The processor will be targeted at four-socket servers, says Shannon Poulin, Xeon platform director at Intel. Each physical core will be able to run two threads simultaneously, giving the chip 64 virtual processing cores on servers.

Intel’s CEO Paul Otellini has described Nehalem-EX as the company’s fastest processor to date. The chip maker announced the processor last year, and said it would release the chip in the first half of this year, but did not provide an exact release date.

Poulin declined to provide the clock speed of the chips. However, the company has said it will include 24MB of cache and 2.3 billion transistors.

Intel is targeting the chip at high-end systems running data-intensive applications such as databases. IBM has said when the chip is released it would implement Nehalem-EX chips in the company’s System x EX5 servers.

The chip will be made using the 45-nanometer process and be based on the Nehalem microarchitecture, which integrates the memory controller and improves system speed by cutting data bottlenecks that plagued Intel’s earlier chips.

Intel is also including new technologies like MCA recovery error correction, which could make servers more fault tolerant and provide greater uptime, he says. The processor will be able to detect system errors originating in the CPU or system memory and correct them by working with the operating system. Some of these technologies have been adapted from Intel’s high-end Itanium processors, which are based on a separate chip architecture and go into fault-tolerant systems.

The new processor will also contain separate buffered memory chips, which can store data temporarily alongside the main memory for faster task execution.

Intel will also offer four memory channels per processor, Poulin says. That will put it on par with AMDs’ twelve-core Opteron server processors, code-named Magny-Cours, which also offer four memory channels per processor. More channels provide more memory bandwidth to run programs faster.

AMD has started shipping Magny-Cours processors, with Intel’s Nehalem-EX set to intensify the battle between the rivals, as both companies reach out to claim benchmark crowns, says Nathan Brookwood, principal analyst at Insight 64. Intel will build large amounts of cache inside Nehalem-EX that could help the processor deliver faster performance, but AMD’s Magny Cours has more physical cores per chip.

Benchmark results could vary depending on the type of task assigned, but performance is just one part of the story, he says. Nehalem-EX could reach out to new markets with additional features like fault-tolerance, Brookwood says. It could tread into territory of high-end servers dominated by Intel’s Itanium and chips based on RISC (reduced instruction set computer) architecture, which includes IBM’s Power and Sun’s Sparc processors.

Last year, when the Nehalem EX was first announced, Boyd Davis, Intel’s general manager for server platforms in the Intel marketing department, said: “This is going after a market that was limited to being served by RISC architecture”, and “We think Nehalem-EX will represent a pretty significant opportunity on the overall server and hardware market”.
But AMD could hold an advantage over Intel’s Nehalem-EX in pricing. Intel could charge a premium for Nehalem-EX chips, while AMD chips could deliver better bang-for-buck per core with Magny-Cours, Brookwood says.

Intel’s Poulin also says the company would release its next-generation Xeon server processors based on the Westmere microarchitecture later this month. The processors will be targeted at two-socket servers, Poulin says. Chips for four-socket servers will be released next year.

Intel says server chips based on Westmere will contain up to six cores. The company last month said the six-core chip contained 1.17 billion transistors and 12MB of cache. The six-core chips will deliver improved performance and power savings compared to earlier quad-core chips, according to the company.

The Westmere-EP chips will fall under the Xeon product line and will be made using the 32-nanometer process. The last refresh for server chips was in March last year, when the company announced a range of Xeon 5500 series and 3500 series chips based on the Nehalem architecture. The chips were made using the 45-nm process.

www.computerworld.co.nz

Advanced Micro Devices plans to release a processor in its “Fusion” line that will be positioned for the netbook market, putting it in competition with the Intel Atom, and, to a lesser degree, the ARM processor.

The “Fusion” program is AMD’s (NYSE: AMD) long-term project to integrate its CPU cores with graphics processor cores from ATI, which it acquired in 2006. The first Fusion processors are expected some time early next year. Intel (NASDAQ: INTC) has its own integrated processors, the Westmere family of Core i5/i7 chips, which feature integrated graphics in dual-core CPUs.

