VentureBeat

Advanced Micro Devices is taking a swing at Intel today with the launch of its Phenom II microprocessor for the consumer desktop market, as well as a new Dragon platform that will house the processors in a variety of new high-end computers.

The Phenom II is based on AMD’s new 45-nanometer manufacturing process, which will let AMD stay competitive with Intel’s newest chips. In the case of consumer desktops, analysts now believe that AMD will leapfrog Intel for the first time in a few years. The chips have larger cache memories — up to eight megabytes — and the processors start at 3.0 gigahertz. Previously, the fastest Phenom I chips could only hit 2.6 gigahertz. With the improvements, AMD says its new chips will be 20 percent faster than previous chips.

The company is unveiling the product at the International Consumer Electronics Show in Las Vegas next week, where AMD chief executive Dirk Meyer will be giving one of the keynote speeches.

The new chips will make AMD credible again at the high-end of the game PC market. The chips capitalize on a new production technology that comes along every couple of years for chip makers. The 45-nanometer technology lets AMD make smaller, faster, and cheaper chips because it shrinks the width betwen circuits, allowing for lower power and less material use. Intel launched its own 45-nanometer processors some time ago and refreshed its line-up in November. AMD launched its first 45-nanometer server chips in November.

AMD has shown that the new processor can be overclocked to 4 gigahertz with air cooling or 5 gigahertz in certain circumstances with liquid-nitrogen cooling. That shows the chip has plenty of headroom for future performance improvements. In gaming mode, the chips can run reliably at 3.5-gigahertz and 3.7 gigahertz.

The Dragon platform is a collection of key chips that PC makers can assemble into a computer. It includes the AMD Phenom II X4 processor, the ATI Radeon HD 4800 series graphics, and the AMD 7 series chip sets. With the graphics chips launched last year, the new machines based on Phenom II and the Dragon platform should be far faster than AMD’s previous line-up, with games running 40 percent to 180 percent faster.

David Schwarzbach, AMD’s division marketing manager for desktops, said he hopes that do-it-yourself PC builders will come back to the AMD platform with the launch of the Phenom II. Many of those PC builders, who are often hardcore gamers, defected to Intel’s Core chips starting in 2006, when the world’s biggest chip maker pulled ahead of AMD in performance.

The picture here shows a gamer pouring liquid-nitrogen into a cooling system for a PC at a preview event AMD held in December. Schwarzbach said the company will likely bump up the performance of the chips as it proceeds with new versions throughout 2009. (That’s not always the case with designs that don’t have much design headroom, or room for improvement). That means it’s time for Intel to be at the top of its game.

The chips are in production and computer makers are announcing a variety of models today. They will be listed on AMD’s web site.

Advanced Micro Devices is introducing technology today that can make notebook computers lighter, faster, and more versatile.

The Sunnyvale, Calif.-based chip maker’s Yukon platform is a collection of chips aimed at the ultraportable notebook computer market. Such laptops will likely blur the line between full-fledged laptop computers and lesser machines known as netbooks.

The first major customer for the Yukon platform is Hewlett-Packard, which will use Yukon in its new high-end notebook, the HP Pavilion dv2, which has a 12.1-inch screen and weighs 3.8 pounds. Bahr Mahony, director of notebook product marketing at AMD, said the heart of the Yukon platform is a new single-core microprocessor, the Athlon Neo processor. It also has an ATI Radeon X1250 integrated graphics chip (or an optional ATI Mobility Radeon HD 3410 standalone graphics chip).

The ultraportable computers are aimed at business users who need full Internet capability. While netbooks (stripped down computers aimed at cruising bare-bones Internet sites) often don’t perform well in loading sites with rich media, the ultraportables are intended to be full-fledged web cruising machines, like Apple’s Macbook Air model. With high definition multimedia interface (HDMI) and DVI connectors, Yukon-based laptops can display their images on big screens. AMD is thus shooting above the netbook market and is targeting ultraportables with price tags ranging from $500 to $1,500.

“We felt that many of the netbooks make too many compromises,” Mahony said.

In particular, the Yukon platform will use AMD’s latest Mobility Radeon graphics chips so that the machines can play games and run full high-definition video. Of course, good luck to AMD figuring out what to call this category: ultrathin, ultraportable, ultra whatever.

One of the secrets of the fast-growing category of “netbook” computers — the mini laptops meant for email or web surfing only — is that they’re being returned like hot cakes. These slimmed-down computers look sexy, but they can’t really do that much. They hang on a lot of web pages with rich media and can’t play most games.

That’s why graphics chip maker Nvidia is introducing its Ion platform, a chip set with integrated graphics that can be paired with Intel’s processor for netbooks, Atom. (Get it? An ion is a charged atom). The Ion platform uses an Atom chip with Nvidia’s GeForce 9400M motherboard GPU (a graphics processing unit that sits on a computer’s main board, in contrast to GPUs that have their own board on a PC).

Netbooks debuted in 2008 with the launch of Intel’s Atom processor, which was specifically designed as a low-power chip for use in portable devices that are smaller than laptops. Some netbooks also used Via’s Nano processor. The machines sell for $400 to $1,000, sometimes cannibalizing sales of more capable laptop computers. Successful models include the HP Mini and its successor released in October, as well as models from Asus, Ace and Slyvania.

