AMD’s FidelityFX Super Resolution (FSR) and Nvidia’s Deep Learning Super Sampling (DLSS) are fast becoming must-have features in the latest PC games. They help the best graphics cards by rendering the game at a lower resolution to improve performance. But among them, which one reigns supreme?
DLSS may be the older of the two, but some clever technology helps it keep its edge. However, AMD has improved FSR with a new version that provides similar image quality to DLSS, and without the need for an AMD GPU. From performance to image quality to gaming support, here’s how FSR and DLSS stack up.
FSR vs DLSS: What’s the difference?
Image used with permission of the copyright holder
FSR and DLSS, despite achieving the same goal, are very different technologies. At a high level, both tools render your game at a lower resolution to improve performance. After that, algorithms enhance the image to match your monitor and fill in the missing information based on various inputs. The specificity is where things get messy.
DLSS is an AI-enhanced supersampling algorithm. Dedicated Tensor cores on RTX graphics cards run an AI model that helps upscaling, which helps fill in information missing from the lower internal resolution. The model is fed with temporal or temporal data that shows the movement of objects from one frame to the next. This helps the algorithm see new details that it couldn’t achieve with a frame.
Despite the AI, DLSS is similar to Temporal Super Resolution (TSR), which we first saw in Ghostwire Tokyo. FSR uses a similar approach, but only in a different version.
As if the differences between FSR and DLSS weren’t confusing enough, there are two versions of FSR. FSR 1.0 is pretty simple. It uses an established algorithm to upscale the image and fill in missing details before applying the sharpening filter. Critically, it happens after anti-aliasing, so it tries to reconstruct an image that has already been refined. This generally leads to much poorer image quality.
FSR 2.0 uses the same sharpening algorithm and filter, but it happens before anti-aliasing and takes more input from the game. It’s basically a TSR with an AMD label. It includes temporal data for more information and works with clean rendering before anti-aliasing. This leads to much better image quality than the first version.
FSR versus DLSS: Performance
The whole point of FSR and DLSS is to improve your gaming performance, so it makes sense to start there. AMD currently has two versions, so FSR, and there are quite a few differences between them. Deathloop is one of the few games that has both versions of FSR and DLSS, so we used it for performance evaluation.
Image used with permission of the copyright holder
Using the quality mode for both, which applies a scaling factor of 1.5x, FSR 1.0 provided the greatest benefit. While every game is different, our FidelityFX Super Resolution review found that FSR 1.0 generally delivers better performance than DLSS — but at the cost of image quality. This is especially true in more aggressive quality modes.
Switching to Performance mode, which applies a 2x scaling factor, the same set is present: FSR 1.0 leads in performance, DLSS is second, and FSR 2.0 is barely behind. However, it is important to keep in mind native resolution performance. DLSS is about 2% faster than FSR 2.0, but both offer nearly double the frames per second of native resolution.
Image used with permission of the copyright holder
Although FSR 1.0 leads the way in terms of performance, this does not mean that it is the best option. Both FSR 2.0 and DLSS are much better at preserving image quality, which is important to keep in mind. FSR 1.0 can improve your performance even more, but lowering the resolution will improve performance even more — it’s a balance between performance and image quality.
FSR vs. DLSS: Image quality
Although performance is the purpose of FSR and DLSS, you cannot consider it separately from image quality. We use each other Deathloop again as an example, but we tested FSR and DLSS in a wide range of games, including God of war and Rainbow Six Extraction. We recommend clicking or tapping on the images below to see all the visual differences.
FSR 1.0 (left), FSR 2.0 (middle), DLSS (right) Image used with permission of the copyright holder
Even in one screenshot you can see that there are two camps here: FSR 2.0 and DLSS and FSR 1.0. FSR 1.0 is based on an outdated supersampling algorithm and sharpening filter. Under the hood, it’s less sophisticated, and this is reflected in the image quality with messy objects in the distance and a softer overall image.
