14 DAY TRIAL // Intel's Larrabee is the codename for a so-called many-core architecture to shoulder a multitude of hardware products as well as some other implementations. The Larrabee technology will be included in some products with different number of cores to target different markets.
During yesterday's conference call, Paul Otellini was asked about the evolution of the Larrabee project. Nobody would have expected to squeeze some really important information regarding Intel's many-core implementation, yet Otellini gave a complete update on the project roadmap.
"There's two thrusts there. One is to significantly increase the performance of our integrated graphics line, our existing product line and you'll see a bump in that product line in 2008 that takes the performance up substantially, and then a couple more bumps over the next couple of years. We are on track to deliver the 10X performance improvement in integrated graphics by 2010 and that's in a fairly linear fashion from 2006 to 2010. Larrabee first silicon should be late this year in terms of samples and we'll start playing with it and sampling it to developers and I still think we are on track for a product in late '09, 2010 timeframe."
Larrabee has ceased being a simple prototype, but it is far from gaining universal support. It will be a long way until game and GPGPU creators master its intricate architecture and programming model, but this is the proof that Intel is decided to make Larabee a long-term investment.
It is assumed that, since the final product is scheduled to go public in late 2009, it will be built using the 45-nanometer technology, but this will conflict with the GPUs produced at Taiwan Semiconducors Manufacturing Co, as far as the companies' roadmaps show. The other scenario assumes that the chip will be produced at the 32-nanometer node, which will bring numerous functional advantages.
The time interval the chip will hit the market is another critical aspect, since it will directly compete with NVIDIA and AMD's brand-new DirectX 11 architectures, which is likely to generate an interesting and profitable fight on the GPGPU market.