But Intel had to restrict the devices to use its 945Gx and its ICH7 chip set, which had weak integrated graphics performance. Early netbooks had analog displays, SD flash memory instead of hard disks, a standard operating system such as Windows XP or Linux, no DVD drives, and screens smaller than 10 inches.

David Ragones, a product line manager at Santa Clara, Calif.-based Nvidia said that most netbooks with the Ion platform could have 10 times better graphics and video performance. They can run full high-definition video and use a premium operating system such as Windows Vista or Microsoft’s upcoming Windows 7.

The devices can have a rich set of input-output functions. You could plug an Ion-based netbook into a flat-panel TV via an HDMI (high-definition multimedia interface). You can use peripherals via the universal serial bus 2.0 and plug into the wired Internet at gigabit Ethernet speed. You can also add an external hard disk.

In a demo, Ragones showed how the hit game Call of Duty 4 won’t even play on an Intel Atom-based netbook but runs just fine on a machine with the Nvidia Ion platform. If the netbook manufacturers want, they can now add screens with displays that are more than 10 inches diagonally.

The Ion platform may add about $50 in cost to a computer, but it can be used in netbooks that cost as little as $450. Machines with the Ion platform will debut in the first half of 2009. They will be paired with Intel’s Atom chips and will compete against the upcoming Yukon platform from Advanced Micro Devices, which is pairing its own processors with ATI graphics.

What does Intel think about this? Well, the processor team likes it because it’ll help sell a lot more netbooks, but the chip set folks… that’s another story.

For the past couple of years, graphics chip maker Nvidia’s big strategic thrust has been to make its chips capable of more than just graphics processing. It unveiled its own CUDA processing architecture and programming language to allow software programmers to use graphics processors to handle various parallel computing tasks beyond drawing pretty pictures on screens.

Today, Santa Clara, Calif.-based Nvidia is announcing that its CUDA architecture will be compatible with a new technical standard for graphics-based computing, dubbed OpenCL (short for open computing language). As its name implies, OpenCL was originally started by Apple as a neutral, open platform for computing tasks that are handled by a combination of microprocessors (such as those made by Intel and Advanced Micro Devices) and graphics chips (made by Nvidia and AMD). Now OpenCL is the product of an industry consortium led by the Khronos Group.

CUDA-based programs are proliferating and enable some software to run much faster by taking advantage of a microprocessor and a graphics chip at the same time. Elemental Technologies’ Badaboom, launched in October, dramatically reduces the time it takes to transfer a video from one format to another, as you transfer a video from a computer to an iPod.

Nvidia has made great strides with CUDA. There are more than 100 applications for it and 25,000 developers working with its programming tools. The company considers CUDA to be an advantage over AMD. But with OpenCL, programmers no longer have to worry about being locked into Nvidia solutions. AMD and Nvidia are both big supporters of OpenCL.

With a standard, programmers will be able to write their applications to run on OpenCL and they will run on any machines, whether they have Nvidia or AMD graphics chips. This doesn’t mean that CUDA is shut out. Rather, applications written in OpenCL will sit atop the CUDA architecture. As CUDA and OpenCL share similarities, translating code from one to another is simple and Nvidia will release a guide for developers wishing to do so, says Manju Hegde, head of the CUDA division at Nvidia.

OpenCL is in its initial alpha testing stage now. A test version is expected to debut to the public in the first quarter of next year and a final version in the second. Meanwhile, Hegde said that the roadmap for CUDA will continue to evolve with full support for OpenCL as well as Microsoft’s new graphics-microprocessor technology, dubbed DX11 Compute. Apple plans to include OpenCL in its upcoming OS X Snow Leopard release.

Advanced Micro Devices caught up with Intel today, announcing that it is slashing its fourth-quarter revenue forecast because of a sharp drop in demand.

The company said it expects revenue from continuing operations for the fourth quarter ending Dec. 27 to be 25 percent lower than its third-quarter revenue of $1.58 billion. That doesn’t include technology license revenue. The shortfall is due to weaker than expected demand across all geographies and businesses, particularly in the consumer market.

Curiously, AMD did not change its guidance at its annual analyst meeting on Nov. 13, a day after Intel slashed its fourth-quarter revenue forecast by a billion dollars. At that time, AMD expected a flat quarter, but it said back then it would offer new guidance today. AMD’s executives said they wanted to wait until they had more visibility into the quarter.

AMD will report final results on Jan. 22. The outlook matches the gloomy forecast from Intel, which foresaw a bad quarter much earlier. However, Hewlett-Packard, a major customer of both companies, surprisingly has not changed its forecast. HP doesn’t have the same fiscal quarter dates — its fourth will close at the end of January. But other tech giants like Cisco have joined in the steady drumbeat of forecast slashing.

The recession is hitting all chip makers. The Semiconductor Industry Association’s annual forecast predicted the first decline in annual sales for 2009 since the crash in 2001. AMD is in the process of cutting costs by dividing itself into two companies, one that makes chips and another that designs them.

