Nanotech - The Circuits Blog

Any gamer worth his or her salt is quick to decry gaming on Intel graphics silicon. But wait. The platform is taking off, according to Intel.

Empire: Total War

(Credit: The Creative Assembly)

"So you want to know what's so compelling about making sure your game runs on Intel integrated graphics?" Aaron Davies, a senior marketing manager in the Intel Visual Computing Software Development group, asked in a video on the Intel Software Network Web page. "Here's your answer: Mercury Research showed that in 2008, for the first time, integrated graphics chipsets outsold discrete (graphics chips), and in 2013, we expect to see integrated graphics chipsets outsell discrete by three to one," Davies said.

Intel is the leading supplier of integrated graphics--which are integrated into its chipsets--while Nvidia and Advanced Micro Devices' ATI unit are leaders in the discrete (standalone) graphics chip market. Discrete chips are the most powerful engines for running games but Intel is the leading supplier, based on market share, of graphics silicon for laptops.

Davies said he wants to help developers "capture" where the mobile-game market is going to be in 2013.

"There are games actually targeting integrated graphics chipsets at this time," according to Davies. "We found through engaging with these Triple A (AAA) game studios that within a relatively short amount of time, they can identify graphics bottlenecks in their code and resolve that to have their games running on integrated graphics chipsets."

Davies mentioned a few of the beta members: Terminal Reality, which is slated to come out with Ghostbusters later this year; Gas Powered Games, which is building Demigod, and Empire Total War, which is put out by The Creative Assembly.

Intel is back, pitching its processors for gaming graphics.

The chipmaker will attempt to promote its silicon for sophisticated game effects at the upcoming Game Developers Conference in March, as it strives to make a case for quad-core processors in lieu of graphics chips from Nvidia and Advanced Micro Devices.

The pitch goes like this: "Learn how to easily add real-time 3D smoke, fog and other fluid simulations to your game without using up the GPU." That's according to an Intel Web page entitled Intel at Game Developers Conference. (The CPU is the central processing unit, or main brains of a computer; GPU stands for graphics processing unit.)

The session abstract goes on to say that the "source code to a fluid simulator optimized for multi-core CPUs...can easily be integrated by game developers into their engines to produce unique 3D effects."

Intel's argument raises the question, how should the CPU and GPU divvy up their tasks? In games, the CPU can handle things like physics and AI (artificial intelligence), and certain older games actually run some graphics on the CPU. Generally, however, the GPU is much more efficient (that is, faster) at handling most of the high-end effects that the gamer sees on the screen.

But there are exceptions. "Not all algorithms and processes map well to a GPU," said Jon Peddie, president of Jon Peddie Research. "You have to have a problem that is naturally parallel, and except for the rendering of, say, a water surface and subsurface and reflections, the wave motion equations will run just fine on a CPU," Peddie said.

Intel may also be seeking ways to make better use of its quad-core processors, according to Tom R. Halfhill, an analyst at the Microprocessor Report. But, he added: "I need to be convinced that a CPU can do those 3D effects better than a GPU can."

Then, there's also the Larrabee factor. Larrabee is an upcoming high-end graphics processor due late this year. "I'm sure some of it may also relate to Larrabee, which will include x86 cores, if or when it comes to market," said Jim McGregor, an analyst at In-Stat.

(This Mythbusters demonstration at an Nvidia conference is oversimplified and self-serving but it crystallizes the difference between CPUs and GPUs.)

In another GDC session, Intel is also pushing the CPU for physics and AI: "How can your game have more accurate physics, smarter AI, more particles, and/or a faster frame-rate? By threading your game's engine to take advantage of multi-core processors. Intel has built a threaded game engine and demo called 'Smoke' that shows one way of achieving this goal," the abstract states.

It continues: "This presentation examines the Smoke architecture and how it is designed to take advantage of all CPU cores available within a system. It does this by executing different functional and data blocks in parallel to utilize all available cores."

