Intel reveals 14nm discrete GPU design at ISSCC

Intel reveals 14nm discrete GPU design at ISSCC

Intel reveals 14nm discrete GPU design at ISSCC

At the IEEE International Solid-State Circuits Conference (ISSCC) Intel has unveiled a new 14nm dedicated GPU design, showcasing the company’s intentions to enter the graphics market in the future. 

These GPU ambitions became clear after Intel hired Raja Koduri, the former head of AMD’s Radeon Technologies Group, who now leads INtel’s newly formed Core and Visual Computing Group as Chief Architect and Senior Vice President.    

Over the past while, there has been a lot of misinformation spreading about Intel’s discrete GPU design, with many speculating that it will become a consumer product. This is not the case, as Intel has made it clear that this is a prototype chip that is designed to examine power regulation and performance control technologies using an integrated voltage regulator. Intel’s 14nm discrete GPU is a test chip for research purposes, not an upcoming product. 

To put this chips deficiency into further perspective, the frequency range of this chip is 50-400MHz, whereas the iGPU in the i7 8700K can operate at up to 1.2GHz. Intel has made it clear that their plans with this chip are to research potential efficiency savings to their GPU architecture, allowing execution units to be enabled or disabled at will and offer boost clock speeds whenever and wherever they are required on the GPU. 

If Intel wants to scale up their GPUs to create viable discrete offerings that can compete with AMD or Nvidia, the company needs to increase their power efficiency and introduce more advanced core features like Turbo boost and similar technologies. AMD only started using core boost technology with Polaris, with Nvidia offering the function since their GTX 600 series.  

Intel reveals 14nm discrete GPU design at ISSCC  

At this point comparing this GPU to what is currently offered by AMD or Nvidia is pointless, not only because this is an Intel test chip but because it was never designed with high levels of performance in mind. To create a competitive GPU offering Intel needs rapidly improve their graphics technology, both in terms of efficiency and performance, which means that Intel is likely to develop a lot more prototype graphics processors in the near future.  

In the future Intel will use what they have learned with future execution unit designs and develop products on a new process node, perhaps creating a viable product for the consumer GPU market. This development is likely to take years to get off the ground, though it would be great to see the GPU wars become a three-horse race between Intel, AMD and Nvidia. 

Intel reveals 14nm discrete GPU design at ISSCC  

Update – Intel has released a statement which confirms that this chip is for testing purposes only, offering an opportunity to test future possible power management designs. 

      Last week at ISSCC, Intel Labs presented a research paper exploring new circuit techniques optimized for power management. The team used an existing Intel integrated GPU architecture (Gen 9 GPU) as a proof of concept for these circuit techniques. This is a test vehicle only, not a future product. While we intend to compete in graphics products in the future, this research paper is unrelated. Our goal with this research is to explore possible, future circuit techniques that may improve the power and performance of Intel products.

You can join the discussion on Intel’s 14nm prototype discrete GPU on the OC3D Forums.Â