Rick Bergman is the senior vice president and general manager of the graphics products group at chip company Advanced Micro Devices. His team of graphics chip designers –- formerly known as ATI Technologies until AMD acquired the company in 2006 — has scored well with the recent launch of the ATI Radeon HD 4800 series family. The so-called “sweet spot” strategy caught rival Nvidia flat-footed. While Nvidia created a high-end graphics chip that burned a lot of power, AMD created a smaller chip that targeted the middle of the market. It decided that it could put two such chips into a board to create a low-cost, top-performance solution for high-end gamers. Meanwhile, the single chips could be easily targeted toward the mid-range and low-end of the markets. Nvidia, by contrast, has had to wait longer to redesign its chips for the mid-range and low-end markets. The result has been a big shift in market share from Nvidia to AMD, whose graphics chip division is now profitable after a tough 2007. I spoke with him at AMD’s recent analyst meeting at its headquarters in Sunnyvale, Calif., after Bergman gave an upbeat outlook for AMD’s graphics business.

VB: How long ago did ATI started thinking about its “sweet spot” strategy?

RB: It was three or four years ago. We recognized we could not continue on with huge die sizes. It was before the merger with Advanced Micro Devices. We were still working on a chip called the R600. We were thinking of what we could do for the R700. We decided we couldn’t do another chip that was so big because of power consumption. You can just plot how die sizes have grown generation after generation.

VB: Did it take graphics chip designers longer than microprocessor designers to realize that power was going to be a problem?

RB: The microprocessor designers ran into it first because they were pushing frequency. When you do that, it pushes the circuits to the limit and consumes a lot of power. We started from a lower frequency base. Graphics processing units (GPUs) started requiring more and more performance to do better 3-D images. We weren’t able to lower voltage, and so it was clear we were going to run into the fundamental laws of physics.

VB: You looked at this problem from a different perspective when you were an underdog?

RB: About three years ago, we were around 50 percent plus market share position. We were not in a tailspin. We were looking at where the problems were. Where the growth was in the market. We had a chip called the R300. That was the first 256-bit GPU. It let us get the crown. It cascaded into multiple years of success for ATI. As we made this decision, we had chips like the R580 which were the best in the industry.

Read the rest of this entry »

As Nvidia’s graphics chips are enlisted to handle computational tasks in supercomputers, the chip maker is finding there is a new class of machines emerging: the personal supercomputer.

Scientists have been using workstations for years to make calculations or share a slice of a multimillion-dollar supercomputer. But Nvidia says that the era of personal supercomputing has arrived for scientists who need to conduct experiments on their own in a timely manner. By adding graphics processors to machines that have only had microprocessors in the past, scientists can speed up their calculations; 3-D ultrasounds can be done 20 times faster, while financial simulations can be done 149 times faster. In computational chemistry, one task that took 4.6 days can now be done in 27 minutes.

Vivoli said that the company was finding that scientists were cobbling together their own supercomputers in their offices to supplement their access to supercomputing resources (see picture below).  Nvidia’s newest Tesla graphics processing units (GPUs) are aimed both at this emerging market and at the supercomputing clusters that use hundreds of machines. Scientists may find the new personal supercomputers with the Nvidia Tesla S1070 chips appealing because they won’t have to wait for their share of a big supercomputer’s computing resource.

The new products are expected to find a welcome reception because graphics chips can be far more efficient at certain kinds of processing. To date, most supercomputers consist of collections of central processing units (CPUs) such as those made by Intel and Advanced Micro Devices. But with the launch of its CUDA programming language, Nvidia has made it possible to run general-purpose programs on GPUs. Since GPUs process in a highly parallel fashion, while CPUs work serially, different kinds of tasks can be done more quickly on GPUs. That has opened up new opportunities for graphics chip makers.

Dan Vivoli, senior vice president of marketing at Santa Clara, Calif.-based Nvidia, said that these new personal supercomputers have the computing power of a small cluster at 1/100th the price. And they come in a smaller form the size of a standard desktop workstation. These machines have 250 times the processing power of a conventional workstation.

Intel, of course, may not like this idea, as it means a supercomputer will have fewer CPUs and more GPUs. But the architecture has the support of Burton Smith, a technical fellow at Microsoft. The combination of CPUs and GPUs is what makes such a breakthrough possible, he said. The personal supercomputers also got an endorsement from Jack Dongarra, director of the Innovative Computing Laboratory at the University of Tennessee.

A number of universities are already using personal supercomputers equipped with GPUs, including MIT, the Max Planck Institute, University of Illinois at Urbana-Champaign, Cambridge University, and others. AMD has its own offerings in this space, including a board launched last week. Nvidia, for its part, uses the new Tesla C1070 GPU processor. There are 28 supercomputer/workstation makers using the Nvidia GPU, including Dell, Lenovo, and Asus, and Nvidia estimates there are 15 million scientists and students who could use the personal supercomputers.

The chip is also being used in high-end supercomputer clusters. One of the first GPU-based supercomputers to break into the world’s top 500 most-powerful supercomputers is a Tesla-powered machine from the Tokyo Institute of Technology. The institute’s Tsubame supercomputer will use more than 170 Tesla S1070 chips.

Just in the time for the recession, Intel is rolling out Monday its latest high-end family of microprocessors, the Core i7.

The chips have four cores, or processing brains, and are based on the company’s code-named Nehalem architecture. There will be three models for high-end consumer desktop computers and gaming machines.

It’s a big step forward for Intel, but the big question at the moment is whether anyone will buy it, given the severe recession. Intel has already shaved a billion dollars in revenue from its fourth-quarter revenue estimates.

More than 18 computer makers are expected to begin selling high-end consumer and gaming computers based on the i7 Core. The chips are the second generation based on Intel’s most advanced manufacturing process, which creates circuits that are 45 nanometers in width.

