The new multimedia processor, touted as a "supercomputer on a chip," is well on the way to completion, IBM says. The chip could end up inside the PlayStation 3 game console, and elements of its design are expected in future server chips from IBM.
Cell has nearly "taped out"--an industry term meaning that the chip's pen and paper design and layout have been completed. Soon these will be handed over to engineers in manufacturing, who will craft samples. Meanwhile, the engineers have been testing various elements of the processor, both separately and together, before the manufacturing unit connects them inside actual Cell chips.
At this rate, commercial production of Cell could come as soon as the end of 2004.
While details remain vague, Cell will differ from existing microprocessors in that it will have multiple personalities. The chip will not only perform the heavy computational tasks required for graphics, but it also will contain the circuitry to handle high-bandwidth communications and to run multiple devices, sources say.
Ultimately, Cell will provide a "much more interactive way of delivering content, including advertising, sports and entertainment such as video," to a wide range of Internet-ready devices, said Jim Kahle, director of broadband processor technology and a research fellow at IBM.
This would be possible because a single chip will contain multiple processing cores, or cells--an esoteric design concept that's rapidly emerging from the labs, sources said. Communications features expected to be in the chips might also allow devices to form powerful, peer-to-peer networks, some analysts believe.
"It's sort of like having a group of handymen who are able to raise the roof (on a building) or do plumbing if it's needed," said Richard Doherty, analyst with Envisioneering Group, a research firm.
More than games
So far, the chip triumvirate of IBM, Sony and Toshiba, which pledged $400 million to the project and sent engineers to a joint development center in Austin, Texas, has been short on details of how Cell could benefit each company.
The processor has always been associated with Sony's PlayStation 3 and peer-to-peer computing, but will do more than allow players to battle opposing characters in multiplayer Internet games, Kahle said.
But Cell will go "beyond gaming to just entertainment in general," Kahle said.
Shin'ichi Okamoto, chief technical officer for Sony Computer Entertainment, seemed to confirm Cell's presence in the next version of the PlayStation during aearlier this year at the Game Developers Conference when he mentioned that Sony and IBM were working on a processor called "cell" that would help Sony achieve its goal of boosting performance of the next PlayStation a thousandfold.
"We are working for the third-generation (PlayStation) with this very aggressive and crazy goal," he said. "Moore's Law is too slow for us."
From his own analysis, Doherty believes Cell will create a new extensible computing platform. A set-top box containing a Cell chip could, for example, combine to share processing power with a Cell-powered high-definition television to render the graphics of an animated movie.
"It's like a beehive--cell components can also be ganged together," he said.
This ability to change rapidly between states will make devices more flexible, but also give the living room a big boost in computing power when devices interact, making for much livelier games, movies and other entertainment-related experiences. That spells opportunities for Sony and its PlayStation line.
For IBM, Cell represents a technology showcase. The new chip will not only illustrate IBM's design prowess, but it will also display the company's manufacturing expertise. To mint Cell, IBM will use its bag of chipmaking tricks, including silicon-on-insulator (SOI) processes and low capacitance dielectrics.
Analysts say it's not as clear what Toshiba will get from Cell. The company could also use Cell to create new consumer devices such as high-definition televisions. Or, it could use Cell in its components business; Toshiba sells a wide range of components for set-top boxes and other consumer electronics products.
What's behind Cell?
While the processor's design is still under wraps, the companies say Cell's capabilities will allow it to deliver one trillion calculations per second (teraflop) or more of floating-point calculations. It will have the ability to do north of 1 trillion mathematical calculations per second, roughly 100 times more than a single Pentium 4 chip running at 2.5GHz.
Cell will likely use between four and 16 general-purpose processor cores per chip. A game console might use a chip with 16 cores, while a less complicated device like a set-top box would have a processor with fewer, said Peter Glaskowsky, editor in chief of influential industry newsletter Microprocessor Report. Some of these cores might perform computational functions, while others could control audio or graphics.
But not everyone thinks this approach is groundbreaking, given that some processors already use inter-chip multiprocessing. "I just don't see that Cell is revolutionary, except in its marketing impact," Glaskowsky said
Indeed, a host of other chipmakers produce chips--known asprocessors--that use multiple processor cores to power systems ranging from networking equipment and automobile electronics devices to cellular phones.
IBM entered the dual-processor core market with the, the first server processor to use a multiple-core design. Power4 pairs two 64-bit PowerPC cores on the same processor, linked by a high-speed communications pathway.
But efforts to create similar, more generic multiple-core processors--including MAJC (pronounced "magic"), a very similar effort by Sun Microsystems--have missed their intended mark.
The firstSoftware--the road ahead chip was originally slated for multimedia processing, a job similar to Cell's. But instead of selling the chip to set-top box and game machine manufacturers, Sun repositioned the dual 500MHz MAJC 5200 chip as a high-end graphics processor for workstations.
While Cell's hardware design might be difficult, it's creating software for the chip that will be the trickiest part of establishing it in the market.
"It's going to take an enormous amount of software development," Doherty said. "We believe the chip architecture is going to be on time and ahead of the software wizardry that is going to really make it get up and dance."
Furthermore, creating an operating system and set of applications that can take advantage of the Cell's multiprocessing and peer-to-peer computing capabilities will be the key to determining if Cell will be successful, he said.
Knowing this, the three chip partners have so far set a goal of crafting Cell as a system, creating operating system and application software alongside Cell hardware.
Cell's designers are engineering the chip to work with a wide range of operating systems, including Linux.
But the chip triumvirate is also developing a purpose-built Cell operating system and applications, which Cell's developers will use to test the chip's various features, such as its multimedia processing capabilities. They are also likely to form the basis of a Cell software development kit and also the Cell OS and applications for end-devices, such as game systems, sources said.
Still work to do
While much of the work on Cell is complete, there's still a lot left to do.
Together, the hardware and software teams will continue testing the chip's inner workings.
The last stage of development work, which still lies ahead, includes completing circuit layout and then eventually testing actual sample chips.
IBM is expected to begin manufacturing Cell as soon as 2004 or possibly early 2005. But as with many other details about the chip, Kahle will confirm only that the Cell project is on track to meet it's a 2005 introduction, which was set forth at its initial announcement.
The rest of the chip's schedule is a secret, at least for now, he said.