What AMD’s FidelityFX Super Resolution means for GPU-starved gamers
For many people, inflated video card prices due to the cryptocurrency boom are the biggest obstacle towards upgrading. As such, many may see AMD’s announcement that FidelityFX Super Resolution works on older cards, including Nvidia cards, as a godsend.
But what is AMD’s FidelityFX Super Resolution in the first place, and what does it do for video cards?
FidelityFX Super Resolution, or FSR for short, is a form of AI upscaling. It uses artificial intelligence and machine learning to be able to increase the resolution of an image.
Using AI to upscale images isn’t a new concept. Nvidia, AMD’s rival in the GPU space, already has its own version of this, called Deep Learning Super Sampling (DLSS). And this is on top of the various AI upscaling programs and apps available for artists.
That said, FSR has one big advantage over DLSS—it’s not tied to any specific hardware.
AI upscaling on (almost) any video card
When Nvidia revealed DLSS a couple of years ago, it tied the feature to its then-brand new RTX cards. That remains true up until today as DLSS uses the Tensor Cores found in the Nvidia’s GeForce RTX 20 and RTX 30 series of video cards.
With this in mind, many people were fully expecting AMD to follow suit. Most people thought FSR would be tied to the RDNA2 architecture of the new Radeon RX 6000 GPUs.
Imagine most people’s shock when at Computex 2021 in late May, AMD revealed that FSR wouldn’t be tied to RDNA2. In fact, it wouldn’t be tied to AMD hardware at all. Instead, AMD announced that FSR would work even on older video cards, even Nvidia ones.
But how exactly did AMD pull this off? How did it enable real-time AI upscaling without the need for dedicated silicon?
The answer is that AMD is using a slightly different method of AI upscaling compared to DLSS. Specifically, FSR uses a spatial rendering algorithm at the end of the graphics pipeline when a scene has already been rendered.
However, there’s a cost to what AMD is doing. Nvidia also used a spatial rendering algorithm in the panned first version of DLSS. The latter only achieved mainstream acceptance following the DLSS 2.0 upgrade. This made the algorithm take motion data from the game into consideration, resulting in better image quality.
That said, it’s hard to call what FSR puts out as “bad,” at least based on what AMD has shown. More importantly, it does look to provide significant performance boosts in games that support it.
The one game demoed by AMD, Gearbox’s Godfall, achieved a 41 percent performance boost while running on an Nvidia GTX 1060. Meanwhile, on a more modern card, such as an RX 6800XT, the game was able to hit 150 FPS while running at 4K with ray-tracing and graphics set to epic.
Will FSR save you from having to upgrade?
With AMD’s FSR allowing for AI upscaling on even older cards, the question then is it will allow you to play newer games at decent framerates on older video cards. Sadly, the answer to this is not as straightforward as it seems.
AMD did not give any specific details on what video cards would support FSR. Yes, the GTX 1060 it used is the most popular video card on the Steam Hardware Survey. But questions remain on whether it can work on less powerful cards or even integrated graphics cards.
The other, more important factor is whether developers actually add support for FSR into their games. When it announced the technology, AMD did not mention which games would support it (aside from Godfall).
AMD has also stated that over 10 “game studies and engines” will be implementing FSR in 2021. It also confirmed that more details on these would be revealed following FSR’s June 22 launch.
In addition, AMD also announced that FSR will be open-source, allowing developers to integrate it much more easily in their games.
What games end up supporting FSR that will ultimately decide whether the technology will extend the life of your existing video card. Should FSR support be added to some of today’s hottest titles and most anticipated games, then the argument for it becomes even more compelling.
If you like reading our content, why not show your appreciation by treating us to a cup of coffee? (or two, if you’re feeling generous)