The chips will be a big competitive shot at Advanced Micro Devices and will keep Intel at the top of the consumer high-performance market. AMD, meanwhile, launched its latest chips at the server market. Intel’s Nehalem-based server and mobile chips will ship sometime next year. Hence, while Intel has the crown in consumer chips, AMD will be more competitive in server chips.

The Intel chips are significant because they have built-in memory controller and Intel QuickPath interconnect. Those features steal a page from the architecture popularized by AMD with the Opteron server chips that launched in 2003. So after five years, Intel is finally closing a gap with AMD on memory and power efficiency.

Intel will also have advantages in being able to run two threads, or the equivalent of two programs, on each core. It will also support a faster memory known as DDR3. In its marketing collateral, Intel is calling the chip family the “fastest microprocessor on the planet.”

The chips include a 965 Extreme Edition, a $999 chip running at 3.2 gigahertz for gamers. The cheapest part, dubbed the 920, runs at 2.66 gigahertz and sells for $284.

Despite a warning from Intel that shook tech stocks yesterday, rival chip maker Advanced Micro Devices said it would not change its guidance for the fourth quarter.

Speaking at the company’s analyst day at its Sunnyvale, Calif. headquarters this morning, AMD chief financial officer Bob Rivet said it’s a tough environment for tech companies. “We’d all like to forget October,” he said.

But he added that he would offer an update to the company’s earnings outlook in the first week of December. That means a downward revision is possible, but it’s not going to happen just because Intel’s outlook has fallen off a cliff.

Revenues for the fourth quarter are expected to be flat with the third quarter at $1.58 billion, Rivet said. That compares to Intel’s new revenue forecast, which was revised downward from 3 percent growth to a 12 percent drop in revenues in the fourth quarter. Rivet said consumer confidence is not good. “We are staring at that kind of environment,” he said. “It’s a storm. It will pass.”

The goal is to come out of the storm in a different position. Thanks to the company’s deal to divide itself into two parts (announced 36 days ago), one with a manufacturing focus and the other doing chip design, AMD is in the process of offloading its debt and reducing its capital spending requirements.

“Clearly, this transaction is a game changer,” Rivet said. “It takes the bankruptcy notion off the table.”

Much of the expected turnaround plan is based on the Shanghai 45-nanometer chips being launched today as well as continued growth in graphics that started earlier this year. Dirk Meyer, chief executive of AMD, said the company has a clear-eyed plan to become a profitable company. “On Shanghai, I can’t wait to hear the cash registers start ringing,” Meyer said. “What excites me as a business guy is making this company cash flow positive as a business. We’ve got the opportunity for the first time ever to turn this company into a cash-generating machine. We’ve collectively had enough of losing money.”

PC unit sales have always grown, even with the slowdown after the tech bubble. The company estimates that PC unit sales will grow 5.3 percent in 2009, which is about as slow a year as 2002. That’s not terrible, given the economic environment, Rivet said. Graphics chip units are expected to grow about 4 percent to 5 percent.

Rick Bergman, general manager of the graphics chip business, said that the company has gained market share in graphics across the board, from low-end to high-end chips, because it came up with a better strategy than competitor Nvidia. In short, AMD focused on the sweet spot of the market, designing a good $200 to $300 chip that could be doubled up in a graphics card to offer a high-end solution. Nvidia, on the other hand, stuck to a traditional strategy of designing a high-performance but power-hungry $400 chip. AMD’s strategy has paid off and its graphics chip division has returned to profitability.

In 2009, the company is targeting 45 percent gross margins and 8 percent net income. Other goals are 10 percent operating margin and 20 percent R&D costs as a percentage of revenues. All of the near-term debt is being moved to its sister organization, The Foundry Company, which will manufacture chips going forward. Capital spending will drop from $1.6 billion in 2007 to $300 million in 2009, thanks to the splitting of the company. The company will be free cash-flow positive.

The company plans to hit breakeven results in the first quarter, after a couple of years of losses. AMD promises to deliver its Yukon low-power processor for Netbook computers (meant to surf the web only) in the first half of 2009. Yukon is actually not the company’s answer to Intel’s low-power Atom processor, launched this year. The Yukon platform will be higher power and higher cost than Intel’s Atom. But the release date for its first version of Fusion, a chip design which combines a graphics chip and a microprocessor, has been moved from 2010 to 2011.

Full told, AMD added six new processors to its roadmap for the next couple of years. After AMD announced the split of the company, Intel raised questions as to whether AMD’s cross-license for patents with Intel would still be valid. But Meyer said such talk “raised my blood pressure” and that, based on everything AMD has disclosed publicly about the split, it is clear that there is no legal issue. Specifically, AMD’s agreement with Intel allows AMD subsidiaries to be licensed and that The Foundry Company, which is 43.5 percent owned by AMD, qualifies as a subsidary as defined by the agreement with Intel.

Just after taking a swing at Intel, Advanced Micro Devices is opening up another front in the battle with graphics chip rival Nvidia.

Today, Sunnyvale, Calif.-based AMD is announcing that it will release a new version of its ATI Stream technology, which is aimed at using a graphics chip’s processing capabilities to accelerate applications beyond just graphics.

With a new free release of its ATI Catalyst driver software (which enables a graphics card to work with a computer) on Dec. 10, AMD will essentially take on Nvidia’s own CUDA programming environment, which has a much stronger foothold in the market.

