Polaris, Vega, Ellesmere and Baffin here is the code names of future GPU AMD

The commercial debut of the next-generation video cards AMD is still at least several months away, but already some rumors emerge on those who might be the technical characteristics. AMD will present in the coming months 3 distinct new generation GPU, linked like this:

Vega 10: GPU is the higher end among those that AMD will make available on the market, intended to take the place currently occupied by the proposals of the Fiji family. Debut, according to the updated roadmap of AMD, at the beginning of 2017:
Polaris 10 is the most powerful GPU in the Polaris family, destined to be used in high-end cards but not top of the line;
Polaris 11: is the proposed smaller, the first that AMD showed in operation in December 2015 and that presumably will be used in mid-range video cards as well as mobile systems.
The naming scheme used by AMD has no performance references: the numbers 10 and 11 shall apply only to the design period, with the GPU Polaris 10 which was then developed internally before 11 Polaris solution.

Latest AMD drivers for linux systems also contain some information related to the next generation of Polaris family cards, and help to better identify the specific names of various family GPUs. The chip 10 has the name of Polaris Ellesemere, while that Polaris 11 is indicated as Baffin. For these chips we expect different variations, as occurred earlier for the various architectures developed by AMD: Ellesemere XT for the most complex GPU and Ellesemere Pro for the declination with fewer stream processors, and similar dynamic for the Baffin GPU.

We do not know at the present time the number of stream processors integrated in these GPUs, let alone the clock frequencies or other architectural features. And 'likely that some details might leak closer to the launch date, but without being able to have access to architectural specifications only know the number of stream processors does not allow to derive performance data, especially with architectures using for the first time a production technology as sophisticated as one to 14 nanometers. We just have to wait, therefore, because the debut of these solutions such as that of competitive offerings from NVIDIA will not happen before the second half of the year.