AMD FidelityFX Super Resolution Source Code Coming Next Month

[phoronix]

Phoronix: AMD FidelityFX Super Resolution Source Code Coming Next Month

AMD today released FidelityFX Super Resolution that was announced earlier this month at Computex. Today it's Windows-only with no Linux support being introduced. FidelityFX Super Resolution is open-source but the code drop will not be until next month...

https://www.phoronix.com/scan.php?page=news_item&px=AMD-FXSR-Source-Next-Month

[obri]

Hmm,

I am curious how well this performs and if it can compete with NVidias DLSS performance wise.

And will it come to proton? Should be possible, when it is open source, shouldn't it?

[Termy]

Hmm,

I am curious how well this performs and if it can compete with NVidias DLSS performance wise.

And will it come to proton? Should be possible, when it is open source, shouldn't it?

the Computerbase-Review seems quite positive. Not quite the fidelity of DLSS2.0, but FAR ahead of DLSS1.0 and a nice performance boost. I haven't expected it to be that good at quality after seeing the first videos from the announcement tbh.
Let's hope it gets adapted to Linux in time for any big games making use of it ^^

[oiaohm]

Hmm,

I am curious how well this performs and if it can compete with NVidias DLSS performance wise.

And will it come to proton? Should be possible, when it is open source, shouldn't it?

Being once open source hopefully not too far out.

the Computerbase-Review seems quite positive. Not quite the fidelity of DLSS2.0, but FAR ahead of DLSS1.0 and a nice performance boost. I haven't expected it to be that good at quality after seeing the first videos from the announcement tbh.
Let's hope it gets adapted to Linux in time for any big games making use of it ^^

https://www.youtube.com/watch?v=yFZAo6xItOI

Termy is reviews are not coming down even. I think it will depend on the game in the end. There are particular bad artefacts that DLSS 2.x method generates that fidelityfx super resolution method cannot generate of course this also means particular bits of image reconstructions fidelityfx super resolution cannot do. Yes those bits of image reconstruction that is generating the DLSS artifacts problem is not having those bits of image reconstruction will result in particular games looking worse compared to DLSS 2.x but DLSS 1.0 where the reconstruction was generally bad the AMD solution is way ahead . Yes it will be possible to be playing game that you will be hitting DLSS 2.x artefact over and over again that is ruining you play that the amd solution is not going to-do. There is also the case the lack of image reconstruction also means if the game generates a artifact itself amd solution is not going magically fix it as a DLSS image reconstruction could.

Interesting point is that it appears that amd method has lower gpu overhead.

We have a mixed bag here. How mixed will come clear as more reviews come out.

Absolutely better than the DLSS1.0 on its best setting vs the worst setting of fidelityfx super resolution. What I am interested in is once open source will the method in amd solution be useful for items like gamescope for generic up-scaling of old games and programs.

I would say AMD here has put out something in the competitive range. It will be very interesting once this is open source to see how amd is doing this. There is something importantly different between fidelityfx super resolution and DLSS all versions. fidelityfx super resolution uses a single frame to generate a single output frame where DLSS is in fact using multi frames to of past and present to generate the current current frame out output. So amd method in theory should be able to be used with your normal photo editing where you only have a single frame.

[Jedibeeftrix]

Interesting that Hardware Unboxed said that AMD had told them that no part of FSR exists in the driver (technically - you don't need 2021.6.1 to use FSR), as it is entirely a function of game code.

What this means for linux support...?

[evil_core]

There is something importantly different between fidelityfx super resolution and DLSS all versions. fidelityfx super resolution uses a single frame to generate a single output frame where DLSS is in fact using multi frames to of past and present to generate the current current frame out output. So amd method in theory should be able to be used with your normal photo editing where you only have a single frame.

Does it mean lower input lag with AMD FidelityFX solution?

[arun54321]

Is this a shader like program? Why does it need OS specific code?

[Floturcocantsee]

Is this a shader like program? Why does it need OS specific code?

Its a shader that is added between the finished image and UI processing steps. It works similar to any other shared library.

[jrch2k8]

I'm just saying here, since there is no code yet BUT BUT it may be possible to use it with VkBasalt + CAS and that basically fix all games BUT well we have to wait until the code drop

[middy]

Being once open source hopefully not too far out.

https://www.youtube.com/watch?v=yFZAo6xItOI

Termy is reviews are not coming down even. I think it will depend on the game in the end. There are particular bad artefacts that DLSS 2.x method generates that fidelityfx super resolution method cannot generate of course this also means particular bits of image reconstructions fidelityfx super resolution cannot do. Yes those bits of image reconstruction that is generating the DLSS artifacts problem is not having those bits of image reconstruction will result in particular games looking worse compared to DLSS 2.x but DLSS 1.0 where the reconstruction was generally bad the AMD solution is way ahead . Yes it will be possible to be playing game that you will be hitting DLSS 2.x artefact over and over again that is ruining you play that the amd solution is not going to-do. There is also the case the lack of image reconstruction also means if the game generates a artifact itself amd solution is not going magically fix it as a DLSS image reconstruction could.

Interesting point is that it appears that amd method has lower gpu overhead.

We have a mixed bag here. How mixed will come clear as more reviews come out.

Absolutely better than the DLSS1.0 on its best setting vs the worst setting of fidelityfx super resolution. What I am interested in is once open source will the method in amd solution be useful for items like gamescope for generic up-scaling of old games and programs.

I would say AMD here has put out something in the competitive range. It will be very interesting once this is open source to see how amd is doing this. There is something importantly different between fidelityfx super resolution and DLSS all versions. fidelityfx super resolution uses a single frame to generate a single output frame where DLSS is in fact using multi frames to of past and present to generate the current current frame out output. So amd method in theory should be able to be used with your normal photo editing where you only have a single frame.

to add another here is gamers nexus: https://www.youtube.com/watch?v=KCzjQ4qP124

steve did say, technically, FSR can actually work further back than polaris 470 for amd and further back than the 1060 from nvidia. its just not officially supported. but steve said its "not bad." which is pretty big since he usually nitpicks everything.