It may not be the most exciting development. But AMD needs to launch this technology to counter a growing advantage that Nvidia has been developing in the past couple of years by moving into Intel’s territory and making its graphics chips far more useful for everyday applications.

Graphics chips have a lot of processing power these days. A few years ago, Nvidia began promoting the idea of the general-purpose graphics processing unit, or GP-GPU. It would work in tandem with a microprocessor to speed up things that took too long to do on a central processing unit (CPU) alone. Nvidia launched a programming language, CUDA, to make it easy to create applications that run on the GP-GPU. Developers used that to create applications such as the Elemental Technologies’ Badaboom video conversion program, which can convert video from one form to another.

AMD will directly take on Badaboom, a third-party application that runs on Nvidia’s chips, by making its own Avivo Video Converter software available at no cost with the Catalyst driver. Avivo speeds the conversion of high-definition video files by 17 times compared to CPU-only conversion. A one-hour video that once took three hours to convert to run on a portable media player now takes only 12 minutes. Vendors taking advantage of the new ATI Stream software include ArcSoft, Adobe, Microsoft and Cyberlink.

The new Catalyst version 8.12 software works on the millions of graphics cards based on the new ATI Radeon HD 4000 series of graphics chips. AMD is also going to create its own programming environment to counter Nvidia’s CUDA language.

However, instead of using a proprietary language, AMD will create an open platform that will be based on industry standards. It will use a set of tools developed at Stanford dubbed Brook+ as well as a development platform dubbed OpenCL. Both of these, essentially, will allow developers to quickly create GP-GPU applications that run on AMD’s graphics chips.

Lastly, AMD said it would release a new workstation-oriented graphics board, the AMD Firestream 920, which is tuned for AMD Stream processing.

Advanced Micro Devices has been losing out to Intel for a couple of years in server microprocessors. But it hopes to make a comeback with a series of new microprocessors for servers in data centers.

Today, the Sunnyvale, Calif.-based company is introducing the latest version of its Opteron microprocessor with four computing cores. The server chip is the first to use the 45-nanometer manufacturing process that is critical to keeping up with the manufacturing might of Intel.

Of course, this AMD turnaround could run smack into the recession. Intel pre-announced lousy results today for the fourth quarter, saying that sales would now fall about 12 percent from the third quarter; previously, the company expected sales would rise 3 percent. Intel now expects $9 billion in revenues for the fourth quarter, plus or minus $300 million. It previously expected sales between $10.1 billion and $10.9 billion. Analysts were looking for $10.3 billion.

The AMD processor is code-named Shanghai and will now be known as the 2000 series and 8000 series Opteron processors. The processor speeds range from 2.3 gigahertz (which is currently the top speed for Opteron chips) to 2.7 gigahertz.

Intel will offer its own arguments as to why its code-named Nehalem processors are superior to the latest Opterons. But Steve Demski, an AMD product manager, said that AMD will continue to emphasize the performance per watt — or power efficiency — of its processors. The new chips have features such as better virtualization (the ability to run numerous types of software on one server) and range in price from $377 to $989 in 1,000-unit quantities. AMD will launch a desktop version of the chips in the first quarter.

It remains to be seen if the new Opterons will help AMD regain market share that it lost to Intel in the past two years. AMD grabbed the upper hand in servers in 2003 with the launch of the power-efficient Opteron family. But Intel struck back in 2006 with the Core 2 server chips. AMD launched its rival chip, code-named Barcelona, in 2007. But that chip had a bug and was delayed, resulting in big losses for AMD.

“We didn’t execute in our last generation and that caused pain for AMD and its customers,” Demski said. “We are in much better shape in this generation.”

AMD had originally scheduled the Shanghai launch for the first quarter of 2009, but it was able to speed up the launch to today. Last week, AMD hosted a roundtable of server makers who said they would use the new chips. Included were Penguin Computing, Appro, Colfax International, Verari Systems, and Rackable Systems. Those aren’t the biggest customers, but there was a day when AMD couldn’t get any server makers to defend it.

John Lee of Appro said that in certain segments such as high-performance computing, AMD has remained ahead of Intel. Appro uses AMD chips in a variety of supercomputers, he said. In the last couple of years, Intel has made inroads into the product line but now AMD is more competitive, he said.

Bill Calder, a spokesman for Intel, however, said that AMD remains a generation behind Intel. He noted that AMD doesn’t have a critical feature for its 45-nanometer production: high-K dielectric, which produces lower power chips. He also said that Intel chips currently hold the top six spots on the SPECPower benchmark for high performance and low power. It isn’t clear if AMD will be able to beat Intel’s new generation of chips, let alone Intel’s current chips, based on leaked reports.

Shane Rau, an analyst at IDC, said that Shanghai represents AMD’s opportunity to regain its competitive position against Intel in server processors, where customers genuinely want two legitimate suppliers to ensure fair pricing and product diversity. Intel has 85.6 percent of the server processor market in the third quarter of 2008, compared to AMD’s 14.3 percent. That compares to Intel’s share of 75.2 percent and AMD’s 24.6 percent share in the third quarter of 2006.

Beyond competitive concerns, a more pressing problem for both AMD and Intel is the economy. Although the recession is hurting information technology sales, Demski said that big companies will pay more attention to getting more for their server purchases now.

