Gamers Will Lag Then Lead Multi-Core Switchover

On Thursday, Microsoft confirmed more than 25 titles to launch with its next-generation console, with a further 160 in development. Games planned for the Xbox 360 include flagship title Perfect Dark Zero, a prequel to the Rare game Perfect Dark; Kameo, a family-orientated game; Project Gotham Racing 3; Dead or Alive 4; Quake 4; FIFA 2006; and Ghost Recon 3. Microsoft was more than happy to discuss the power of the machine. Graphics are controlled by a 500MHz ATI graphics chip, and the console is powered by three 2GHz cores.

Most PC games today are not well suited for the dual-core processors that are now on the market -- and will next year become mainstream. There's a lot of history behind this fact. For the rest of this year, both Intel (Extreme Edition) and AMD (FX 57) are saying that best gaming performance will come from the company's fastest uniprocessors. Both companies plan follow-on uniprocessors into 2006, as well.

This uniprocessor-for-gaming argument suggests that a processor split is occuring, since the mainstream is going dual core by next year. I think the future will be the opposite.

By the end of next year's holiday season, game developers will have dozens of game titles out for the next-generation, multi-core consoles from Microsoft and Sony -- or they will be left behind in a dying legacy console market. These developers are now rewriting their game engines to support multiple processors. The smart ones are already planning for more than two processors (e.g., dual core PCs) since the xBox 360 and PSP 3 will have three processors. Moreover, the 2007-2008 products from AMD and Intel are likely to have four cores.

The multi-core games will first intro on the consoles because that is where the marketing dollars will be. But the standalone and online PC gaming markets are attractive enough that I expect many titles will be ported to dual-core PCs by the second half of 2006. This new content, in turn, will fuel dual-core purchases by PC gamers. Looking forward, more processors will allow more game threads to be running at the same time, allowing for real-time physics, motion, and rendering. All we need are the games to be multi-threaded to take advantage of multiple processors cores. And when the games are multi-threaded, even a much faster uniprocessor won't be able to keep up to a multi-processor.

That's the way I see the game ending.