Microsoft Talks About Using AMD and Nvidia GPUs Simultaneously in DirectX 12

/ 1 year ago


It’s no secret that DirectX 12 is going to make a huge change in the way we game, it’s going to bring improved performance, optimisation and a whole lot more when it launches alongside Windows 10 later this year. What is interesting, however, is how the API will be able to utilize all available hardware, even if it’s from different hardware manufacturers, something that hasn’t really been hit upon with current generation hardware and software.

DirectX 12 is seeing widespread adoption, at a rate that hasn’t been seen since the switch to DirectX 9. “It’s been the fastest adopted API in more than a decade.” said Microsoft’s Principal Direct X Development Lead Max McMulle.

McMulle also went on to explain the DX12 Multiadapter functionality, which will support both linked and unlinked GPUs. Linked is what we’re familiar with, Crossfire and SLI, but unlinked is quite different, as it’ll accept GPU hardware such as Intel, AMD and NVidia simultaneously and even allow for the use of each cards unique feature sets. “This feature exposes direct control of two linked GPUs as well. So you might have known as Crossfire or SLI from AMD and NVIDIA, respectively, but you’ll get direct control of both GPUs in DX12 if you so choose.” said McMulle.

The discretion lies with the developer, if they code their game or software to handle both types of hardware at the same time, it can be done, what this means for gamers remains to be seen, but the prospect of upgrading to a new GPU and no having to get rid of your old one is no bad thing, allowing to use all the rendering power at your disposal. Even more importantly, it’ll allow independent memory allocation in Crossfire and SLI systems, giving you the VRAM of all cards (linked or unlinked), something that is currently unavailable in current multi-GPU configurations.

Thank you DSO for providing us with this information.



Topics: , , , , , , , , , ,

  • Danpapua

    Here is a major question: Will it allow cross architecture support? For example, Nvidia’s Fermi to Nvidia’s Kepler or Maxwell Architectures?

    • DynamiteSJ

      if its able to go as extreme to using AMD, Intel, and Nvidia together.. i would think different architectures of the same make is a not a problem.

      • Danpapua

        Similarities in architecture can cause more issues than differences.

  • Gnuclear

    I bet Nvidia is going to try to disable the ability to use Nvidia GPU’s with AMD GPU’s in their drivers, and try to spin it as “giving you the best experience”.