The graphics chip market defied gravity in the third quarter, showing a 22.5 percent in unit shipments compared to a year ago.

More than 111 million GPUs (graphics processing units) shipped in the quarter, up from 91 million a year ago and 94 million in the second quarter. The third quarter increase of 18 percent from the second quarter was the largest increase in six years, according to Jon Peddie Research.

The numbers show that GPUs are growing beyond the market for game computers and everyday computers. They’re also being used in video game consoles, industrial and medical systems, cash registers, kiosks, and digital signs.

The big three graphics vendors — Intel, Advanced Micro Devices and Nvidia — took market share from the smallest three: Matrox, SiS, and VIA/S3. But every vendor showed an increase in shipments. Intel took first place with 49.4 percent market share, up from 33.4 percent a year ago. (Intel ships chip sets, used in most computers, with graphics built into them. These so-called integrated graphics chip sets often sit unused alongside systems with AMD or Nvidia GPUs).

Nvidia’s share slipped to 27.8 percent from 36.4 percent a year ago. AMD’s share was 20.6 percent, up slightly from 20.5 percent a year ago.

In the desktop market, Intel had 43.9 percent share, while Nvidia’s position slipped to 32.6 percent and AMD climbed up to 20.3 percent. Overall, desktop GPUs rose only 4.7 percent to 61.9 million unit. In notebook computers, Intel lost one point to 56.2 percent, while Nvidia had 21.8 percent and AMD had 20.9 percent. Notebook GPUs grew by almost 40 percent quarter-to-quarter to 49.4 million units. They’re now 44.4 percent of the market. The sales rise was expected in part because AMD and Nvidia launched new chips during the summer, giving a big boost in graphics performance.

It’s not all good news. The third quarter typically shows growth as computer makers stock up to build inventory for fourth quarter computer sales. That means a lot of inventory was in the pipeline as computer makers got ready for a big fourth quarter. Clearly, however, with the economic turmoil hitting at the end of September, those sales may not materialize. Jon Peddie said that Q4 results are usually up from Q3, but he wouldn’t be surprised if the fourth quarter this year is flat compared to the third.

Nvidia is launching a chip set for low-end computers in a shot at Intel. In a familiar battle, Nvidia argues that low-end computers will function much better if users spend more on graphics and less on Intel processors.

In low-end computers and laptops, there is no stand-alone graphics chip. Rather, graphics is integrated into the chip set, a two-chip solution that is mounted directly on a computer’s main board, or motherboard. To date, the so-called motherboard graphics of Intel’s chip sets have been weak. Jeff Fisher, senior vice president at Nvidia, notes that 14 of the top 30 PC games won’t even run on a machine with Intel’s latest integrated graphics technology.

Santa Clara, Calif.-based Nvidia is launching a GeForce 9 Series motherboard GPU (graphics processing unit) that it says is five times faster than Intel’s fastest G45 integrated graphics chip set. Nvidia says that computers using the new chip set, dubbed the GeForce 9400, can run all of the top 30 PC games at acceptable speeds. These machines can also run with the coolest lighting features, such as the colors that make up a vibrant sunset, fully operational.

The new chip sets will be able to run high-definition video, including Blu-ray discs (as long as the machines also incorporate Blu-ray drives). It can also run certain programs such as Adobe’s high-end Photoshop better because of support for Nvidia’s CUDA programming language, which is built into the chip set. The chip set also supports other technologies previously built only into high-end graphics chips, such as physics support and support for hybrid SLI (using more than one graphics chip in a computer).

The company is starting to make a case that visual computing is becoming so ubiquitous across a range of non-game applications that spending more on a graphics processor than a CPU makes sense. But it remains to be seen if Nvidia can break Intel’s bundles, where it sells chip sets and CPUs together.

As examples, Nvidia showed side-by-side comparisons of systems running on Intel integrated graphics and Nvidia’s new chip set. The Intel machines just didn’t look good — or couldn’t even run — games such as “Spore,” “Portal,” or “The Sims 2.” The argument: parents may buy a computer for work or their kids’ homework, but most people will still want to play the occasional game on a computer.

Among those buying the argument: Apple, which is including Nvidia chip sets in its new line of laptops. But it remains to be seen if Nvidia can pull out of a slump. The company laid off 360 employees recently and Advanced Micro Devices launched a more competitive product than Nvidia did in the latest generation of hardcore graphics products.

Jon Peddie, an analyst at Jon Peddie Research, said that the rumors of Nvidia’s chip set business dying can now be put to rest. Nvidia has steadfastly denied the rumor. It should also help all of the bad news about Nvidia’s flawed chips, since Apple wouldn’t likely use the new ones if there were still flaws.

Intel’s third-quarter net income grew 12 percent thanks to a competitive edge over rival Advanced Micro Devices as well as expense cuts. But the technology bellwether painted a fuzzy picture of how the financial crisis will affect demand in the fourth quarter.

Intel’s strategy is always to use its financial might to put pressure on rivals during a downturn. You can expect the company’s posture in this one to be no different. Intel Chief Executive Paul Otellini said he expects the company “to outpace peer companies at a time when business levels are difficult to predict.” On Intel’s side: Growth of laptops and the ever-spreading use of the web.

Aside: Intel’s conference call started with a hilarious echo that turned Otellini’s voice into a kind of echoing and vibrating robot sound. Then the sound came back online. Otellini joked, “Let me apologize for that glitch. That system probably doesn’t have Intel product in it.”

