|
DIGITAL ENTERTAINMENT
By Dean Takahashi
The accelerator
The PlayStation 3 could conquer the home-entertainment and computing markets--if the chip inside it can deliver 1,000-fold processing improvements.
Sony's vision for home entertainment may not look much different from the aspirations of other consumer-electronics companies. Like Sony, these companies hope to build a machine for the living room, capable of computing, communications, and entertainment functions. But Sony hopes to differentiate its machine--in this case, the PlayStation 3--by equipping it with a chip of unprecedented computing power, one that would make it as much as 1,000 times more powerful than the PlayStation 2.
The soul of Sony's new machine is a cell-computing chip. These chips enable a distributed style of computing (known as cell computing) that performs computing tasks in much the same way a cell phone network routes calls from base station to base station. Due for release in 2005, the PlayStation 3 will thus be able to use its broadband Internet connection to reach across the Internet and draw additional computing power from idle processors. And if still more horsepower is needed, the PlayStation 3 can use a home network to enlist support from other available machines to tackle big computing jobs. Pieces of a computing task--for example, creating realistic 3D graphics that simulate entire worlds--will be distributed among available processors to harness their combined power.
Buoyed by so much processing power, consumers will be able to interact with these worlds without worrying about hackers, viruses, or lost connections. Instead of using a mouse or game controller, players might wave their hands in front of a Web cam, showing what they want to do through gestures. They might play games without ever putting a disc into the console machine, downloading games from the Internet instead. They could tap into vast networks of movies and music, or they could record shows on the PlayStation 3 hard drive, which, by 2005, might hold 12,800 hours of music or 2,000 hours of video. And, starting with buying games from Sony, consumers will also be able to use the PlayStation 3 to engage in all sorts of e-commerce, through either a Sony ISP or a potential ally like AOL Time Warner.
Sony's plan to build a box that could be the nexus of home entertainment was revealed in a speech by Shinichi Okamoto, senior vice president of research and development at Sony's game division, at the Game Developers Conference in March. Mr. Okamoto said that Sony's next box will make good on the unfulfilled promise of the PlayStation 2--that the PlayStation 3 will be a broadband-enabled computing machine. As such, it will compete not only with game consoles from Nintendo and Microsoft, but also with PCs from the likes of Dell Computer and Hewlett-Packard, and with TV set-top boxes from Motorola and Philips.
It's a grand vision, and it won't be easy to pull off. "The notion of a game box becoming a universal 'everything box' is architecturally very difficult," says Mike Ramsay, CEO of the digital video recorder pioneer TiVo. "The demands for processing that gamers have are too high. They can't be interrupted by an email message or have a game slow down while they're recording a TV show."
Faced with such a challenge, Sony is not going it alone. The consumer-electronics giant has formed an unlikely alliance to design the needed cell-computing chip and to perfect its manufacturing process. The company's game division, Sony Computer Entertainment, headed by PlayStation business creator Ken Kutaragi, is partnering with IBM and Toshiba to develop the PlayStation 3's cell-computing chip.
Technical concerns aside, Sony faces other obstacles. The company's plan contains no mention of how it will handle Microsoft's software applications, which are widely used for home computing. Also, neither broadband subscriptions nor the cell-computing chips are likely to become ubiquitous in just a few years--and ubiquity of these two things is critical to making this vision a reality. Still, the network effect applies here: more processors acting together equals more computing power. Sony is sending out the message: "Match what we're doing by 2005, or we're going to race ahead of you," says Richard Doherty, an analyst at the Envisioneering Group, a market research firm. "The PlayStation 3 is clearly going to be a replacement for your PC."
HOME INVASION
The battle lines are already forming, and it may well become a war of competing chips. For instance, by partnering with IBM and Toshiba, Sony suddenly finds itself competing with Intel. Craig Barrett, CEO of Intel, notes that his company has so much capital (more than $11 billion in cash) that, even with a $5.5 billion capital-spending budget, it doesn't need partners to make chips for its ally, Microsoft. "Joint ventures never work," he grouses, referring to the alliances of rival chip makers. "It's like tying three legs together to try to win a race."
Microsoft, Sony's main competition in this field, has placed two bets. It continues to work with Intel on its eHome project, which will enable a PC to communicate wirelessly with the TV or stereo. It has also allied with chip makers Intel, Nvidia, and Taiwan Semiconductor Manufacturing Company in the manufacture of chips for the Xbox, Microsoft's game console that sports a hard drive and a broadband Internet connection. Either the PC or the Xbox, a new version of which will be ready when Sony launches the PlayStation 3, will provide the hardware for Microsoft's MSN subscribers to receive all sorts of broadband services over the Web and to engage in e-commerce.
