According to NVIDIA co-founder Chris Malachowksky, the next generation Kepler GPU will ship by end of this year. The Kepler GPU is expected to be manufactured on TSMC's 28nm process. Compared to the Fermi in terms of raw computing power, Kepler is expected to push GPU performance to 4-6 DP GFLOPS per watt; Maxwell, targeted for 2013, will push it up to 14-16 DP GFLOPS per watt. Under Fermi, NVIDIA GPU performance reached 1.5 double-precision gigaGLOPS (DP GFLOPS) per watt, according to the company. The release of CUDA has made it possible for many to develop applications that can take advantage of GPGPU (general purpose processing on GPU) abilities to autonomously process data without the assistance of the CPU.
The shipping of a GPU usually does not mean the video card hitting the retail shelves; hence, consumers would have to wait a while for the Kepler-based graphic cards. Compounding this issue is the alleged low yields of TSMC's 28nm process. As a result of such possible impediments, consumers may have to fork out more dollars for these cards.
Source: ( Xbitlabs )