AMD hasn’t yet hammered out all the specifics, but the netbook part will run in the 10-15 watt range, which is similar to the Intel Atom, and “it will have a good processor integrated with graphics, so you won’t need the [Nvidia] Ion graphics to give it half-decent performance,” according to Nigel Dessau, chief marketing officer at AMD.

The processor is likely from the “Bobcat” line of chips first disclosed last November. It will be designed for 12-inch and smaller screens. Intel has defined netbooks by screen size, and only supports 10-inch and smaller devices. AMD chose to make a new part rather than whittle down an existing one for the low-power, low-profile environment of netbooks.

www.internetnews.com

They also used the two-qubit superconducting chip to successfully run elementary algorithms, such as a simple search, demonstrating quantum information processing with a solid-state device for the first time. Their findings will appear in Nature’s advanced online publication June 28.

“Our processor can perform only a few very simple quantum tasks, which have been demonstrated before with single nuclei, atoms and photons,” said Robert Schoelkopf, the William A. Norton Professor of Applied Physics & Physics at Yale. “But this is the first time they’ve been possible in an all-electronic device that looks and feels much more like a regular microprocessor.”

Working with a group of theoretical physicists led by Steven Girvin, the Eugene Higgins Professor of Physics & Applied Physics, the team manufactured two artificial atoms, or qubits (“quantum bits”). While each qubit is actually made up of a billion aluminum atoms, it acts like a single atom that can occupy two different energy states. These states are akin to the “1″ and “0″ or “on” and “off” states of regular bits employed by conventional computers. Because of the counterintuitive laws of quantum mechanics, however, scientists can effectively place qubits in a “superposition” of multiple states at the same time, allowing for greater information storage and processing power.

For example, imagine having four phone numbers, including one for a friend, but not knowing which number belonged to that friend. You would typically have to try two to three numbers before you dialed the right one. A quantum processor, on the other hand, can find the right number in only one try.

“Instead of having to place a phone call to one number, then another number, you use quantum mechanics to speed up the process,” Schoelkopf said. “It’s like being able to place one phone call that simultaneously tests all four numbers, but only goes through to the right one.”

These sorts of computations, though simple, have not been possible using solid-state qubits until now in part because scientists could not get the qubits to last long enough. While the first qubits of a decade ago were able to maintain specific quantum states for about a nanosecond, Schoelkopf and his team are now able to maintain theirs for a microsecond—a thousand times longer, which is enough to run the simple algorithms. To perform their operations, the qubits communicate with one another using a “quantum bus”—photons that transmit information through wires connecting the qubits—previously developed by the Yale group.

The key that made the two-qubit processor possible was getting the qubits to switch “on” and “off” abruptly, so that they exchanged information quickly and only when the researchers wanted them to, said Leonardo DiCarlo, a postdoctoral associate in applied physics at Yale’s School of Engineering & Applied Science and lead author of the paper.

Next, the team will work to increase the amount of time the qubits maintain their quantum states so they can run more complex algorithms. They will also work to connect more qubits to the quantum bus. The processing power increases exponentially with each qubit added, Schoelkopf said, so the potential for more advanced quantum computing is enormous. But he cautions it will still be some time before quantum computers are being used to solve complex problems.

“We’re still far away from building a practical quantum computer, but this is a major step forward.”

Authors of the paper include Leonardo DiCarlo, Jerry M. Chow, Lev S. Bishop, Blake Johnson, David Schuster, Luigi Frunzio, Steven Girvin and Robert Schoelkopf (all of Yale University), Jay M. Gambetta (University of Waterloo), Johannes Majer (Atominstitut der Österreichischen Universitäten) and Alexandre Blais (Université de Sherbrooke).

www.sciencedaily.com

they are all capable of driving HD content, can surf the web for 8 hours, and will cost round and about 250 USD. They are all focused on Ubuntu; ARM and several of its licensees have worked with Canonical to bring their ARM-port up to speed. I must say that I am pretty much impressed.

www.zdnet.co.uk
www.osnews.com

AMD’s shares jumped 18 percent to $5 on the news on Tuesday after Wall Street waited for months for the struggling chip maker to formulate its so-called “smart asset” strategy to focus more on chip design and less on costly factories.