Chip startups are forming to arm the existing camps or to create their own skirmish lines. Other consumer-electronics companies, from TiVo to Scientific-Atlanta, will tap makers of custom chips like Broadcom and TerraLogic, or makers of media-processor chips like Equator Technologies, LSI Logic, and TriMedia Technologies, to make sure that their machines hold their own.
Sony's vision for the future is plausible and frightening to its rivals. Mr. Okamoto says the "third-generation platform," a temporary moniker for the PlayStation 3, will have 1,000 times the processing power of the PlayStation 2. That may be bluster; Sony's PlayStation 2 didn't quite have the horsepower that Sony originally claimed it would. But in 2000, Sony managed to deliver a PlayStation 2 that was several hundred times faster in some respects than the original 1993 PlayStation.
The performance the PlayStation 3 promises to deliver is far beyond the progress almost guaranteed by the chip- manufacturing advances codified by Moore's law. "If Sony gets 12 to 36 months ahead of other companies on Moore's law, it could be very threatening" to chip makers, Mr. Doherty says.
Yet as much as anyone tries to race ahead of Moore's law, few succeed in pulling that far ahead of other major chip makers, says Peter Glaskowsky, an analyst at In-Stat/MDR, a market research firm. Faced by a rival's advance, the best chip companies, like AMD and Intel, hold nothing back to improve their products. Sony, primarily a consumer-electronics company, will do well not to enter such a contest.
SILICON VOLLEY
Still, Sony is doing a couple of things to hit this 1,000-fold processing-power improvement and pack the advances of 20 years of Moore's law into just 5. The architecture of the cell-computing chip, which promises huge performance improvements, is one piece. To this end, the company is trying to develop a second partnership with IBM and Toshiba, this one to devise a custom manufacturing process for its PlayStation 3 chips. In East Fishkill, New York, hundreds of IBM, Sony, and Toshiba engineers are working to tailor IBM's silicon-on-insulator process to the new chip design. This process for making chips allows transistors, the basic building blocks of circuitry, to be packed extremely densely. Mr. Glaskowsky says this type of process is going to be required in the next few years because it's the only way to pack 500 million or more transistors onto the chip. (The PlayStation 2's Emotion Engine microprocessor has 13 million transistors.)
"This means that Sony will be able to design its chips to take advantage of a manufacturing process that doesn't yet exist," says Bijan Davari, vice president for technology and emerging products at IBM. "By combining improvements in chip architecture, software, circuit design, and manufacturing, this is how we move toward a thousand times current performance."
Sony's opening gambit in the next-generation chess game will have repercussions, effectively accelerating the plans of rivals to launch competing game consoles. This has happened before. In 1999, when Sony announced the huge performance leap of the PlayStation 2, Microsoft reacted by conceiving the Xbox. Rick Thompson, a vice president who would later manage Microsoft's Xbox project, told Bill Gates at a strategic retreat that an alliance between Sony, AOL Time Warner, and AT&T could create a game console that would be able to surf the Web and be given away for free at the local supermarket. Part of that alliance has been formed. Sony and AOL Time Warner have partnered to provide Internet connectivity and instant-messaging capability for the PlayStation 2--an alliance that might extend to the PlayStation 3.
When it developed the PlayStation 2, Sony allied itself with Toshiba to design and manufacture its chips. But it also spent $1.2 billion to build its own chip factory in Japan to manufacture the PlayStation 2's microprocessor and graphics chip. The strategy backfired when hiccups at the plant forced Sony to cut back on the number of machines it had planned for its launch. Microsoft, seeking to catch up quickly with the Xbox, used off-the-shelf PC microprocessors from Intel and a slightly customized graphics chip from Nvidia, which uses TSMC to make its chips. By turning to these third-party chip suppliers instead of building its own chip, Microsoft was able to leapfrog the performance of the Sony machine.
As it began creating the PlayStation 3 platform, Sony toyed with different strategies. It put 16 PlayStation 2 microprocessors in one machine--all working in parallel--but found that the performance of this so-called GS Cube wasn't good enough to provide the processing power the company wanted, Mr. Okamoto says. So Sony formed the alliance with IBM and Toshiba to focus on cell computing.