Intel won't stop there. It will also focus on the bane of many PC game developers: gaming on Intel integrated graphics silicon--a relatively low-performance platform that prohibits game titles from being displayed in all their glory at higher resolutions. The session will focus on "programming for scalable graphics applications" and cover "performance considerations when programming for integrated graphics in general with specific tips for Intel Integrated graphics."

When Apple converted to Intel in 2005 that was big. But 2008 Intel Atom converts make this look like a small-town baptism.

Overall, it was a good year for the Intel faithful despite the Wall Street financial crisis. Intel handily beat Advanced Micro Devices in the PC processor performance war. (Not coincidentally, AMD was forced to spin off its manufacturing operations to save itself.) But that really was last year's news since AMD had not been delivering competitive processors for almost two years.

iBook G3: Apple's conversion from IBM-Motorola to Intel pales against the conversion of PC makers to Intel's Atom

(Credit: CNET Networks)

The tectonic shift in 2008 came as one PC maker after another adopted Intel's new Atom processor. Count 'em: Acer, Asus, Dell, Hewlett-Packard, Lenovo, Toshiba--to mention only the largest vendors. (Atom shipments in the third quarter were strong and expected to hit between 10 and 20 million units this year.)

This wasn't one sole convert (like Apple), this was a Pentecostal conversion of biblical proportions. Almost overnight, the entire top tier of the PC industry got the Atom religion. In fact, it happened so quickly and so massively that companies like AMD and Qualcomm didn't know what hit them.

Wait a minute, Qualcomm seemed to say, we specialize in making chips for small devices, why is Intel running away with this market? (Even Intel was a bit surprised at the swiftness of Atom adoption in Netbooks.) And though AMD had helped pioneer the market by supplying its Geode processor for the progenitor of the Netbook, the One-Laptop-Per-Chip XO laptop, the Geode never came close to the commercial success (or performance) of the Atom.

AMD took notice, however, and said it plans to deliver a processor for the ultraportable market (an upscale Netbook or cheap notebook--however you want to look at it) at the Consumer Electronics Show.

And Nvidia followed suit. And seemed to be posing the same questions. Hey, if everyone's doing this, is this the Second Coming of the PC? Or, at least, a restructuring of the traditional price structure of the PC market? (The other question Nvidia is asking itself is whether it can bust the Intel bundling Juggernaut).

Oh, and we almost forgot Microsoft. Not initially enthusiastic about the Netbook market because of its XP-centric nature, Microsoft seems to have also gotten the Netbook religion with Windows 7 which will be ready for Netbooks from day one.

On Tuesday, an industry consortium ratified the OpenCL 1.0 specification, a standard that started as an Apple proposal but has gained many supporters, including graphics chip companies Nvidia and Advanced Micro Devices.

OpenCL, or Open Computing Language, is essentially an open industry standard for 3D graphics and computer audio and is meant to extend the capabilities of the graphics processing unit (GPU).

Not surprisingly, graphics chip companies have been quick to pick it up, including Nvidia and AMD's ATI graphics unit, which both made separate announcements Tuesday, along with the broader announcement from The Khronos Group consortium.

OpenCL has been developed on Nvidia GPUs and the company was one of the first to show working OpenCL. Nvidia is the graphics chip supplier for Apple's MacBooks.

Other computer graphics-related companies supporting OpenCL include Imagination Technologies and ARM. The complete list is long and includes 3DLABS, Broadcom, Electronic Arts, Freescale, IBM, Intel, Nokia, Samsung, and Texas Instruments.

Nvidia combines support for OpenCL with its CUDA language

(Credit: Nvidia)

The language itself--based on the C programming language--is designed for programming parallel computing across both GPUs and CPUs (central processing units). Apple proposed OpenCL in June for standardization work, targeting Mac OS 10.6 "Snow Leopard."

Apple's goal for Snow Leopard is to let any application tap into the prodigious parallel computing capabilities of GPUs, which integrate hundreds of computing cores.

"We are excited about the industry-wide support for OpenCL," said Bertrand Serlet, Apple's senior vice president of Software Engineering, in a statement. "Apple developed OpenCL so that any application in Snow Leopard, the next major version of Mac OS X, can harness an amazing amount of computing power previously available only to graphics applications."

