Nanotech - The Circuits Blog

Advanced Micro Devices discussed the Hemlock high-end graphics card due next week and third-generation ultra-thin laptop technology, among other topics, at the AMD Financial Analyst Day on Wednesday.

AMD Vice President Rick Bergman holds up the 'Hemlock' graphics card at AMD Financial Analyst Day on Wednesday. The product is due next week.

(Credit: AMD)

"Hemlock will get launched next week," said AMD Senior Vice President Rick Bergman, speaking Wednesday morning at the conference which was streamed live. "It's in production. You'll be able to buy it at e-tailers around the world. You can see there are two GPUs. Five Teraflops out of this baby," he said. (GPU stands for graphics processing unit. A teraflop is a trillion floating point operations per second, a key indicator of graphics performance.)

Hemlock is expected to be appear as an HD 5900 series product--what some reports have called the HD 5970.

Bergman also addressed AMD's third-generation "Nile" ultra-thin laptop platform. "Bring the real PC experience into the ultra-thin. Battery life well north of seven hours," Bergman said. This is due ... Read more

Despite persistent rumors, Nvidia's chief executive says the graphics chip supplier is not working on an Intel-compatible chip.

Nvidia CEO Jen-Hsun Huang

(Credit: Nvidia)

In an exclusive interview with CNET Thursday, I asked CEO Jen-Hsun Huang about the possibility of Nvidia coming up with its own x86 (Intel-compatible) chip technology, after the company reported strong third-quarter earnings. A recurring rumor has it that Nvidia is developing a chip that would be able to run the same software that runs on all Intel- and AMD-based PCs worldwide.

"No," he said when asked if there was any truth to the rumor. "Nvidia's strategy is very, very clear. I'm very straightforward about it. Right now, more than ever, we have to focus on visual and parallel computing."

Huang went on to describe where the chip supplier sees its best opportunities for growth. "Our strategy is to proliferate the GPU (graphics processing unit) into all kinds of platforms for growth," he said. "GPUs in servers for parallel computing, for supercomputing--and cloud computing with our GPU is a fabulous growth opportunity--and streaming video."

"And also getting our GPUs into the lowest power platforms we can imagine and driving mobile computing with it," Huang added, referring to its Tegra chip, which, for example, powers Microsoft's Zune HD media player.

Despite Huang's denials, Doug Freedman of Broadpoint AmTech is the latest to postulate that Nvidia will enter the x86 central processing unit market. "We feel Nvidia could become a supplier of x86 CPUs by necessity, perhaps in the next 12 months (if not sooner) to preserve both GPU and chipset revenue," Freedman said in a note recently.

"We believe the company has hired former Transmeta staff extensively," Freedman said. Transmeta was at one time a low-power x86-compatible chip supplier. Earlier this year, Intellectual Ventures acquired the patent portfolio of Transmeta.

Huang also dismissed the the possibility of Nvidia using Globalfoundries as a manufacturing partner--typically referred to as a "foundry" or a "fab"--for its chips, after saying in the earnings conference call that Nvidia's longstanding foundry partner--Taiwan Semiconductor Manufacturing Company (TSMC)--was not allocating it enough capacity.

"Globalfoundries is an AMD fab, right?" he said. "Globalfoundries is AMD's fab. Our strategy is TSMC."

Buoyed by Netbook sales, shipments of Intel graphics chips surged and Advanced Micro Devices gained on Nvidia in the third quarter.

Third-quarter shipments of graphics processors jumped 21.2 percent over the second quarter, according to market researcher Jon Peddie Research. Graphics chips drive the images produced on PC users' screens.

A total of 119.45 million units were shipped in the third quarter, exceeding the record 111 million units that shipped in the third quarter of 2008, according to Jon Peddie, president of Jon Peddie Research. "So the market has caught up with, and exceeded, last year's highs. The crash of fall 2008 is now behind us," he said in a statement.