He said that the financial crisis had caused “signs of stress” in the business but it is difficult to quantify and the outlook is mixed. He noted that consumer sales appear to be light overall, but demand for laptops and “net books” is strong. Corporate sales were softer than usual, while consumer sales were seasonal.

Net books are handheld, more limited computers for surfing the web. Intel will offer a rare mid-quarter update on Dec. 4 to clarify how the quarter is going. In the third quarter, Intel sold $200 million of its new Atom microprocessors for handheld, ultramobile PCs.

The Atom processors are aimed at gadgets, dubbed mobile Internet devices, which can access the web wirelessly. Otelini said Atom sales were off to a good start. Without Atom, microprocessor sales were flat and overall prices per microprocessor were down from the prior quarter.

Otellini said that the company was already well prepared for a downturn since its headcount is 20,000 lower than the peak in 2006 and the company has cut $3 billion in expenses. He also said the company has $12 billion in cash. (That’s enough to pay for about four giant chip factories). Intel had allowed itself to grow too large and diverse, so Otellini refocused the company on its core chip business.

Shares rose 4.65 percent to $16.70 a share in after-hours trading. (That’s down a third since the start of September). The world’s largest chip maker reported net income of $2.01 billion, or 35 cents a share, up from $1.79 billion, or 30 cents a share, a year earlier. The results included a net $162 million write-down of Intel’s investment in Numonyx, a flash-memory joint venture with STMicroelectronics. A year earlier, the included $125 million in restructuring and asset-impairment charges. Revenues rose 1.3 percent to $10.22 billion. In July, Intel predicted revenue of $10 billion to $10.6 billion.

The company has had a heavy investment in overseas chip sales for a long time. It reported growth in all regions, except North America, where sales fell 8.7 percent. Intel expects fourth-quarter revenue of $10.1 billion to $10.9 billion. Analysts expected $10.77 billion in the fourth quarter. Inventories are flat, with the Taiwanese and channel customers cutting back and computer makers building inventory.

Asked how bad it could get, Otellini said the range of revenue was broader than ever before because of the uncertainty. He said that gross margin percentage is expected to be 59 percent, plus or minus two percentage points. That’s fairly typical for Intel, meaning that the company isn’t counting on the world falling apart.

Otellini said he would not comment on the legal ramifications of the recent changes at Advanced Micro Devices, which is splitting itself into two companies. As far as competition goes, he said the changes at AMD have not changed the fundamental dynamics of the industry. Consequently, he said that Intel would not change its competitive strategy.

“This is not a post dotcom kind of downturn,” Otellini said. “People stopped buying computers, particularly because so much was available on eBay because” of companies that shut down. “I think this will have a different pattern. I’m of the opinion that technology will do well during a downturn because we sell tools that lift productivity.”

Come November, Intel will launch a new generation of microprocessors, dubbed Nehalem. These new chips have a better memory design that should help erase an architectural advantage that AMD has had for more than five years. Or so Intel says.

Jerry Sanders, the founder of chip maker Advanced Micro Devices, once said, “We’re the oil of the eighties” in reference to the red-hot semiconductor industry so many years ago. So it’s only fitting that oil money has come to bail out AMD in its hour of need.

The New York Times is reporting that Sunnyvale, Calif.-based AMD will split into two companies. One will focus on designing microprocessors, while the other will focus on the business of manufacturing them. The plan, articulated by AMD management earlier this year, is what the company must do to survive against the newly resurgent Intel.

Under the deal, two Abu Dhabi investment companies will invest $6 billion into both companies so that AMD can build a planned factory near Albany, N.Y., and upgrade a Germany factory. AMD will own 44.4 percent of the new entity, to be named the Foundry Company. The Advanced Technology Company, formed by the Abu Dhabi government, will own the rest.

AMD put up a valiant fight against Intel, starting in 2003, when it launched the Opteron processor. The design was both energy and memory efficient, helping to cut electricity costs just as they were spiraling out of control for data center managers. It took years for Intel to recover.

But the world’s biggest chip maker had many times more engineers than AMD and far more money to pour into factories. In 2006, Intel came up with a better chip design, based on its Core architecture. It retook just about all of the market share that AMD had gained.

Intel’s resurgence made this outcome inevitable. AMD lost money and was on the ropes; in June, it had $5.3 billion in debt and $1.6 billion in cash. The investment in the chip-making process for upcoming 45-nanometer factories is so big that chip makers would have to generate $13 billion over five years to justify the investment.

Now it has come to this. Intel will be competing against foreign governments. They are so far willing to put up the multibillion-dollar ante that it takes to compete head on with Intel. The Foundry Company will make chips for AMD and other customers too. It will in effect be a contract chip manufacturer and it will license chip manufacturing processes from IBM.

It takes about $13 billion in revenues over five years to justify a big investment in next-generation chip manufacturing technology. That explains why the ranks of chip makers are thinning out.

Walter Ng, a vice president at Chartered Semiconductor Manufacturing, one of the big contract chip manufacturers, said that the math just doesn’t add up for many companies to keep their own chip factories. Big companies such as Intel can afford to spend a couple of billion dollars on a chip manufacturing process — which is the recipe for making chips. Then they have to spend $2 billion or so on each factory and its accompanying chip-making equipment.

