updated 04:55 pm EDT, Fri August 5, 2011
Clarifies when commercial introduction will begin
Last week rumors flew about NVIDIA delaying the launch of its next generation Kepler GPUs from late 2011 until some time in 2012. Then, Chris Malachowsky, senior vice president of research and co-founder of Nvidia, reaffirmed that the Keplers would indeed be shipping by year's end. NVIDIA has now backtracked and claims that, while it's on target for receiving the silicon to build the GPUs, actual production of commercial units won't start until 2012.
The confusion appears to have arisen over semantic differences between what is meant by production of prototypes and other pre-release units and commercial production.
Kepler, based on a 28nm assembly process, is expected to be about three times more powerful and efficient than NVIDIA's current Fermi architecture, generating five gigaflops per watt versus Fermi's 1.5 gigaflops per watt.
NVIDIA names its architectures after notable scientists. Fermi, the current GPU product set, is named after famed nuclear scientist Enrico Fermi. The next gen Kepler is named for German mathematician and astronomer Johannes Kepler, best known for postulating the laws of planetary motion. The follow on architecture, Maxwell, is now expected out in 2014. It bears the name of James Clerk Maxwell, the Scottish physicist and mathematician credited with developing modern electromagnetic theory. [via Xbit labs]