AMD, which supplies both CPUs and GPUs, is well situated to exploit OpenCL. "The potential benefits of having applications run on both the CPU and GPU within a system are enormous," said Rick Bergman, senior vice president and general manager, Graphics Products Group, AMD, in a statement. "Unfortunately, up until now programmers could only choose proprietary programming languages that limited their ability to write vendor-neutral, cross-platform applications. With today's ratification of OpenCL 1.0, I'm happy to say those days are over."

Nvidia confirmed Friday that it has acquired RayScale, a small company that develops ray-tracing technology. Financial terms of the deal have not been disclosed.

Ray tracing has been mentioned frequently by Intel over the last six months. An Intel blog titled "Real Time Ray-Tracing: The End of Rasterization?" and later comments by Intel executives that the company is looking at doing ray tracing on its processors set the stage for debate on the viability of ray tracing in mainstream gaming.

PC graphics technology today uses rasterization. (A discussion of ray tracing vs. rasterization.)

Ray Tracing is a technique for rendering three-dimensional graphics with extremely complex light interactions, allowing the creation of transparent surfaces and shadows, for example, with stunning photorealistic results.

Ray tracing is a highly parallel process. And the GPU (graphics processing unit) provides high level of parallelism, according to Nvidia officials speaking at a conference on Thursday. The GPU has special function units that were desgined for doing graphics operations that are perfect for ray tracing, said Nvidia Chief Scientist David Kirk.

At the conference, Kirk and RayScale scientists discussed "GPU ray tracing." It's not clear how soon this technology would be used commercially by Nvidia.

RayScale, which provides interactive ray tracing and photo-realistic rendering solutions, says its technologies "dramatically increase the speed and realism at which graphics professionals can produce high quality three-dimensional computer graphics and photorealistic computer images."

RayScale is a product of the decade-long interactive ray-tracing research at the University of Utah, according to RayScale.

At the Intel Developer Forum in Shanghai in April, Senior Intel Vice President Patrick Gelsinger spelled out Intel's vision: essentially that ray tracing-based rendering technologies can be used in high-end gaming. "An intro of these capabilities into mainstream gaming we believe is possible in the future," Gelsinger said at that time.

Intel though has provided little proof beyond the running of some aspects of Quake IV (a relatively old game), for example, on an eight-core system.

Jen-Hsun Huang doesn't pull any punches. And Intel is a favorite punching bag these days.

I had a chance to sit down with the Nvidia CEO as he described his company's philosophy and what sounds like the first volleys of a long battle with Intel.

Nvidia CEO and founder Jen-Hsun Huang

(Credit: Nvidia)

A quick backdrop: Nvidia is the largest graphics chip company in the world, with quarterly revenue in the $1 billion range. Although Intel and Nvidia seem to exist in symbiotic bliss inside many PCs, this doesn't reflect the two companies' business models, which are in many respects far apart.

Intel is a chip manufacturer. Nvidia is not; it's a fabless company. Intel supplies the central-processing unit (CPU), a general-purpose processor. Nvidia supplies the graphics-processing unit (GPU), a special-purpose chip.

Huang is relentless in driving GPU performance--and fearless when challenging Intel. This is admirable, if anything. Even the world's largest PC makers treat Intel with great deference--publicly--because the chipmaker is so instrumental in supplying and defining the core electronics in their PCs (And partly due to the fact that they use Intel advertising dollars).

But Huang will tear into Intel when he thinks it's warranted. And Intel may have reason to be worried about the content of Huang's candor. Despite Intel's colossal size and and clout, Nvidia--not Intel--supplies the defining chip for the most savvy computer users: game enthusiasts. They depend on Nvidia graphics chips to deliver the spectacular visuals of games like Crysis.

And few people will deny that computing is becoming more visual. The GPU is essentially a parallel-computing engine that is extremely efficient at running visual (and scientific) software--that is, many of the popular graphics, video, and photo applications now running on PCs.

