At its Investor Day presentation AMD took a good part of it to talk about GPUs. They announced that they will continue to make GPUs for every segment of the market. The company plans to continue to improve on its Graphics CoreNext architecture, but is really betting on the huge performance / watt increase with its GPUs in 2016. The reason for this increase will be its move to the 14nm FinFET process. It is interesting to point out that AMD did not specifically mention “14nm”, but mentioning FinFET really gives it away. AMD is expecting a 2x or 100% gain in performance / watt over its current generation of GPUs.
AMD’s new GPUs will focus on CPU-efficient graphics API’s like DirectX 12 and Vulcan, Windows 10, and the fact that Ultra HD displays will becoming more affordable. The company also announced a major design win, the was unnamed, but will take shape this quarter and will hit markets in 2016.
In 2015 we will see AMD’s first GPU to leverage HBM. It will offer double the memory bandwidth of GDDR5 and have half its power draw. It was announced that the companies HBM implementation will see memory dies placed on the same package as the GPU die, so this means we will no longer see memory chips surrounding the GPU on the PCB. The company is expected to announce a few high-end graphics cards around Computex 2015.
Source: TechPowerUp! | News Archive