This time, Sony's decision to use the custom silicon-on-insulator process from IBM could pressure Microsoft into spending a lot more money on the next Xbox. Along with the amount it will spend in its alliance with IBM and Toshiba, Sony is planning to spend billions more on designing and building the PlayStation 3 console, not to mention the advertising, marketing, and expensive broadband connections. Sony had $5.1 billion in cash at the end of fiscal 2002 (ended March), enough to finance its ambitious plans. Its game division accounted for 62 percent of the $1 billion in operating income that Sony reported, thanks to strong sales of the PlayStation 2.
Sony hopes its cell-computing chips will be useful in other kinds of devices, from camcorders to TV sets. That could result in such high production volumes that overall chip costs could fall much more quickly than if the new chips were used only in the PlayStation 3.
Meanwhile, the Xbox will probably not be lucky enough to get a chip designed especially for it, nor is there an array of Microsoft consumer devices awaiting such a chip. Intel can't yet match IBM's silicon-on-insulator manufacturing process, and it is not likely to tailor its manufacturing processes for chips to be used in a future version of the Xbox. Indeed, Intel's strategy is more generic; it doesn't customize manufacturing processes to the design of specific microprocessors. And it typically doesn't shoot for 1,000-fold performance increases with each new generation. By contrast, Nvidia and other Xbox chip suppliers could further customize their graphics chip to TSMC's manufacturing process, but only if they got funding from Microsoft for such a project. So far, Microsoft is expected to lose hundreds of millions of dollars on the first iteration of the Xbox. Investing more money in chip production and design would only exacerbate those losses. But the company might have to take this hit just to stay in the battle with Sony.
DELAYSTATION
There are obstacles to Sony's plan to own the living room. The problem is that a game machine has to marshall all the resources of its computing and storage functions to deliver the ultimate gaming experience to hard-core gamers. That doesn't leave room for simultaneous processing tasks like video recording and email, which would be required of an all-in-one box.
"We believe the device that can be built to handle all sorts of simultaneous processing tasks, with software controlling what is happening in the background, is the PC," says Michael Toutonghi, vice president in charge of Microsoft's eHome division. "The PC already does that well, and we're going to improve it in the years to come." Microsoft will also add improvements like simplified wireless networking that can enable the PC to control lots of living room devices.
Multiple processors and storage could alleviate the performance problem that would otherwise be created by, say, recording a TV show or receiving email. But that would add extra cost to the machine. Custom-chip makers, like Broadcom, say they will always be able to beat general-purpose chips, like the cell-computing chips envisioned by Sony or general-purpose PC microprocessors, by customizing their chips to specific tasks, like decoding video or sound. "Right now, the cheapest way to do something is to have this mix of custom chips and low-cost processors, not having all of the functions performed by some kind of Pentium," says Mr. Ramsay of TiVo.
In the end, Sony may be reaching too far as it tries to go beyond the game console. "I think the flaw in Sony's thinking is they are trying to anticipate computing performance too far ahead," says Dave Orton, president and CEO of the graphics-chip maker ATI Technologies. "The rivals will simply use what is available as they need it, and even with this reliance on general-purpose processors, they may move ahead of Sony's custom-processing aims."
Adding multiple functions to a game box would mean weighing it down with expensive features on an already expensive graphics-laden box. It's also a recipe for complexity, which has been a hindrance for the PC as it tries to move into the living room.
Sony might argue that there will be plenty of leftover processing power and storage in a machine that will make its debut in 2005. But to beat Sony in this scenario, a competitor will have to dedicate its chip, hard drive, and Internet connection to gaming. That would make Sony vulnerable to losing its edge in gaming, and that isn't a risk that Sony is likely to take. So, for all of its talk of creating a universal box, Sony may have to live with the idea of a legion of living room devices. Indeed, in addition to the PlayStation 3, Sony is working on advanced set-top boxes, as well as better PCs.
"I think it's much more likely that we'll see a proliferation of devices for consumers," says Marc Andreessen, chairman and cofounder of Loudcloud. "Just look at the evolution of the kitchen. If everything converged into one appliance, we'd have a machine that did everything in the kitchen, instead of separate dishwashers and stoves and refrigerators."
That's the best updates I've seen on the PS3. It sounds so cool That piece of shit Xbox2 is going down in flames before it even on the market.
  |
 |
Hybrid Mode
