AMD FSR vs. Nvidia DLSS: Which Upscaler Reigns Supreme?

[JarredWaltonGPU]

I would expect DLSS to die in the future. Anyone who has been around long enough knows it's not the better technology that wins, it's the one that reaches the most customers. An open source product that can run on anything is more desirable that a proprietary feature on a limited number of GPUs.

Not a chance. G-Sync isn't dead yet, and DLSS will be around for a very long time. You can get by with having proprietary software features when 80% of graphics cards use your hardware. RTX is at about 25% of the market (according to Steam) I think, and in another year or two it will probably be 35% -- including nearly all mid-range or high-end gaming PCs built in the past five years. I figure if G-Sync can hang around for several years after its primary competition mostly closes the gap, DLSS — which doesn't really have any direct competition yet, because FSR is quite different — will stick around for at least five more years. XeSS could kill it off, but only if it works as well as DLSS and also runs on Nvidia's tensor cores.

 

[watzupken]

In my opinion, there is no need for such comparison test. At least for me, I've seen enough to know what is the conclusion when the main focus is on image quality. There is no doubt that FSR tend not to do as well as DLSS in most cases. FSR wasn't design to be the best looking upscaler in the first place. It is more about easier to adopt, and works across different GPU architectures. In some games where DLSS is not supported, Nvidia hardware should also be able to utilise FSR to improve performance if required. And to me, FSR have met its objective where it may not look as good as DLSS, but does preserve good enough image quality and performance even at the higher tier settings. If I am forced to lower native resolution to 720p to get smooth frame rates, then I feel I would rather use 1080p with FSR at best or high quality settings, which should achieve a better IQ.

 

[watzupken]

Not a chance. G-Sync isn't dead yet, and DLSS will be around for a very long time. You can get by with having proprietary software features when 80% of graphics cards use your hardware. RTX is at about 25% of the market (according to Steam) I think, and in another year or two it will probably be 35% -- including nearly all mid-range or high-end gaming PCs built in the past five years. I figure if G-Sync can hang around for several years after its primary competition mostly closes the gap, DLSS — which doesn't really have any direct competition yet, because FSR is quite different — will stick around for at least five more years. XeSS could kill it off, but only if it works as well as DLSS and also runs on Nvidia's tensor cores.

G-Sync isn't dead yet. But if you look at the number of monitors with G-Sync, its miserable. It won't do Nvidia any good to hang on to it since I don't believe it brings in a lot of revenue.

In my opinion, it may be harder for DLSS to "die" as the vast majority of dedicated graphic solution out there are from Nvidia. That goes to show consumers' preference, and chances of them picking up a Nvidia card going forward with support DLSS is quite high. I don't believe AMD or Intel can do their market share much damage in the short run.

 

[-Fran-]

Except nvidia is very serious when it comes to pushing their tech. That's why with DLSS nvidia do it a bit different. They try to make it as easy as possible for game developer to implement them and they work with as many game engine developer to make DLSS being part of the game engine from the get go. For game developer that use those engine all they need to do is enable the plugin and it will work automatically for them. The heavy lifting on DLSS were all done by nvidia.

You mean like PhysX? Or like GameWorks? Or the "3D" glasses push?

There's plenty tech they've created and is being used, for sure, but most of the "surviving" tech can be used cross-brand and not using dedicated hardware from them like DLSS.

Regards.

 

[renz496]

You mean like PhysX? Or like GameWorks? Or the "3D" glasses push?

There's plenty tech they've created and is being used, for sure, but most of the "surviving" tech can be used cross-brand and not using dedicated hardware from them like DLSS.

Regards.

Did Bullet end up being the "PhysX killer" like what Roy Taylor said before? What game developer end up using GPU open effect like TressFX without being sponsored by AMD? (the case with Crystal Dynamics is a bit different). What happen to AMD 3D stereoscopic solution where they want monitor makers to compete among themselves to provide the drivers?

The thing you mention even when they were open source or can work on any hardware they were something are not popular even among game developer or consumer.

Actually it doesn't matter if DLSS continue to live or not. For nvidia user with capable hardware DLSS is just something extra they can take advantage of if they want to improve their performance. The way i see it people are angry because the tech did not work on their older hardware or their preferred brand. Then we saw AMD pushing things like FSR that can benefit everyone. And yet some people still hope DLSS to die because it will not work on certain hardware or they simply hate Nvidia.

 

[-Fran-]

Did Bullet end up being the "PhysX killer" like what Roy Taylor said before? What game developer end up using GPU open effect like TressFX without being sponsored by AMD? (the case with Crystal Dynamics is a bit different). What happen to AMD 3D stereoscopic solution where they want monitor makers to compete among themselves to provide the drivers?

The thing you mention even when they were open source or can work on any hardware they were something are not popular even among game developer or consumer.

Actually it doesn't matter if DLSS continue to live or not. For nvidia user with capable hardware DLSS is just something extra they can take advantage of if they want to improve their performance. The way i see it people are angry because the tech did not work on their older hardware or their preferred brand. Then we saw AMD pushing things like FSR that can benefit everyone. And yet some people still hope DLSS to die because it will not work on certain hardware or they simply hate Nvidia.

Not sure TBH. It'll depend on what libraries most are using, which may be HavoK instead? I wouldn't be surprised each graphics engine comes with their own pre-cooked physics one. And what you mentioned about TressFX is basically the important take away. Some technologies do make sense to include and others are just either over the top and not cost effective or just would segregate your user-base and you don't want that as a game developer. DLSS is such tech, much like any other technology that requires exclusive hardware to run. Even nVidia is having issues with G-Sync to the point where they had to bend over and lower their certification standard and low-key accept Freesync.

So, I do agree some people doesn't like DLSS for the wrong reasons (hate, bias, fanboism, etc), but there are real reasons why you'd want such a tech to either be accessible to others (they can perfectly license it) or just make it open (something nVidia will never ever do) so any vendor can implement the algorithm with their own take of the hardware need (much like an API). I personally do not like any technology that upscales images, because I want them to run at acceptable frame-rates natively instead, but I can see the want/need for such tech. So, it makes sense for it to be available for such users, which I'm willing to bet, are mostly people using lower end hardware instead of $600+ GPUs.

I hope that makes sense? It's a mixed bag of rambling and real reasons, but I guess the short version would be: whatever tech fits the shoe of the developer will be implemented, unless they're showered with money (and you're right there). I like tech to be available for as many people as possible and not hiding it behind unnecessary paywalls.

Regards.