NVIDIA CEO Jen-Hsun Huang showed off a CUDA GPU roadmap at his GPU Technology Conference in San Jose. The roadmap gives an indication of the expected double-precision gigaFLOPS per Watt of the company's next major GPU releases, with Huang claiming that he chose this metric because power is the primary constraint in parallel computing.
The roadmap reveals Kepler will be NVIDIA's next major GPU architecture, this 28nm chip is expected to be released in 2011 and promises to be three times more efficient than Fermi, while the 22nm Maxwell promises to deliver eight times the double-precision FLOPS per Watt of Fermi in 2013.
In between these major architectures, NVIDIA plans to launch a "mid-life kicker" or an enhancement to existing architectures. For instance, Fermi will receive a refresh sometime next year, and between now and Maxwell, NVIDIA also plans to introduce several GPGPU computing focused features into its architectures, including pre-emption.
In between these major architectural generations, which are planned to track with the introduction of new process technologies, Huang said Nvidia will continue to produce a "mid-life kicker" or an enhancement to its existing microarchitecture. The next such update will be a refresh of the Fermi-generation GPUs, coming next year.
As before, the firm aims to do a top-to-bottom refresh of its GPU lineup, whenever it introduces a new generation of GPUs, within about three months. Huang admitted the Fermi generation has taken longer than he'd like because of some challenges in producing the first chips.