AMD FSR 3.0 – everything we know so far

As games get more and more intensive, FSR and upscaling might be the way forward

Sebastian Kozlowski

Share this article...

What might be a saving grace for a lot of gamers, especially those of use with aging hardware, is the implementation of upscaling technologies. As such FSR 3 is the next iteration of AMD’s solution to the problem able to be implemented into games.

Although it was announced with the RTX 7900 XTX GPUs, and likely in response to Nvidia’s DLSS 3 feature, it would offer quite the improvement over previous iterations like its competitor’s option. And since then, AMD held a GDC talk for developers to go over further details and implementation into games.

Today’s best GPU deals

When is AMD FSR 3 coming?

Currently, there is no timeline for when you can expect to see the new generation of AMD upscaling. As it is still being developed and worked on, it might be still some time before it comes out. Especially as it is coming with new features to implement and bring forward some good competition to what Nvidia has to offer.

Although since its announcement in November, we’d like to think it would come out within a year’s time frame. So we might yet see it sometime in Q4 of 2023 but possibly sooner as we hope for some good options for our aging hardware and getting the most out of it.

What does AMD FSR 3 add?

Looking at the key differences in what FSR is going to be adding, AMD said these may be subject to change as of yet. Since some of these might take a bit more time to implement or are just too much to implement straight away. Especially if it is expected to be released anytime soon to actually be used.

Up to 2x performance over FSR 2

FSR 3 would benefit from what FSR 2 already has. With the current implementation computing more pixels than samples in the current frame, more can be generated from interpolated frames. similar to Nvidia’s frame generation, this can produce the in-between parts and produce a smoother experience with faster performance for the graphics.

So using motion vectors and AMD Fluid Motion is what will produce these new frames. As with good motion, estimation is important for interpolation as that prevents the jittering you could see if it interprets the object in the wrong place and takes you out of the experience. So also clarifying that FSR 2 information can be used as well to improve what 3 can do.

And in the workflow timeline for the technologies, AMD has outlined how it will change (as seen in the above image). Where 3 differs from 2 is the addition of interpolation after upscaling, post-FX, and UI. These also present the frames more often so in that same scale, you get 2x more than 2, which was already doubling how many over native rendering.

Latency reduction

One of the bigger reasons to skip over any upscaling is the effect it has on the responsiveness of the gameplay. This does mean quite a lot of work is required to implement it, but it does mean that it should be a better experience as you don’t have that delay that might cost you something when playing it. So that would mean not having to choose over the framerate or the feel, although an alternative to Reflex from AMD should be coming too to help.

Easy transition

Another important feature to the developers of the games is the easy transition from one to the other. If a game already has FSR 2 implemented in it, going to 3 should be no harder than adding it in. As such it provides a better offering for the audience of the game and makes life much simpler for the devs.

Permissive license

Another important feature for accessibility is the software coming in under the MIT license. This means it is a permissive free software license allowing anyone to use, copy, modify, and so on for free. Therefore, developers don’t need permission to implement FSR 3 into their games and rather have it as soon as they need, allowing it to be featured in more games with ease.

FAQs