Today during the NVIDIA GPU Technology Conference (GTC2010), CEO Jen-Hsun Huang revealed the codenames of the next two generations of GPU parts: Kepler in 2011 and Maxwell in 2013.
Kepler, built on a 28nm process, will introduce virtual memory and pre-emption, and is expected to bring 3-4x performance per watt over Fermi. Maxwell, at least according to the bar graphs we saw, look to double Kepler’s performance and provide “16 gigaflops per watt.”



Articles RSS