The $13 billion figure represents a six-fold return on the investment into a chip process investment of $2 billion, which is the estimated cost for the newest 45-nanometer factories being built now. Ng was one of the speakers today at the Global Semiconductor Alliance Intellectual Property conference in Santa Clara, Calif. If chip makers can’t make that much revenue in five years, then they have to consider farming out their chip production to companies such as Singapore-based Chartered or its Taiwanese rivals, Taiwan Semiconductor Manufacturing Co. or UMC.

It’s no surprise that a lot of chip makers are joining together in alliances to jointly develop the chipmaking processes. It’s a smart use of resources to share the costs with someone else. IBM leads an alliance of chip makers that includes Advanced Micro Devices and others. Texas Instruments said it would develop a 45-nm process but after that it would likely ally itself with others.

The risks involved in doing chip R&D have led to foundries doing a lot of the production for hundreds of fabless chip companies. Nvidia is one of the most successful of these fabless companies. But much of the growth in the chip business is now in IP companies, or those that develop a component that can be reused in a number of chip designs.

Jordan Selburn, an analyst at iSuppli, said that the IP market hit $1.5 billion in revenues in 2007, with ARM, the mobile processor IP developer, accounting for 34 percent of market sales. Adam Traidman, a group marketing director at chip design software company Cadence Design Systems, said that there are more than 300 companies that focus on making IP for chips.

Selburn said that IP revenues should grow twice as fast as the overall chip market in the coming years. Logically, then, the best place to invest in the chip industry these days is the IP market.

That way, you don’t have to cut a $2 billion check.

Intel launched new six-core server microprocessors today, hoping to widen the performance gap with rival Advanced Micro Devices and help corporations with ever-increasing Internet-related processing chores.

More than 50 server makers are launching new models of servers with the Intel chips today. Those models range from four-socket machines to 16-socket machines. Each socket can hold a six-core chip, known formally as Intel’s Xeon 7400 series processors and previously code-named Dunnington.

Tom Kilroy, general manager of Intel’s Digital Enterprise Group, said that the new chips deliver 50 percent better performance and 10 percent lower power consumption than the previous line of Xeon 7300 series chips. He spoke at a press conference in San Francisco. The event was attended by server makers and customers from big companies such as Oracle, MySpace, Verisign and Yahoo.

The chips are aimed at enabling big companies to deal with growing Internet traffic. Richard Buckingham, MySpace director of system engineering, said that the company has more than 12,000 servers that must be updated every couple of years to keep up with growing traffic requirements. Power-efficiency and more performance per power consumed are key concerns in adding new equipment, he said.

The Intel chips are based on 45-nanometer manufacturing technology that Intel has been shipping for some time. The introduction shows that the world’s biggest chip maker still has an advantage over AMD in server chips. AMD’s server market share hit its peak in 2006, but ever since then Intel has become more competititive and now has about 90 percent of the server chip market, according to International Data Corp.

Kilroy said the new chips can run software anywhere from 14 percent to 48 percent faster than the previous Xeon chips. Buckingham of MySpace said that the company would likely adopt the new chips and that would enable it to tackle future processing tasks such as letting users play and share high-definition videos.

Jen-Hsun Huang, chief executive of Nvidia, started his graphics chip company in 1993 and is now the last man standing. Back then, no one could have predicted that PCs and game machines would spawn the powerful visual computing we have today. In a speech in San Jose, Calif., Huang talked about how video games and movie special effects are only the tip of the iceberg for visual computing, which encompasses everything from digital art to medical imaging. Huang is celebrating the growth of this ecosystem this week with his own new visual computing conference, dubbed Nvision 08. But it’s a turbulent time for Nvidia as the company struggles against competitors and its own product bugs. After his speech, Huang took questions on a wide range of topics at a press conference. I’ve blended questions from the general press Q&A with my own one-on-one questions in this edited transcript.

VB: I remember when I interviewed you 14 years ago. You talked about how your graphics chips could be used as “Windows accelerators.” It was like there was no other use for a 3-D chip.
JH: The real breakthrough for our industry came when we at Nvidia discovered this perspective: graphics is not just putting pixels up on the screen, but graphics can be a medium for artistic expression. That was when we decided we had to build programmable shaders (subprograms that add custom special effects to a 3-D scene). We didn’t want graphics to just all look the same. This medium has a real artistic element. To deal with that, we had to create an infinite palette. To do that in a computer architecture, we had to make it programmable. That was the insight that allowed us to see that programmable shaders were the future. We blindly believed in it and made everyone believe it. Now, when you say that computer graphics is an artistic medium, you don’t sound like a psycho. That notion isn’t more than 10 years old.

VB: A lot of people in the world get excited about computer-animated art in movies or video games. But you’re excited about digital still art.

JH: If you think about what these people are doing, computer-generated art is part math, part imagination, and part programming. You have to know what the technology is capable of doing. It’s a complicated thing. It’s not for your average artist. Yet, as extraordinary as it is and as beautiful as it can be, it is really complicated to make worthy. Most people think all of the valuable things in life are expensive. But it’s hard to make a digital art piece expensive because it’s easy to replicate. The whole point is that it is written in software so that it can be generated flawlessly over and over again. There are complicated issues to solve with digital art. It doesn’t take away from how amazing digital art can be. I don’t know how artists even come up with the combination of skills to do it. I admire them for it and I hope that someone can figure ou