Nvidia graphics cards could soon lose the most important selling point over AMD

On May 19, 2022By Garmin Philip

Many users who want to buy a graphics card are currently using Nvidia. Because here you get more features. But that could change now, because AMD is taking a huge step forward in terms of software.

If you want to buy a graphics card, you often turn to Nvidia. Because Nvidia offers some features that can significantly increase your FPS in games. We are talking about DLSS, DLAA and Co – features that you have had to search in vain for in this form at AMD.

But now AMD is bringing an update for its FSR technology, which makes the gap to Nvidia significantly smaller. MeinMMO explains what it’s all about and what that means for you as a gamer.

So far, Nvidia graphics cards have offered technology that AMD outperformed

What technology is it? With DLSS (deep learning super sampling) low textures are scaled to higher resolutions with the help of AI. This allows performance to be improved even more in high resolutions when the graphics card is already under load. According to Nvidia, there is talk of an increase in the frame rate of up to 50% (via nvidia.com).

DLSS is also a technology that uses the tensor cores optimized for AI math, which are only built into Nvidia graphics cards. So far, DLSS has been Nvidia’s strongest argument why you shouldn’t buy a graphics card from the competition. This is shown, for example, in the Review overview of the GeForce RTX 3060that thanks to DLSS you can increase the performance in games by around 30%.

Because while AMD’s RX 6000 series now also offers ray tracing, AMD has not been able to offer an alternative to DLSS for a long time. The first attempt with “Fidelity Super Resolution”, FSR for short, could not keep up. But that is exactly what is changing now.

What is FSR anyway? The idea behind Fidelity Super Resolution (FSR) is the same as with DLSS: low textures are scaled to higher ones without making the game look worse.

With FSR 2.0, however, AMD has changed a few things. The latest version uses a new scanning method to get the data. With FSR 2.0, AMD works like Nvidia with temporal upsampling, with FSR 1.0 a spatial upscaler was also used.

With spatial upscaling, every single frame is analyzed. In the case of temporal upsampling, the data from each previous frame is also looked at. So there is more data available to improve image quality and more importantly stability.

The special thing about it: AMD’s FSR also works with graphics cards from Nvidia. Because unlike DLSS, you don’t need tensor cores for FSR. Because such tensor cores only have RTX graphics cards.

Where was the problem so far? AMD’s FSR 1.0 didn’t come close to DLSS. AMD also knew this and is therefore working on a second version. And the update to 2.0 was really worth it. Because many testers explain that the new version is a huge step forward.

First tests say: AMD is damn close to Nvidia

The testers of the online magazine TechPowerUp describe FSR 2.0 as a “DLSS killer” (via techpowerup.com). The new technology has made a huge step forward and offers significantly better performance than FSR 1.0. And the distance to Nvidia is getting much smaller. This is how the testers from TechPowerUp write in conclusion:

AMD has done the unthinkable – the new FidelityFX Super Resolution FSR 2.0 looks fantastic, as well as DLSS 2.0, actually DLSS 2.3 (in deathloop). Sometimes even a little better, sometimes a little worse, but overall this is a big win for AMD.

The testers from Computerbase see the improvement very similarly, but are a bit more cautious (via computerbase.de). Both techniques have their weaknesses and in terms of performance, neither DLSS nor FSR is really better. However, AMD has the advantage that its upsampling method also works on graphics cards from the competition. Sometimes one technique performs better or worse, sometimes the other.

If you want to see the differences between FSR 2.0 and DLSS 2.0, check out Digital Foundry’s video. Here you can clearly see the differences and similarities in the video: