Cell chips with everything

Feb 9th 2005
From The Economist Global Agenda

IBM, Sony and Toshiba have unveiled a new sort of microprocessor. Will the “Cell” chip cement Sony’s dominance of the computer-games market? And can it eventually challenge Intel’s dominance of chips that run PCs, as well as finding a home in a vast array of electronic devices?

“MOORE’S LAW is too slow for us,” declared Shinichi Okamoto, Sony’s chief technical officer, in 2002, in reference to the chip the company was developing to power its next-generation computer-game console, the PlayStation 3. The famous law says that the amount of computing power available at a given price doubles every 18 months. Even this exponential rate of growth was not thought rapid enough to satisfy the thirst for speed and realism demanded by today’s discerning gamers. The February 7th unveiling of the technical specifications for IBM’s “Cell” chip, developed in conjunction with Sony and another Japanese electronics giant, Toshiba, shows that the claims of Sony’s leading boffin may not be misplaced. The chip contains eight processing units that can work simultaneously on different tasks and will run at a speed of 4 gigahertz, at least ten times faster than the best chips available at the moment, according to IBM.

Described as a “supercomputer on a chip”, the features of the new device show that single microprocessors are giving way to a new, “multi-core” design. This allows chips to run more efficiently, using less power and generating less heat. IBM, Sun and Hewlett-Packard already produce high-end computers that use multi-core chips. But no competitor has yet produced a chip as cheap, powerful and versatile as the new Cell chip.

That the chip is initially intended for a computer game may strike some as underplaying the technological advance. But these games are huge moneyspinners for both Sony and its competitors, including Nintendo and Microsoft, whose Xbox is now powered by an Intel chip. Sony is the leader in the console market with the PlayStation 2, which has sold over 100m units worldwide since its introduction in 2000. In the year ending March 2004, sales of games accounted for around 10% of Sony’s revenues of $72 billion.

Some think that Sony’s healthy lead over Microsoft and other rivals could soon narrow, as the PlayStation 3 is set for launch in 2006, later than the next version of the Xbox. But the PlayStation is likely to remain an important part of the “digital home”, where a variety of electronic devices will be used to deliver data and digital home entertainment, such as films.

However, much more is expected of the new chip than just powering a new version of a video game. Toshiba plans to put the chip in a high-definition television. Eventually, mobile phones and other hand-held devices could feature it too. Most importantly, the chip poses a threat to Intel, the dominant figure in the personal-computer microprocessor market. Rather than relying on all-out processing speed, the new chip will use scale: the design will allow many chips to be stuck together to provide massive processing power.

Intel’s dominance in computer microprocessors is impressive. It is the world’s biggest chipmaker and its branded wares are an essential component in most PCs. Over 80% of the world’s PCs do indeed have “Intel inside”; its nearest rival, Advanced Micro Devices (AMD), can boast just over 15% of the market. And microprocessor sales are booming once again, growing by 11% last year as the computer business bounced back after the dotcom crash and the heavy retrenchment of technology budgets that followed. Intel made net profits of $7.5 billion in 2004 on revenues of $34 billion (up 13.5% on the year before).

But not all is well at the chip giant. Despite its dominance of the PC market, Microsoft is set to ditch Intel’s Pentium chip from its Xbox in favour of a processor from IBM (though not one as advanced as the Cell chip). And AMD is now perceived to have a technological edge over its much bigger rival. Some of Intel’s most important allies, such as Hewlett-Packard and Microsoft, have looked favourably on AMD’s Opteron chip. In addition, Intel suffered as a result of the vast sums and resources it expended on its long-overdue and ill-fated Itanium chip, developed in partnership with Hewlett-Packard. Like the Cell chip, this was intended to revolutionise the business but has ended up as a small-volume chip with only specialist applications. It is a lesson that cheerleaders for the Cell chip would do well to note.

Intel is trying to put things right. It announced its own “dual-core” chip to coincide with the launch of the Cell chip. And in January the company said that it would restructure its business and split into five units in an effort to improve its ability to operate in its different markets. But its attempts to expand its chip expertise into home electronics and mobile phones are fraught with difficulty, especially with the likes of IBM and Sony in the enemy camp. Many analysts think Intel should stick to maintaining its dominance of the PC market, though that too will prove difficult if the claims made for the Cell chip prove true.

Copyright © 2005 The Economist Newspaper and The Economist Group. All rights reserved.