The third quarter exceeded a robust second quarter. "Q2 was already a great quarter clearly signaling the holidays will be robust for PCs and the industry in general," Peddie said.

AMD gained on discrete graphics chip leader Nvidia in quarter-to-quarter growth.

(Credit: Jon Peddie Research)

AMD showed the biggest jump in quarter-to-quarter growth at 30 percent, followed by Intel at 21 percent. But Intel dominates raw shipments. "Intel shipped the most parts at 63 million, over twice as many as its nearest competitor Nvidia," according to Peddie, who said Intel had a 53 percent share of the market in the third quarter. Nvidia was second with 24.9 percent, followed by AMD with 19.8 percent.

Surging Netbook shipments are behind the big Intel numbers. Integrated graphics in notebooks, which includes Netbooks, increased 27 percent over the second quarter. Integrated graphics are built into supporting Intel silicon called chipsets.

"Netbooks will remain popular but they will not have the high market share they had during the recession when they were just introduced. Rather, consumers are expected to 'buy up' in the next quarter," according to Peddie.

Fourth-quarter shipments may not be as strong as the third quarter, however. "The channel is full...That suggests that while Q4 is typically a good quarter for PCs, the quarter-to-quarter growth in Q4 may not be as robust as Q3. Graphics are a great leading indicator. The graphics go in before the PC is built or shipped," Peddie said.

Advanced Micro Devices posted a third-quarter loss of $128 million, lower than Wall Street projections, while also reporting revenue that beat expectations.

The loss, at 18 cents a share, compares with a loss of $134 million, or 22 cents a share, for the same period last year. Analysts had expected a loss of 42 cents a share.

Revenue was $1.4 billion, an 18 percent increase over the second quarter of this year, while falling 22 percent compared to the third quarter of 2008. Forecasts had called for only $1.3 billion in revenue.

"There was strength in notebooks and China," said Dirk Meyer, AMD president and CEO, speaking during the company's earnings conference call on Thursday afternoon. He added that there is "an increased focus on small form factor" laptops at AMD and that upcoming inexpensive, thin laptops based on AMD processors should be priced lower than Intel-based offerings. And Meyer said AMD will broaden its processor offerings in this area going into the holiday season.

Meyer also spoke to AMD's future 32-nanometer silicon. Products codenamed "Fusion" that combine the graphics function with the main processor will be based on 32-nanometer technology and ship in the second half of 2010, Meyer said.

AMD is currently moving most of its production to 45-nanometer-based processors. Intel, on the other hand, will begin to move to 32-nanometer by the end of this year. Generally, the small the geometry, the faster and more power-efficient the chip is.

Addressing graphics processing units (GPUs), Meyer said that its recently-introduced 5800 series products have been well received but that the average selling prices of GPUs were down compared to the prior quarter and are still below central processing units or CPUs, which are higher.

"Growth in microprocessor and graphics unit shipments drove an 18 percent sequential revenue increase, while improved factory utilization rates, higher microprocessor average selling price, and an increase in 45 (nanometer) product shipments resulted in a gross margin improvement from the prior quarter," Meyer said in a statement.

AMD expects its product company (non-manufacturing-related) revenue to be up modestly for the fourth quarter of 2009.

AMD was the world's second-largest seller of microprocessors in the second quarter of 2009 with an 11.9 percent share of global revenue behind market-leader Intel, according to market researcher iSuppli.

Updated at 3:30 p.m. PDT: adding comments from earnings conference call.

Updated at 4:40 p.m. PDT: adding to discussion of next-generation Nvidia Ion chip.

As graphics kingpin Nvidia tries to reshape itself into a broad-based computing company, it is taking big gambles with potentially big payoffs, while it fends off challenges from rivals Intel and Advanced Micro Devices.

The world's largest supplier of standalone graphics chips for PCs needs to grow. Established markets have matured and Nvidia must seek out other ways to make money.

Nvidia's Fermi-based GF100 graphics processor.

