There are some rules to launching a successful line of microprocessors. These are just a few of them:
The launch should be international in scope, preferably tied in to a large, well-known IT industry event such as the Consumer Electronics Show (CES). The chip’s debut should be backed up by
pledges of support by industry OEMs, who announce plans to include it in their corporate products. Most important, you have to know your market — the chip should be aimed at a specific kind of hardware platform, customer segment or computing task. That IBM and Toshiba are breaking all these rules with a possibly breakthrough chip design called the Cell is the first sign they may win the game.
Although the Cell hadn’t officially launched as I wrote these words, so many details had surfaced prior to an event scheduled by IBM and Toshiba that you knew the chip had already entered the big leagues. When a chip is greeted with the same mix of optimism and speculation that surrounded the Pentium III, somebody’s got to be doing something right.
Several published reports suggest the Cell could reach a peak performance of 256 billion mathematical operations per second. We’re talking about capabilities that could go anywhere from supercomputing to cell phones.
With a potential reach that broad, the Cell could end up anywhere — or nowhere. Its chief asset may not be in its architecture, which seems to have impressed several processor analysts, but the market prowess of its creators. IBM knows how to create high-performance systems that could benefit from the Cell’s ability to virtualize applications. Now that Lenovo is taking over IBM’s PC business, Toshiba may provide the best link to the desktop and mobile computing worlds.
The Cell may be IBM and Toshiba’s rebuke to Intel and AMD’s product roadmap, which will engineer performance enhancements by splitting up tasks to be handled by various parts of a dual-core processor. The Cell is composed of several self-contained processing units that handle tasks independently. This is not as similar as it sounds, because the “”software cells”” can be accessed from distributed pieces of a network when additional resources are required. A server in one part of the data centre, for example, could draw upon a software cell residing in the chip on another server during a peak period. This could appeal not only to companies such as IBM which are pushing on-demand strategies but even longtime Intel partners such as Oracle, which is aggressively promoting grid computing in the enterprise.
Intel has faced a number of setbacks over the past year, but until recently its biggest threat seemed to come from AMD, which has capitalized on its rival’s missed deadlines to introduce more features into its own line of chips. With the Cell, however, IBM and Toshiba are making their start for the consumer processor market.
The video game console market may provide the perfect playground for chip vendors to prove their products can handle the heavy graphics, streaming and workload demands of future enterprise applications.
The real test will come if the Cell reaches the corporate desktop. For years, Intel and AMD have been arguing that processor speed is no longer a valuable benchmark. IBM and Toshiba seem prepared to take them at their word.