AMD FidelityFX Super Resolution 2.0 - preview - Final Words & Conclusion

Concluding

We've been very skeptical about FSR 1.0 when it was released, and perhaps still are. FSR 2.0 however simply looks great whilst it can offer substantial performance increases. As a technology, we do not need to devote pages to explaining how AMD's latest feature, FSR 2.0, operates. If you need quicker framerates and implement this technique, your framerates will be significantly increased. This remains to be a temporal scaler, which has existed for years. In games, they have developed a reputation for image quality sacrifices. The challenge is that some people will dislike it, while others will not even notice it. We believe the quality mode settings are incredibly sufficient.

AMD has disregarded AI-based machine learning techniques for years now, failing or ignoring to implement such technologies at the hardware level. Within this mindset, they sought a remedy because it has been biting them recently. With the debut of the Radeon RX 6000 series, this became significantly more obvious in regards to Hybrid Raytraced games. There is where they fall behind. FSR as implementation works great, and FSR 2.0 actually looks to become pretty amazing. And it can be especially beneficial in instances where Raytracing performance is extremely important. In the early rounds of DLSS, nothing was new for NVIDIA. The first version of DLSS was a deep learning algorithm, but DLSS 2.0 is a vast improvement because NVIDIA can now train its algorithms. AMD's Super Resolution is a purely spatial upscaler, and it is debatable how far that can take you without changing the input resolution and so reducing performance. The only information that AMD's technology consumes is the original image and a depth map. DLSS can utilize thousands of photos for training purposes. In this perspective, FSR will be an interim solution to close a performance gap at a cost; we believe AMD should embrace deep learning. In the end, we conclude that the quality setting is pretty darn impressive.

AMD's biggest challenge will be finding support at a developers level, game devs need to implement this tech, and as FSR 1.0 showed, support has been minimal. However FSR 1.0 non is complemented as RSR into the driver, and that brings support to almost any game for it. That in the end will happen with FSR 2.0 too, and we feel that can't happen quick enough.

There you have it, we're very excited about FSR 2.0, below is an extra table of internal AMD results. We stopped further testing as some websites decide to breach the embargo, ergo we did not feel the urge to spend more editorial hours on it.

• Sign up to receive a notification when we publish a new article.
• Or go back to Guru3D's front page

- Hilbert, LOAD"*",8,1.