Larrabee and Gesher could walk hand in hand under a beautiful sunset

Jan 18, 2008 09:59 GMT  ·  By

Intel is usually not much of a talker, and that is why it surprised everybody when Paul Otellini spilled the beans over the company's plans for 2008. Although the conference was focused on how poor Intel feels because it had some slightly lower-than-expected earnings during the Q4, Otellini began talking about one of Intel's top priorities, the many-core Larrabee project.

"Larabee first silicon should be late this year in terms of samples and we'll start playing with it and sampling it to developers," said Otellini. "I still think we are on track for a product in late 2009, 2010 timeframe."

Larrabee is Intel's codename for a many core architecture that will get implemented into a multitude of hardware products, and will sustain Intel's integrated graphics division. It is part of the Terascale initiative, that uses multiple cores for enhanced computing performance.

However, Larrabee seems to be running late and it is likely to arrive just in time to keep warm company to Intel's 32-nanometer processor, codenamed "Gesher". It will be a good choice of time, as both Larrabee and Gesher will be multiple-core processors, with the cores situated on a high-bandwidth ring bus. Larrabee will feature 16 to 24 small cores, while Gesher will come with four to eight massive cores.

If the estimations are right, and Larrabee won't show up until 2010 (although Otellini previously stated that it will arrive in 2009), this will take out the stress off the GPU makers' shoulders. They will have enough time to think of a multi-core strategy to boost exception handling and process switching capabilities.

Sure, it will take some time for Larrabee to gain universal support, since GPGPU creators and game developers might take some time to understand the intimate workings of its architecture. Intel will have to force the game developers' hands to build their gaming engines with respect to the new architecture, and Havok might kick in just in time.

Another issue that raises questions is how would the gaming world look like until 2010. Currently, more and more A-list games are dedicated to gaming platforms (PlayStation, Xbox and Wii), rather than to the PC sector. There is a slight chance for Intel to design an excellent PC-oriented hardware platform only to find out that nobody's home when it gets released.