Advanced Technology Investment Company (ATIC), an Abu Dhabi state-owned venture capital firm, said it will invest $2.1 billion for a 55.6 percent stake in the venture, of which $700 million will go directly to AMD, which will hold the remaining stake. The two will divide the venture’s board seats equally.

ATIC also committed to investing another $3.6 billion to $6.0 billion over 5 years to fund the venture’s expansion. The 3,000-person new company will hold AMD’s two plants in Dresden, Germany and make all of its central processing units, as well as chips for other companies.

Another Abu Dhabi government company, Mubadala Development Co, will spend $314 million to increase its stake in AMD to 19.3 percent from 8.1 percent and gain a seat on AMD’s board.

Brian Mata, an analyst at IC Insights in Arizona, said AMD desperately needed the boost. The company has posted seven consecutive quarters of losses and is forecast by Wall Street to report another quarterly loss next week.

“The key thing is the financial backing from Abu Dhabi because AMD was essentially out of money,” said Mata.

He said AMD can focus on design and has the money to try to catch up with Intel, but the new venture has a challenge competing with existing contract chip manufacturers.

“The foundry business is already pretty entrenched,” he said, citing Taiwan Semiconductor Manufacturing Co Ltd, United Microelectronics Corp, Singapore’s Chartered Semiconductor Manufacturing Ltd and China’s Semiconductor Manufacturing International Corp.

GOING FABLESS

AMD has lost market share to Intel and in the last few years was forced to weigh the price of its pride in owning the semiconductor fabricating plants, or “fabs,” which most other chip makers gave up long ago.

It has also been hit by problems with its high-end personal computer and server Barcelona chip, and had bumps along the road after acquiring graphics chip maker ATI.

Despite the jump in AMD’s share price on Tuesday, the stock remains far below its 12-month high of $14.73.

“Today is a landmark day for AMD, creating a financially stronger company with a tightened focus,” Dirk Meyer, president and chief executive officer of AMD, said in the statement.

The new venture, temporarily called Foundry Company, will assume all $1.2 billion of AMD’s manufacturing operations debt so the remaining company can compete hard against Intel, which sells about 80 percent of the central processing units at the heart of the world’s 1 billion PCs. AMD makes the rest.

“A duopoly is a very helpful market structure for companies and a very profitable one. AMD needs to improve its execution to enjoy the potential benefits,” said JoAnne Feeney, an analyst with FTN Midwest in Boston, who has “buy” ratings on AMD and Intel. “Separating the companies into design and manufacturing will improve their execution.”

The venture plans to break ground next year for a factory in upstate New York, employing 1,400 people, if New York state will transfer financial incentives to the new company.

The chief executive will be Doug Grose, senior vice president of technology at AMD and its new chairman will be Hector Ruiz, now chairman of AMD.

The venture will upgrade one of the two plants in Dresden and build the plant in New York to the latest technology standards, AMD said. It said the venture, which will be on AMD’s balance sheet, may ultimately build a fab in Abu Dhabi.

ATIC Chairman Waleed Al Mokarrab said on a conference call that this was his company’s first major investment and “though owned by the Abu Dhabi government, ATIC will be directed by commercial principals that will generate commercial returns.”

Mubadala, which holds stakes in sectors ranging from energy to aerospace, purchased an 8.1 percent share of AMD in 2007 for $622 million. It will boost that to 19.3 percent by buying 58 million newly issued shares and warrants for 30 million more.

The deal is to close at the end of 2008 or early in 2009, if approved by stockholders. It will also be reviewed by the U.S. government inter-agency Committee on Foreign Investments in the United States (CFIUS).