Nvidia is set to challenge Intel in the mobile Internet device space. It is getting set to make a big platform play in tiny, fit-in-your-pocket devices with its APX 2500. This "system on a chip" will house everything that comes on a PC circuit board today. Intel is targeting the same market with its Moorestown processor, due in late 2009 or 2010.

One important note: because Huang had made so many references to Intel over the last few months, particularly at the financial-analyst day in April, the interview revolved around this topic. In some cases, I asked pointed questions about Intel and posed hypothetical--i.e. devil's advocate--scenarios. In other cases, Huang volunteered statements about Intel.

Huang has spoken forthrightly in the last few months about Intel. The obvious question is why.

"People thought that I had lost some of my patience with Intel recently. The fact of the matter is that they're out spouting things that are just not true. And I was just correcting that," he said.

"Intel is a big, powerful company," I noted to Huang. "And there aren't many people like you in the industry, who are so blunt about Intel."

His reaction: "Because they are Intel. Because they are a monopoly. Because they are a market-dominant player. They ought to be held to a higher standard. They shouldn't be able to say that other peoples' businesses are going to die."

Huang, here, is referring to a statement by an Intel executive who recently said current graphics technology (sometimes referred to generally as rasterization) will be replaced by another kind of graphics technology (sometimes referred to as ray tracing), on which Intel is working.

Intel has also been dropping more hints about its upcoming high-end graphics chip, called Larrabee, with relatively few specifics. I asked Huang if he thought there was a reason so few details had been given.

"Larabee is a PowerPoint slide," Huang said. "I haven't met a product on my PowerPoint slide that I don't like. You know, they're floating Larrabee out there just to put a shadow over us, cast a cloud over us. They've already slipped it two years from the time they talked about. They would love to slip it another four years and leave a cloud over me."

"Just to play devil's advocate," I said, "Intel sees the success of the GPU. So it has to crank up its skunk works and develop a fast GPU too (Larrabee). Then Intel, being Intel, has to fill its factories and sell these things. Again, I'm playing devil's advocate here."

Huang's immediate reaction: "You and I have a deal. If you're going to write controversial stuff about what I say, can you write what you just said? Here's what I believe: I believe that the entire world believes that what Intel does is build a factory, stuff that people don't want to buy, and then shoves it down its customer's throats. Just like you said."

Huang also spoke about how the PC industry is shifting away from the CPU-centric vision.

"We would love it if people would buy more GPUs, but the fact of the matter is, we don't have Intel's budget to tell you to buy something you don't need. We're going to let the market decide for itself," he said.

"Selecting the right GPU for the right CPU--and having these two processors collaborate. We call it the optimized PC design," he continued. "Notice, we didn't call it 'increase your GPU' design. Notice we didn't go 'buy more quad cores.' It's not a market benefit message. The optimized PC asks what your work flow is. Take the work flow, and benchmark it on the machine. And decide for yourself."

Huang had a few points to make about changes in PC marketing.

"The whole idea that the PC industry is good, better, best, faster microprocessors, more memory--that psychology of the PC industry is so yesterday," he said. "Not a single person believes it. Sony doesn't believe it. Dell doesn't believe it. HP doesn't believe it. God knows Apple doesn't believe it. Nobody believes it anymore."

Huang elaborated, saying that at the other end of the computing spectrum is the minimalist PC, which Nvidia's APX 2500 system-on-a-chip addresses.

"There's a movement toward 'I want the most minimal of PCs': the ThinkPad (X300)," Huang said. "In the future, if it's not thin like a sheet of paper, it's just too much. There should be no electronics. There should be just one tiny chip. And this computer ought to cost nothing. The display should be the most expensive thing. It's not about the CPU. It's not about the GPU. It's about the computer on the chip."

But Intel, and its capacity to integrate more and more of the PC's function into its chipsets, is never far from his mind. Huang gave a number of examples of companies--as smaller and smaller chip geometries have allowed more and more transistors to be packed into a single chip--that disappeared because they were integrated out of existence. (Think sound chip and multimedia chip companies as just a few examples.)

