Comparing DLSS, FSR, and XeSS Input Latency

Table of Contents

1. Introduction
2. Upscaling Technologies Overview
3. Nvidia's DLSS (Deep Learning Super Sampling)
4. AMD's FSR (FidelityFX Super Resolution)
5. Intel's XCSS (Z Super Sampling)
6. Limitations of DLSS Frame Generation in Cyberpunk 2077
7. Latency Comparison in Cyberpunk 2077
8. Latency Comparison in Hitman 3
9. Insights into the Different Technologies
10. Conclusion

Introduction

In the world of gaming, upscaling technologies have become increasingly common. GPU manufacturers like Nvidia, AMD, and Intel offer their own solutions to enhance the graphical quality of games. These technologies aim to improve performance and visual fidelity by utilizing advanced algorithms and techniques. This article explores the different upscaling technologies available in the market and their impact on latency during gaming Sessions.

Upscaling Technologies Overview

Before diving into the specific upscaling technologies, it's crucial to understand the concept of upscaling itself. Upscaling involves taking a lower resolution frame and then using algorithms to increase its size to match a higher resolution display. The goal is to enhance the visual quality of the Game without requiring the GPU to render at native higher resolutions.

Nvidia's DLSS (Deep Learning Super Sampling)

Nvidia's DLSS (Deep Learning Super Sampling) is one of the leading upscaling technologies in the market. DLSS utilizes artificial intelligence and deep learning algorithms to upscale lower resolution frames. It analyzes the previous frame, the next frame, and generates an intermediary frame to double the frame rate. This technique reduces the load on the GPU and enhances the overall gaming experience. DLSS is currently in its third generation, with new features such as frame generation.

However, DLSS has certain limitations. It needs to know what the next frame looks like, causing additional latency as it holds back the new frame to render the intermediary frame. Despite this latency, DLSS offers impressive performance gains compared to rendering frames at full resolution.

AMD's FSR (FidelityFX Super Resolution)

AMD offers its own upscaling technology called FSR (FidelityFX Super Resolution). Unlike DLSS, FSR is an open-source alternative that works on any GPU, including Nvidia's RTX 4090. FSR does not rely on AI magic, making it more accessible to a wider range of users. However, in certain scenarios, FSR can result in a slight decrease in image quality compared to DLSS.

Intel's XCSS (Z Super Sampling)

Intel's upscaling technology is known as XCSS (Z Super Sampling). It employs a neural network to enhance visual quality. Like FSR, XCSS is also an open-source solution. However, it is explicitly designed for Intel GPUs, although there is a fallback option for AMD and Nvidia cards. XCSS may not offer the same level of performance as DLSS and FSR, but it still provides a viable upscaling option for Intel users.

Limitations of DLSS Frame Generation in Cyberpunk 2077

Unfortunately, DLSS frame generation faces certain limitations, as observed in Cyberpunk 2077. The game's engine experiences issues when frame generation is enabled, resulting in unplayable conditions. The rendering process locks up, causing delayed inputs and hindering accurate latency measurements. This issue is not isolated to a particular test system, as reinstalling the game and updating drivers did not resolve the problem.

Latency Comparison in Cyberpunk 2077

Despite the challenges faced with frame generation in Cyberpunk 2077, latency comparisons were made for other upscaling technologies. When tested on a system with a Gigabyte M32Q monitor running at 165Hz, DLSS on auto mode reduced the total system latency from 35.76 milliseconds to 26.14 milliseconds. Switching to DLSS ultra performance mode further reduced latency to 24.65 milliseconds.

AMD's FSR on auto mode offered similar results to DLSS on auto, with a slightly higher latency of 26.7 milliseconds. However, enabling FSR's ultra performance mode increased latency to 30.38 milliseconds. Despite this increase, it still outperformed not using any upscaling technology, which resulted in 35.76 milliseconds of latency.

Latency Comparison in Hitman 3

To further understand the impact of upscaling technologies, Hitman 3 was tested, which supports all three upscaling technologies. Latency measurements were challenging in this game and required specialized tools. FSR on the quality mode exhibited impressive latency reduction, going from 60 milliseconds without upscaling to just 45 milliseconds.

Intel's XCSS, however, performed the worst among the upscaling options, with the ultra quality mode only reducing latency by three milliseconds compared to the stock settings. DLSS quality mode offered a latency improvement of 2.5 milliseconds compared to XCSS performance mode. DLSS ultra performance mode was the fastest, with only a two-millisecond increase in latency compared to stock settings.

Insights into the Different Technologies

From the latency comparisons, it is evident that DLSS is a fast process, upscaling lower resolution frames in the same or less time than rendering higher resolution images. FSR, on the other HAND, seems to be more intensive, resulting in increased latency in certain scenarios. However, FSR's quality mode still offers impressive results, even on Nvidia hardware, considering its open-source nature.

Intel's XCSS, while not performing as well in these tests, may provide a satisfactory experience for users with Intel GPUs. It is worth noting that these results are specific to the tested configurations, and individual experiences may vary.

Conclusion

In conclusion, upscaling technologies have become an integral part of gaming, providing improved performance and visual quality. Nvidia's DLSS, AMD's FSR, and Intel's XCSS offer different approaches to upscaling, each with its own strengths and weaknesses. Despite some limitations and latency considerations, these technologies have shown promising results in reducing latency and enhancing the gaming experience.