"In almost every market they have entered they have become dominant," said Jon Peddie, president of Jon Peddie Research, which tracks the graphics chip market. "Almost 90 percent market share in the workstation business and 55 to 65 percent in the graphics business. But if you're that successful you can't really grow the market anymore, and if you want to keep growing your company, then you have to get into new markets."

Enter supercomputing and Nvidia's brand-new Fermi architecture. "That's a huge market and big margins," said Peddie. Fermi was announced last week at an Nvidia conference to great fanfare when prestigious Oak Ridge National Laboratory said it plans to use Fermi in a future supercomputer.

It would be an understatement to say that the Fermi chip potentially packs a computing wallop. The chip integrates an astounding 3 billion transistors, about three times the number of transistors in Nvidia's most powerful graphics chip now on the market, and it has been designed with features that make it more suitable for high-performance computers, the first time that Nvidia has architected a chip this way.

Fermi GPUs, each containing 512 processing cores, would enable "substantial scientific breakthroughs" that would be impossible without the new technology, Jeff Nichols, Oak Ridge's associate lab director for computing and computational sciences, said last week.

Nvidia hopes to parlay this computing power into the mainstream. (For a comparison of Fermi with AMD's newest graphics chip see: ATI and Nvidia face off--obliquely.)

"Fermi will offer Nvidia the opportunity to grow our consumer business by having the fastest raw graphics power," said Drew Henry, general manager of Nvidia's bread-and-butter GeForce graphics business. "But it's also going to expand our business by allowing people to process better video and photo applications and to use the GPU for many, many more mainstream applications." (GPU stands for graphics processing unit.)

... Read more

Nvidia's new Fermi chip is being billed as a supercomputing chip but Nvidia doesn't want you to forget that it is also aimed at Apple's Snow Leopard and Windows 7.

The Fermi chip was announced with much fanfare on Wednesday as key silicon in a future supercomputer from Oak Ridge National Laboratory. But, wait, Fermi is also going to be great at accelerating stuff in Snow Leopard and Windows 7--not to mention a great gaming chip, according to Bill Dally, chief scientist at Nvidia who spoke during a conference call with analysts on Thursday.

The Fermi graphics processing unit (GPU)--which packs 512 processing cores--will support DirectX-11, a technology for speeding certain multimedia software in Windows 7, and also support an analogous technology in Snow Leopard, OpenCL.

"A lot of (the chip's new) features accelerate key consumer applications. Both Snow Leopard and Windows 7 enable the GPU to be used as a co-processor to accelerate third-party applications," Dally said. With a "discrete (standalone) GPU they can get very good performance on these applications," he said.

Applications that Nvidia says will be accelerated by the Fermi chip

(Credit: Nvidia)

Dally gave examples (see graphic) of consumer titles such as Adobe's Creative Suite, Motion DSP's vReveal (for fixing photographs), and Badaboom (for creating iPod video).

He offered a qualifier, however. "We are paying a bit of a compute tax in that we launched a part where a lot of the consumer compute applications haven't really taken hold yet. But over time as more consumer computer applications are developed that take advantage of our compute (consumer) features...I think it's going to give us a big leg up," he said.

And being an Nvidia chip, games are a big target market. "Fermi adds value to games by doing exactly the same kind of scientific simulations that we use to predict climate and to understand the genome and other things," according to Dally. "A great example of that is our PhysX package that basically does physical simulations to make games appear more real."

He also explained why the chip was billed as a supercomputer chip initially and not a gaming chip. "It's a zero-sum game. You have a certain amount of die (chip) area, a certain power budget. It is the case that we put a bunch of die area into double-precision floating point, a bunch of die area into ECC. And for gaming graphics applications, those give less returns than they do for the scientific applications," he said. Double-precision floating point operations are used heavily in scientific computing. ECC, or error correcting code, is a technology that can correct data errors on the fly.