"Make me...list one single example where Moore's Law is not your enemy today," he said. "At this very moment, the only one we know of is the GPU."

Every year, Huang said, "we're making chips that are twice as big as the (year) before that. And every single year, we deliver an experience that is twice as good as the year before. And every single year, people say, 'It's not good enough. I want more. I want more.'"

Throughout it all, Moore's Law is still Huang's friend, he said.

"Notice in the case of CPUs, people are saying, 'I don't need that many gigahertz,' or 'I don't need that many cores,'" Huang noted. "(CPU makers) are going down that path. And that's why it's possible now to build an Atom CPU. At that point, the technology becomes good enough."

Huang said he is not trying to wish Intel away. He is willing to co-exist. But he doesn't believe that Intel is able to do this. This probably is his biggest beef.

"There are going to be two important processors in the system," he said. "A microprocessor that is used for all kinds of complicated, unpredictable sequential code. And a parallel processor, called a GPU, that is really dedicated toward doing very parallel, very heavy-lifting mathematical operations."

Huang refocused his attention on Intel.

"Intel cannot share the world with someone else. They want the world to have one processor. They don't want the world to have two processors, even if it's good for them. (The Nvidia chip) just happens to be so famous, and just happens to be so popular, and happens to be so delightful that it just really makes them upset. That's an anti-innovation feeling. That's a monopolistic feeling, right? You can't share the world with somebody else."

His attitude borders on paranoid. But in Silicon Valley, the credo "only the paranoid survive," put forth by former Intel CEO Andrew Grove, is followed by many.

"People have been predicting the demise of our company for 10 years now," Huang said. "Intel has been in the graphics business for 10 years. They've been predicting our death for 10 years. They'll be predicting our death 10 years from now."

Nvidia's hostility toward Intel is on a high boil these days. In its latest dig against the central processing unit (CPU) and the company that makes the lion's share of CPUs, an Nvidia VP said in a private missive that the CPU is dead and it has "run out of steam."

Nvidia CEO Jen-Hsun Huang

(Credit: Rico Shen)

But wait. That's not what Nvidia really thinks. The message cited by the The Inquirer is "not a public statement," said Brian Burke, an Nvidia spokesperson. "The views in (Roy Tayler's) e-mail do not mirror the views of Nvidia." (The author of the message, Roy Tayler, is VP of content relations at Nvidia.)

But is the statement that far apart from Nvidia's public sentiment? "You need nothing beyond the most basic CPU," Burke said. Sounds like Nvidia thinks the CPU is, if not terminal, certainly fading.

(The CPU, or central processing unit, is the main processor in a PC. The GPU, or graphics processing unit, handles much of the visual content on a PC.)

This of course is news to Intel, the largest chip company in the world whose main business is making CPUs. "We believe that both a great CPU and great graphics are important in a PC. Any PC purchase--including the capability level of components inside it--is a decision that each user must make based on what they will be doing with that PC," said Intel spokesperson Dan Snyder.

To be sure, Nvidia and Intel have never gotten along famously. But the acrimony (mostly Nvidia's) started to build at Nvidia's fourth quarter conference call and carried over to the company's financial analyst day earlier this month, when CEO and co-founder Jen-Hsun Huang, alluding to comments from game developer Tim Sweeny, said "Intel is incapable of running modern games. Intel's integrated graphics just don't work."

But the crux of Nvidia's marketing message, vis-a-vis Intel, is focused on the graphics chip maker's perceived limitations of the CPU. In short, buy a high-end GPU, not a high-end CPU, and save money. During the earnings conference call, Huang cited the Gateway P series notebook as an example. One model has an Intel 1.6 GHz processor and a GeForce 8800 GPU. He said systems like this with a "higher-end GPU" and "lower-end CPU" are better optimized for today's users. "Relative to a notebook with a higher-end CPU and lower-end GPU, the Gateway FX is twice the performance and yet $200 lower cost."

The graphics processing unit (GPU) is in, the central processing unit (CPU) is out. That was one of the main themes running through the Nvidia fourth-quarter conference call earlier this week. Nvidia is the largest graphics chip supplier.

