In 2012, we’re likely to see even more of these systems focus on GPU computing. Nvidia continues to push its Tesla processors into more systems and says 35 of the top 500 computers in the world already use its processors, typically in a combination of traditional processors and GPU cores. Next year, Nvidia plans an upgrade to its basic architecture, moving from its 40nm Fermi design to a 28nm design known as Kepler. It is almost certain this will be part of future Tesla products, as well. I expect we’ll know more at the company’s GPU computing show next spring. In the meantime, the company has open-sourced its CUDA development platform for GPU computing.
Read the Full Story.