ecosystem on the planet. If I'm moving away from Unix, I don't want to compromise. What they see is there is no compromise going to the ES7000. We really think we are going to have an impact on Unix."
That said, Unisys may still find it difficult to tap the low end market, the analyst said.
"It's going to be tough for Unisys to make a big impact in that broader market...it's still relatively high-end in the scheme of things," Haff explained.
http://linuxpr.com/releases/5740.html
IBM Continues Momentum in High-End Intel-based Computing - Record Setting Performance
Apr 7th, 16:54 UTC
Thousands of Customers Now Using eServer x440 System
ARMONK, N.Y., APRIL 4, 2003 . . . IBM today announced breakthroughs in Intel-based computing with a new record setting performance result on the eServer x440. Thousands of customers are now using this system since it was announced in March 2002.
The IBM eServerTM xSeriesTM 440, using IBM TotalStorageTM FAStT700 storage, recently set TPC-C performance and price/performance records(1) for a 16-way server running the TPC-C online transaction processing benchmark. The x440 achieved 151,744.13 tpmC (transactions per minute/C) at a cost of $11.03/tpmC.(2)
This scalability and performance is a result of IBM's Enterprise X-Architecture, which allows the IBM eServer xSeries 440 to scale up to 16 Intel® XeonTM MP processors. In the future, the x440 will be enabled to support up to 32 Intel Xeon MP Processors. Dell and HP currently only have servers of this type that contain eight Intel Xeon MP processors.
IBM's technology behind industry standard computing has helped produced significant performance results for customers. According to IDC, in fourth quarter of 1999 Compaq owned over 83 percent of worldwide revenue for servers with 8 Intel processors, yet according to the IDC Quarterly Server Tracker, by the fourth quarter of 2002 IBM accounted for 51 percent of worldwide revenue of the same types of servers.(3)
"What we see today as Intel-based servers enter the high end is that technology matters a lot," said Deepak Advani, vice president, IBM eServer xSeries. "This technology will eventually allow you to connect and manage up to 32 processors. Our investments in Enterprise X-Architecture have previously helped enable IBM to achieve time-to-market advantages over other major server vendors with Xeon MP based systems in high end 8-way systems and blade servers."
This fact is borne out by the thousands of eServer x440 systems that are currently in use by customers across a wide range of industries.
NetCreations, a leader in email direct marketing services chose an IBM eServer x440 running Linux to help consolidate their current technology infrastructure. IBM business partner Saturn Business Systems deployed the eServer x440 system at NetCreations.
"The IBM eServer x440 helps manage our more than 80,000 e-mail lists and 48 million double opt-in e-mails as well as push marketing services out to our customers," said Steven Gittleson at NetCreations. "We needed a solution that could give us high performance and scalability. We chose IBM because it has the option to scale to 16 processors. We are looking forward to expanding our 8-way x440 platform to 16 processors allowing us to upgrade processing power rather than replacing processing power."
"IBM's eServer x440 system provides customers with the ability to scale their infrastructure and maintain a high level of availability," said Tom Heger, Sr. Account Manager, Saturn Business Systems, Inc. "Our customers have found IBM hardware to be extremely reliable and our internal testing helps demonstrate the ROI from this unique hardware."
DXP Enterprises, a leading distributor of products and services for industrial maintenance, repair, operations and capital equipment projects, chose the IBM eServer x440 running Windows to host its supply chain management software solution. As a medium-sized business, DXP requires an 8-way system to handle today's computing needs, but has plans to expand to a 16-way system in the future as the business grows.
"The IBM eServer x440 gives us the opportunity to scale up and pay-as-we grow our business, said Suzie Dahle, IT Director, DXP. "Our previous server had processing limitations, so we replaced it with the IBM eServer x440. We needed a powerful server that would allow us to upgrade easily and the eServer x440's scalable architecture and management features allow us that flexibility."
