AMD is working on alternatives to Nvidia’s DLSS upscaling

When AMD unveiled the news in the architecture RDNA 2 and the graphics cards in the Radeon RX 6000 series, the focus was partly on performance and specifications, and partly new functions such as Smart Access Memory. Another feature that was highlighted was that the Fidelity FX function suite can clean up the visual noise that occurs when scenes are to be illuminated via ray tracing, so-called “denoising”.

One feature that could be discerned during the presentation of Fidelity FX that AMD did not highlight further is something called Super Resolution. In addition to the function “to give players the opportunity for better performance with the help of ray tracing”, the function was not further explained. In conversations with The Verge, it now appears that this is a function for upscaling resolution, which is said to provide better performance when ray tracing calculated.

It is thus an alternative to Nvidia’s DLSS function, which uses dedicated Tensor cores on the graphics circuit to calculate the scaled-up image. In the conversation, it emerges that Super Resolution is an open solution that can be used on several different platforms, and is thus not specifically tied to the Radeon RX 6000 series. This suggests that the technology does not have to be tied to functions included in RDNA 2, and could be performed by standard shader processors or dedicated AI / machine learning hardware used on other graphics circuits.

AMD mentions that Super Resolution is expected to receive strong support in the industry, but also that the technology will not be ready for sharp operation when the Radeon RX 6000 series is launched. From that perspective, it seems likely that AMD’s alternative to DLSS will roll out as part of the graphics card drivers only in 2021.

Read more about resolution scaling: