According to news from this website on March 5, AMD Chief Technology Officer Mark Papermaster said in an interview with the YouTube channel No Priors that it is developing AI-based super-resolution technology for gaming devices.
Mark Papermaster said:
This is a special year for us because we have spent so many years developing the hardware and software capabilities of AI. We just finished applying AI across our entire product portfolio: cloud, edge, PC, embedded devices and our gaming devices. We're making our gaming devices support AI super-resolution, and 2024 is really going to be a huge deployment year for us.
The conversation showed that As part of the penetration of AI into the overall product line, AMD is considering integrating AI into the super-resolution solution for gaming devices.
Although Mark Papermaster did not directly mention FSR, he used the term gaming devices in the conversation, so foreign media VideoCardz believes that this should be related to FSR.
Foreign media Tom's Hardware believes that theoretically integrating AI can further improve the image quality of FSR technology.
This site noticed that the two major competitors of AMD FSR - NVIDIA's DLSS and Intel's XeSS have integrated AI technology. However, the former is closed source, and although the latter promises to be open source, its implementation is not in good condition. It is impossible to know what role AI plays in both. VideoCardz understands that AMD is expected to use machine learning algorithms to reduce artifact issues in FSR.
Currently, DLSS uses the Tensor Core unit on NVIDIA graphics cards, and Intel's XeSS algorithm can be better implemented on the XMX hardware of its graphics cards. Tom's Hardware believes that if AMD's AI super-resolution technology also requires hardware support, it will probably be limited to APUs equipped with the XDNA AI engine and RDNA3 graphics cards with improved AI computing power.
The above is the detailed content of AMD is developing AI version of super-resolution technology, which is expected to improve FSR image quality. For more information, please follow other related articles on the PHP Chinese website!