Come on Intel... Turn on the tap...
http://news.zdnet.co.uk/story/0,,t271-s2133080,00.html
Pentium M chips may face supply problems
15:58 Monday 7th April 2003
Matthew Broersma
Some manufacturers have already sold out of their Centrino-based laptops in China due to tight supplies of Intel's new mobile processor, according to a report
Some resellers in China are having trouble getting supplies of laptop computers based on Intel's new Centrino group of chips, according to a report. Demand has been high for the new technology as a result of good reviews and an intensive marketing campaign by Intel, in the UK and elsewhere.
Samsung, Acer and Asustek Computer have all had trouble supplying Centrino laptops to their Chinese resellers, due to the tight supply of the Pentium M processor, according to a report in the Taiwanese industry journal Digitimes. The Pentium M is one component of Centrino, which also includes a chipset and integrated Wi-Fi wireless LAN.
The group of chips is designed to deliver longer battery life for wireless notebooks, which Intel sees as a major selling point for future portables. It is the first time Intel has delivered chips designed especially for laptops rather than customising a desktop chip instead.
Samsung was only able to supply 500 units to China, and Acer 200 units, according to the report. Samsung's Chinese reseller has already sold all the units. Intel is denying any suggestion that the Pentium M is not being manufactured in full production quantities.
<more>
YB -
Why should I contrast commodity desktop Northwood with high-end server Opteron? Can I use Itanium size instead?
Because you're the one who brought up P4 and the need for "wast" manufacturing resources:
I think anyone will agree that P4 (and Itanium too) requres wast manufacturing resources, because it's design needs a lot of cache.
You can compare to Itanium if you want but Opteron will never see the ASPs that Itanium gets.
YB I think anyone will agree that P4 (and Itanium too) requres wast manufacturing resources, because it's design needs a lot of cache.
"Vast", not "wast", and with NorthWood at less than 135mm2 I think your should contrast that with the projected 185mm2 die size for Opteron. So who needs the "vast" manufacturing resources? With the Athlon64, the 256K L2 version, essentially canned because of poor performance, it would seem that Opteron is the one dependent on a very large cache to be competitive, and even still it will fall short of current, much smaller Northwood in SPEC performance if expectations hold. AMD's choice of SOI will prove to be the biggest blunder in the companies history.
Borusa -
"team" Humn, two or more can be a team. You guys were on one side of the issue, Dew on the other. Well, what word would you rather?
Please... "Team" implies a coordinated effort. Do you really think these guys are doing that?
Greg -
Good way to handle it. I'm pleased the early days of disagreeing on thread moderation seem distant now.
EP
YB -
Right. Elmer! Before you start fighting - I was talking about bulk 130 nm process, not soi.
No problem. I knew that.
Joe -
I didn't run this through the translator. Does it say whether these are compiled with a 32 bit or 64 bit compiler?
I didn't run it through either, and as someone else pointed out, it's not new but it is the first time I've seen 1.8GHz numbers. Let's give AMD the benefit of the doubt and assume it's 32 bit but at the same time let's realize that SPEC doesn't thread processes so it doesn't show the benefit of iHT either. It does show that if those numbers hold, and 1.8GHz is the highest frequency shipped in April, that the best available Opteron will fall short of Intel's current shipping Pentium in the benchmark that AMD originally chose to use to highlight Opteron performance. This is in stark contrast to the claims of "world's highest performance" made many times by proud AMD enthusiasts and exactly in line with what I've been predicting ever since AMD started making their claims.
I realize a benchmark on an obscure German website doesn't settle anything, and AMD may present something different but after years of outrageous claims we are getting close to the time when AMD has to put up or shutup. Talk has been damn cheap and it won't sell for much longer. We are coming up on "show time".
Yourbankruptcy -
Say goodnight Gracy...
Spokeshave -
Am I missing something?
Yes. So looked right at it and missed it:
"Assuming these hold, and assuming Opteron's top bin will be 1.8GHz(Best case)"
It beats the P4 2.8 by 24% and 25% in INT and FP respectively, it beats the 2.8GHz Xeon by 31% and 33%, it beats Itanic 800 by 280% and 80%, it beats Itanic 2 by 50% in INT. The only thing that is loses in is FP to the Itanic 2.
Would you like me to post Intel vapor scores that beat the Opteron vapor scores?