Although there were a few leaks earlier, AMD has now officially announced its new Polaris GPU architecture which should be coming in mid-2016.

Radeon Technologies Group, led by Raja Koduri, is describing the Polaris GPU architecture as a historic leap in performance per watt for Radeon GPUs, mostly due to a bunch of new features as well as the new 14nm FinFET manufacturing process.

While AMD still did not unveil all the key details for its upcoming Polaris GPU, it did say that it wil be based on a 4th generation Graphics Core Next architecture which will include Primitive Discard Accelerator, Hardware Scheduler, Instruction Pre-Fetch, Improved Shader Efficiency and new Memory Compression.

In addition to new key 4th generation GCN features, the Polaris GPU architecture will also have a next generation display engine with HDMI 2.0a and DisplayPort 1.3 outputs and feature next generation multimedia engine with h.265 main10 decode up to 4K and h.265 60FPS encode at 4K resolution.

AMD puts a lot of hope into the new FinFET manufacturing process and says that Polaris GPU architecture is basically made for FinFET.

During its presentation, AMD showcased a working Polaris architecture entry-level GPU meant for small form factor and compared it to Nvidia's GTX 950. Both systems were able to get 60FPS in Star Wars Battlefront game with Nvidia drawing 140W and Polaris GPU architecture drawing 86W.

Introducing the Polaris GPU architecture AMD said that "Our guiding light is to power every pixel on
every device efficiently. Stars are the most efficient photon generators of our universe. Their efficiency is the inspiration for every pixel we generate."

AMD expects to have the first Polaris architecture GPUs ready in mid-2016.













Source: AMD.com.