For details, see the sign and wires in the far background. FSR 1.0 shows a jumble of pixels, while FSR 2.0 and DLSS show those details clearly. It’s hard to tell the difference between DLSS and FSR 2.0. These screenshots were taken with the quality preset and DLSS and FSR 2.0 are mostly the same.
FSR 1.0 (left), FSR 2.0 (middle), DLSS (right) Image used with permission of the copyright holder
Moving to the more aggressive Performance mode, the gap widens between the FSR 1.0 and the other two. Here, we’ve zoomed in 267% to really see the differences, and it’s clear how much FSR 1.0 lags behind with image quality. Still, FSR 2.0 and DLSS are shockingly close. If it’s hard to see the differences here, it’s impossible to see them while playing.
In addition to photo quality, supersampling is prone to visual artifacts. For DLSS, especially older versions, the problem is ghosting. Nvidia has been working on a DLSS update to remove these artifacts, but they still occasionally appear in games. FSR 2.0 has a similar problem, although we only had the opportunity to look at it in Deathloop.
DLSS (right), FSR 1.0 (left) Image used with permission of the copyright holder
Here it is easier to call a loser than a winner. FSR 1.0 is so far behind FSR 2.0 and DLSS that the additional performance becomes a moot point. FSR 2.0 and DLSS are too similar to be declared winners, so it will mostly come down to which upscaler you have access to.
FSR vs DLSS: Compatibility
AMD
One of the biggest differences between FSR and DLSS is compatibility. DLSS is an RTX feature, so it’s only available on RTX 20- and 30-series graphics cards. FSR, on the other hand, works with GPUs from AMD and Nvidia because it doesn’t require dedicated Tensor cores.
AMD has recommended hardware for FSR, but it should work on most graphics — including integrated GPUs in many cases. FSR is also available on the Xbox Series X, and since it doesn’t require any special hardware, it can also run on platforms like the PlayStation 5 (although, it’s not officially supported).
FSR is freely available for developers and is built on open source. For a while, DLSS was locked behind Nvidia’s walled garden. But growing pressure from FSR forced Team Green to go the open source route as well. This is a stark change from DLSS 1.0, which required developers to work closely with Nvidia and train an AI model on a game-by-game basis.
Still, there is a clear winner here. While DLSS is now easier for developers to access, the fact that it only works on newer Nvidia GPUs is a big hurdle.
FSR vs DLSS: Game support
Image used with permission of the copyright holder
DLSS is available in many more games than FSR. At the time of publication, we counted about 180 titles that support DLSS, but that number is growing every month. FSR is available in about 60 games, but that list is growing every month.
In addition to raw numbers, DLSS is available in many more popular games than FSR — Rainbow Six Siege and Fortnite are two main examples. Major AAA games with ray tracing tend to favor DLSS as well, with the RTX brand carrying both ray tracing and DLSS together.
FSR doesn’t have the numbers, but AMD is making inroads. Many titles now support both DLSS and FSR, not just one or the other. Recent examples include god of war Marvel’s Guardians of the Galaxy, and Cyberpunk 2077. Older games, such as marvel avengers, they also got FSR support with DLSS through patches.
DLSS has the edge here simply because it’s been around longer. The general trend we see from developers is that DLSS and FSR are included, usually not one or the other. It’s hard to say whether that trend will continue, but the developers want to satisfy as wide an audience as possible. And not everyone owns an Nvidia GPU.
FSR v DLSS: judgment
Image used with permission of the copyright holder
Overall, DLSS is the best option over FSR. This was especially true for the first version of FSR, which still accounts for the lion’s share of supported games. FSR 2.0 is still a serious contender and should become the default option for developers moving forward.
If you have access to DLSS, you should use it. But the days of buying Nvidia GPUs specifically for DLSS are over. With FSR 2.0’s impressive image quality, you can still access high-quality supersampling no matter what brand of gaming PC you have.
Editor’s recommendations
Categories: GAMING
Source: newstars.edu.vn