And Dally explained how Fermi can be scaled down to lower-end chips used in the gaming and consumer segments. "We're not talking about other (chips) at this point in time but you can imagine that we can scale this part by having fewer than the 512 cores and by having these cores have fewer of the features, for example less double-precision," he said.

All the Fermi products, including gaming and professional workstation chips, will be announced "pretty close together." Chips are expected sometime in the coming few months.

And how does Fermi stack up against current public information about Intel's future "Larrabee" graphics chip? "We can't compare anything to Larrabee until it shows up and can actually be measured," said Jon Peddie, president of Jon Peddie Research, which tracks the graphics chip market. "But remember, Larrabee was started over two years ago and both ATI and Nvidia have had two new designs out since then," he said. "So the pressure will be on Intel to chase fast-moving ATI and Nvidia," Peddie said. ATI, which is Advanced Micro Devices' graphics chip unit, already has a chip in stores--the Radeon HD 5800-- that supports Windows DirectX-11.

Updated at 6:40 p.m. PDT: adding additional information about Fermi chip .

Oak Ridge National Laboratory announced plans today for a new supercomputer that will use Nvidia's next-generation GPU architecture, codenamed "Fermi."

The Oak Ridge and Fermi announcements were made at Nvidia's GPU Technology developer's conference, which kicked off Wednesday in San Jose, Calif. The Fermi chip integrates three billion transistors, about three times the number of transistors in Nvidia's most powerful graphics chip now on the market. In the future, the chip will also find its way into Nvidia's GeForce product line for PCs.

Oak Ridge's supercomputer will be used for research in energy and climate change and is expected to be 10 times more powerful than today's fastest supercomputer, according to a joint statement from Oak Ridge and Nvidia. The architecture would use both graphics processing units (GPUs) from Nvida and central processing units (CPUs), according to Nvidia. Intel and Advanced Micro Devices, among others, make the CPUs.

High-end GPUs today typically contain hundreds of processing cores, allowing them to accelerate certain types of computational tasks much more efficiently, and thereby much faster, than CPUs. The new Fermi chip will have a little more than twice as many cores as current high-end Nvidia GPUs, according to Nvidia CEO Jen-Hsun Huang, jumping from 240 cores to 512 in Fermi.

Nvidia CEO Jen-Hsun Huang shows graphics card with new Fermi chip Wednesday

(Credit: Nvidia)

Fermi GPUs would enable "substantial scientific breakthroughs" that would be impossible without the new technology, said Jeff Nichols, Oak Ridge's associate lab director for Computing and Computational Sciences, in a statement.

"With the help of Nvidia technology, Oak Ridge proposes to create a computing platform that will deliver exascale computing within ten years," Nichols said. Exascale computing proposes to go beyond petaflop performance (a thousand trillion) to one million trillion operations per second.

Oak Ridge also announced it will be creating the Hybrid Multicore Consortium focused on computing with different types of processor architectures. The goals of this consortium are to work with the developers to run applications on the next generation of supercomputers built with CPUs and GPUs.

Nvidia's rival AMD recently announced one of its fastest graphics chips yet that packs 2.15 billion transistors and supports Microsoft's upcoming DirectX 11 programming interface for accelerating graphics and general-purpose computing in Windows 7. The ATI Radeon HD 5870 has received consistently positive reviews, beating comparable Nvidia chips and, most importantly, it is available now.

SAN FRANCISCO--Heads up, Nvidia. Intel demonstrated its Larrabee graphics chip for the first time Tuesday at the Intel Developer Forum.

Larrabee will be Intel's first discrete, or standalone, graphics processor in about 10 years and is expected to compete with graphics chips from Nvidia and AMD's ATI unit. The demo used an early "stepping," or version, of Larrabee, which is expected to come out commercially sometime next year.

Larrabee will be targeted initially at the gaming market. The demonstration was based on the game Enemy Territory: Quake Wars from Splash Damage (See video.)