Gateway P series FX PC with Geforce 8800 GPU

(Credit: Nvidia, Gateway Computer)

During the call on Wednesday, Jen-Hsun Huang, President and CEO of Nvidia, repeated one thing often: GPUs are playing more of a central role in PCs, CPUs less so. "The CPU has become fast enough for the vast majority of (PC) users," he said. "PC enthusiasts, gamers, and design professionals have know this for some time." The GPU offers more horsepower for parallel processing, essential for today's visually rich environments, he said.

Huang cited the Gateway P series notebooks as an example. One model has an Intel 1.6 GHz processor and a GeForce 8800 GPU. He said systems like this with a "higher-end GPU" and "lower-end CPU" are better optimized for today's users. "Relative to a notebook with a higher-end CPU and lower-end GPU, the Gateway FX is twice the performance and yet $200 lower cost." In short, Huang was saying that users can save $200 by buying a system with a low-performance CPU and high-performance GPU--and get better performance to boot than the other way around.

Intel, of course, has other ideas. "We feel that the CPU is absolutely vital and you need a fast CPU and a fast GPU for the best experience. Take game AI (artificial intelligence) and physics for example, something that is consuming more and more CPU cycles," an Intel spokesperson said. "Also, the CPU is essential for intensive stuff like hi def video encode, 3D rendering," the spokesperson said.

Huang had a lot to say about physics too in the wake of Nvidia's purchase of Ageia Technologies this week (first announced on February 4th). Ageia's PhysX software is used with more than 140 PhysX-based games on the Sony Playstation 3, Microsoft XBOX 360, Nintendo Wii, and gaming PCs. (Game physics simulate the laws of physics in games.) "We're going to port the Ageia PhysX engine onto CUDA."

CUDA, a programming interface, has now shipped into 50 million GeForce 8 series processors and over the next several years will ship into a few hundred million more, Huang said. "Our expectation is that this will encourage users to buy a second GPU...and for the highest-end gamers, will encourage them to buy three GPUs." One GPU would be used for physics, while two for graphics (or vice-versa), Huang said. "Every single GPU that is CUDA enabled will be able to run the PhysX engine when it comes. In the end, it's just going to be a software download," Huang added.

But Nvidia's CEO returned to his overarching theme again and again. More Nvidia and less Intel. "Rebalance the system so that more GPU horsepower can be dedicated to the (user) experience." Nvidia even has a name for this strategy. The "optimized PC design approach." And Nvidia believes that more and more consumers are coming to know this, resulting in high growth. "The consumption of GPUs is increasing," Huang said, citing 80 percent year-to-year growth in Nvidia's discrete GPU business in the fourth quarter.

"I think I would say that [Huang's argument] has qualified merit. It's completely true that in some applications graphics, rather than CPU, is the limiting factor, and naturally Nvidia would be concerned with those applications most often," said Dean McCarron, founder and Principal of Mercury Research. But Intel and AMD are not standing still. "As far as rebalancing, it's pretty clear the CPU suppliers are actively re-partitioning their products, and graphics capabilities are perhaps the highest priority here. If you look at AMD and Fusion, or Intel and its Nehalem CPUs, both suppliers clearly see advantages to repartioning the PC around graphics -- in this case, moving graphics onto the CPU."

Nvidia's execution is not flawless. It is not competitive in the business segment and at the lower end of desktop and notebook lineups. Large computer segments unto themselves. Here both AMD-ATI graphics and Intel integrated graphics dominate. AMD-ATI is also competitive in the mid-range to high-end.

In related news, Nvidia's shares fell Thursday due to lower gross margins. On Wednesday, the company said that for the first time in 13 quarters non-GAAP gross margins did not increase quarter to quarter. Gross margin shrank to 45.9 percent in the fourth quarter from 46.4 percent in the previous period. In the fourth quarter, the company posted a 58 percent jump in fiscal fourth-quarter net income.

On another front, Nvidia CFO, Marvin D. Burkett, said no new process technology will be needed for the 8800 processors and they will continue to be made on a 90-nanometer process.