As has been said many times here, NVIDIA’s spearhead when it comes to GPUs over AMD is DLSS, which bases its strength on making the most of the computing capacities of tensor units. Which are not found in AMD GPUs and this led them to develop their FidelityFX Super Resolution with a totally different approach than NVIDIA. Not only in the nature of its algorithm, but also in the way it is deployed.
AMD releases source code for its Fidelity FX Super Resolution
The main advantage of FSR over NVIDIA DLSS is that since it is not an artificial intelligence algorithm, it does not require a period of learning and subsequent inference. this allows deploy it directly in all games And while it’s not as good as NVIDIA’s DLSS, it’s great not only for any game, but also for any GPU without an AI unit to boost its frame rate at the highest resolutions. high.
This is a completely brand and architecture agnostic algorithm that will benefit not only AMD GPUs, but NVIDIA GPUs without Tensor Cores and video game consoles as well. Not only of this generation, but also of the previous one by running on FP16, allowing the PS4 Pro to benefit from this algorithm. The fact that it works with any GPU is great news, as its ease of adaptation will breathe new life into older GPUs in an era of complete hardware shortage.
Plus, it’s already fully integrated into Unity and Unreal Engine, so any games you create with these engines will already have FidelityFX Super Resolution built into the code.
New technical details have been disclosed
Comments in the source code of the FidelityFX Super Resolution give us new insight into how the algorithm works, which not only does not use AI, but is also not based on temporality or image information. previous ones. What information we were missing? Know which algorithm you are using to get a higher resolution version of the image, which turned out to be Lanczos’ algorithm, one of the most powerful in the world of linear algebra, the mathematical discipline from which computer graphics are derived.
So now we have a completely rough idea of how the AMD algorithm works, moreover, developers will be able to modify the code to suit their needs at any time. Something that cannot be done with NVIDIA’s DLSS because it is an undocumented black box, proprietary and whose internal functioning we do not know because it is kept by NVIDIA in the greatest secrecy.