"This is a ray tracing demo," said Intel senior research scientist Bill Mark during the demonstration. "We took the content, the textures, and geometry, pulled it out of that game and put it into our ray tracing engine."

Mark described ray tracing technology as allowing "you to simulate the interaction of light with matter in a way that's accurate and makes it really easy to get effects like light and shadows."

"If you look at the water. That's done with only 10 lines of...code," he said. The demo was written in C++.

Mark said the same thing can be done on a standard multicore Intel processor but with Larrabee there is more parallelism--or the ability to do more things at the same time.

Intel is expressing Moore's Law anew as packing key technologies into fewer chips. New "Clarksfield," "Arrandale" and "Jasper Forest" processors, among others, will showcase this theme later this month at the Intel Developer Forum.

Intel Vice President Steve Smith discussed the highlights of the annual marquee Intel event that will kick off September 22 in San Francisco in a phone interview on Friday.

"Contrary to speculation that Moore's Law is slowing down or potentially dying, we're here to demonstrate that it's alive and well," Smith said. "Integration gives you a smaller, better, faster, more mobile compute platform," he said. Moore's Law, named after Intel co-founder Gordon Moore, states that the number of transistors that can be placed on an integrated circuit doubles roughly every two years.

This theme will be manifested in a number of new processors including the first mobile processor based on Intel's new Nehalem microarchitecture codenamed Clarksfield and even more highly integrated processors to follow dubbed Arrandale and Clarkdale as Intel moves to its next-generation 32-nanometer manufacturing process.

True to its rich heritage of codenames, IDF can, at times, slide into little more than a series of codename-riddled Power Point slides, with some names sounding frustratingly familiar such as Clarksfield and Clarkdale. But codenames, for better or worse, are part and parcel of IDF.

Intel codename decoder:

• Clarksfield: 45-nanometer Nehalem mobile processor integrating I/O
  
• Clarkdale: 32-nanometer Nehalem desktop chip integrating graphics with CPU
  
• Arrandale: 32-nanometer Nehalem mobile chip integrating graphics with CPU
  
• Moorestown: 32-nanometer system-on-chip Atom for smartphones
  
• Sodaville: 32-nanometer system-on-chip Atom for consumer
  
• Pine Trail: new Atom for Netbooks integrating graphics with CPU
  
• Jasper Forest: 45-nanometer, first Nehalem embedded chip for uses such as storage hardware
  
• Larrabee: Intel discrete graphics chip that will compete with Nvidia, AMD
  
• Westmere: 32-nanometer manufacturing process technology
  

Smith said that Intel's move to the next-generation "Westmere" 32-nanometer manufacturing process will drive even more integration next year. "We have completed development and certification of the 32-nanometer process, which means our factory is fully qualified to run the wafers. And we are actually running Westmere CPU wafers through the factory in support of our Q4 revenue production. Absolutely on track for that Q4 revenue production," he said, referring to commercial production of 32-nanometer processors.

In the more immediate future, Intel will roll out a new mobile processor based on its current 45-nanometer technology. "We just announced Lynnfield (the Core i5 and i7 chips for desktops), Clarksfield is the equivalent product for notebooks," Smith said. "Quad-core, 45-nanometer. Based on Nehalem technology but optimized with power management and integration of the PCI express I/O. Moving from a three-chip solution in the original Nehalem products to two chips--and that is our path going forward." I/O, or input-output, is silicon that enables a processor to talk, and shuttle data, to other parts of the system and peripheral components.

Speaking more specifically about Clarksfield integration, Smith said that "the key elements are integration of memory controller, integration of PCI Express 'gen' 2, power management." Intel will also be talking a lot about a feature called Turbo mode. "Turbo mode is extremely important. If you're not using all the cores, the cores that are not used are powered down. The cores that you are using can run at a faster clock rate with Turbo mode," Smith said.

Smith spoke about the next-generation Atom processor for Netbooks and Nettops, "Pine Trail," too. This chip will also ... Read more