The deal will be welcomed by IBM, which has worked closely with AMD and other chip makers to improve chips. The company will be part of the IBM technology alliance, making it easier to build chips for members.

www.reuters.com

The 20th annual HOT CHIPS conference at Stanford University is an academic show, not a vendor-driven one. That means the biggest guns share the stage with the littlest pistols; and it gives the little guys a chance to shine.

In this case, a Chinese processor project found itself sandwiched between presentations from AMD and Intel, with its own unique story to tell.

The Chinese Academy of Sciences is a nationally-funded institution with five main areas of focus, one of which is technology. Within that area is the Institute for Computing Technology, which designed the Godson processor, a project that began in 2001.

China decided to support microprocessors because it has come to realize CPU design is important and one of national strategic importance, said Zhiwei Xu, chief technology officer and a professor at the institute during his speech here.

Now on its third generation of Godson, the ICT has managed to triple performance with each generation, although he admitted it still has a long way to go to close the gap with giants like Intel (NASDAQ: INTC), AMD (NYSE: AMD) and IBM (NYSE: IBM), but added “we are doing our best to join the international community.”

Godson-2 and 3 are scalable 64-bit single core processors built on 90 nanometer design. They are instruction-compatible with the MIPS III processor architecture, and Xu said the company has a license from MIPS for such compatibility. The operating system of choice for Godson-powered computers, said Xu, is Linux.

The current Godson-2 generation, E and F, are both 1Ghz processors, consuming from 3 to 7 watts of power. Godson-2E has an on-chip DDR (define) controller while Godson-2F has an on-chip DDR2 (define) controller. A Godson-2G and –2H are also planned for computers as well as System on a Chip (SoC) designs.

Xu then introduced the Godson-3 design, which will be a four-core, 1.0Ghz chip on a 65nm process design and consume just 10 watts of power. The core will be reconfigurable into one of two purposes and it will have on-the-fly x86 binary translation, which Xu said would be ten times as fast as software-only emulation.

The x86 compatibility still needs work. The goal is to reach 80 percent of the native MIPS performance. Currently, it’s between 48 and 58 percent, according to Xu.

Godson-3 can be reconfigured into one of two designs: general purpose core, and multipurpose core. The general purpose core is just as the name implies, and does more x86 emulation, media acceleration, and general processing. The multipurpose core is for LINPACK programming, a Fortran-style of processing, biological and digital signal processing.

Godson-3 will use a crossbar and mesh interconnect to support rapid communication between the cores. Xu said the processor would be capable of 16 gigaflops at 1Ghz, which is a pretty good showing at that speed. It would put it at a comparable performance range to Intel’s mass market quad core processors.

In 2009, ICT hope to introduce an eight-core processor that is essentially a multi-chip module like Intel’s quad core Xeons, which are two dual core processors on one die. It will also be a combination of general purpose and multipurpose cores, rather than all of one style like the four core chip.

This will bump the power draw to 20 watts, still far lower than any other eight-core processor. The ultimate goal, said Xu, is to build a petaflop (define) computer running Godson-3 by 2010.

www.internetnews.com

Products based on the new microarchitecture will deliver high performance and energy efficiency. This “best of both worlds” approach is expected to extend Intel’s processor leadership in future mobile, desktop and server market segments.

“The Core name is and will be our flagship PC processor brand going forward,” said Sean Maloney, Intel Corporation executive vice president and general manager, Sales and Marketing Group. “Expect Intel to focus even more marketing resources around that name and the Core i7 products starting now.”

Intel Core i7

The Intel Core processor brand name has gained broad awareness, preference, and market momentum over the past several years. The Intel Core name remains the logical choice for Intel’s latest family of processors. The Intel Core i7 processor brand logo will be available for high-performance desktop PCs with a separate black logo for Intel’s highest-end “Extreme Edition.” Intel will include processor model numbers to differentiate each chip.

Initial products based on this microarchitecture are expected to be in production in the fourth quarter of this year. These processors will feature Intel® Hyper-Threading Technology, also known as simultaneous multi-threading, and are capable of handling eight software “threads” on four processor cores.

www.intel.com