AMD FSR 3’s success depends on one feature.

AMD FSR 3's success depends on one feature.

AMD’s FidelityFX Super Resolution 3: The Good, The Bad, and the Latency Issue

Image Image: Jacob Roach / ENBLE

AMD recently unveiled more details about its highly anticipated FidelityFX Super Resolution 3 (FSR 3), and it seems to have ticked all the boxes that enthusiasts were hoping for. Not only does it support frame generation, but it also works seamlessly across GPUs from both AMD and Nvidia. Furthermore, AMD also announced a driver-based version of its Fluid Motion Frames technology, potentially extending game support to thousands of titles. These are undoubtedly exciting developments in the world of gaming technology, but there’s one pressing question that remains unanswered: How will AMD address the issue of latency?

Let’s take a step back and define what we mean by latency. In the context of gaming, latency refers to the total time it takes for an action to be executed on the screen from the moment it is initiated by the user. This includes the time it takes for input signals to travel through various components of the system, such as mouse cables, PC processing, and display cables, before finally being displayed on the screen.

When it comes to features like FSR 3 and Nvidia’s Deep Learning Super Sampling 3 (DLSS 3), the focus is on PC latency – that is, the time it takes for the PC to process the input and deliver the final output to the display. Both AMD and Nvidia have implemented solutions to reduce latency by minimizing the render queue, the storage of frames that the GPU needs to render. Instead of storing frames in a queue, these technologies enable direct transmission of frames from the CPU to the GPU, ensuring the latest frame is rendered with minimal delay.

However, AMD and Nvidia have taken different approaches to address latency. Nvidia’s Reflex is a per-game feature, while AMD’s Anti-Lag is implemented through its driver, allowing it to support a wider range of games. Yet, this difference in implementation comes at a cost. For example, in a comparison between Reflex and Anti-Lag using Cyberpunk 2077, Reflex achieved a 49% reduction in average PC latency, while Anti-Lag managed a 38% reduction. Additionally, the tested AMD RX 6700 XT exhibited higher average latency compared to the Nvidia RTX 3060 Ti overall.

An analysis by Igor’s Lab further emphasized the superiority of Nvidia’s approach to minimizing congestion in the render pipeline. It stated that “Nvidia’s approach – from within the game, so to speak – to clearing the congestion in the render pipeline is not only more efficient, but much more effective.” This underscores the challenges that AMD faces in addressing latency compared to its competitor.

So, why does latency matter, and how can it be reduced? Latency plays a crucial role in frame interpolation, as only half of the frames you see on the screen are responsive to your actions. When features like DLSS 3 or FSR 3 are used, the generation of frames introduces additional processing time, potentially increasing latency compared to running the game without these features. This latency can be particularly noticeable when transitioning from low frame rates, where latency is already high, making the game feel sluggish.

AMD claims that FSR 3 already incorporates latency-reducing technology that works across all GPUs. However, the extent of this latency reduction remains to be seen when FSR 3 is released. Additionally, AMD is introducing its new Anti-Lag+ feature, although details about it are scarce. According to AMD, Anti-Lag+ aims to further reduce latency through intelligent synchronization placement that takes into account the holistic rendering pipeline. This suggests that Anti-Lag+ may provide additional latency reduction in areas that matter the most.

It’s worth noting that Anti-Lag+, like its predecessor, is accessed through AMD’s driver software suite. However, contrary to Nvidia’s approach with DLSS 3, which automatically enables Reflex when DLSS Frame Generation is turned on, AMD states that Anti-Lag+ does not require FSR 3 and can be used independently. On the downside, Anti-Lag+ currently only supports AMD’s latest graphics cards, such as the RX 7900 XTX and RX 7900 XT. If you have an older AMD GPU, you will have to rely on the regular Anti-Lag to reduce latency in games.

From AMD’s perspective, Anti-Lag and Anti-Lag+ are supplementary features aimed at further reducing latency. They are not intended to replace the claimed latency-saving benefits of FSR 3 – AMD described them to me as “bonus latency reduction” features. However, concerns remain regarding whether FSR 3 can independently deliver significant latency reduction or if Anti-Lag+ can outperform its predecessor.

AMD faces an uphill battle with FSR 3, as it has always emphasized supporting a wide range of GPUs at the expense of image quality. With FSR 3, the company now also grapples with the latency challenges inherent in frame interpolation technologies. The success of FSR 3 will ultimately hinge on whether latency can be effectively addressed. While discussions about image quality and performance are inevitable once FSR 3 is released, it is imperative that AMD ensures latency remains under control for the feature to succeed.

Fortunately, we won’t have to wait long to see the outcome. AMD has confirmed that FSR 3 will be available in September in games such as Forspoken and Immortals of Aveum, and thorough testing will provide insights into its latency performance. Stay tuned!