Although DLSS has been around for a long time, it was only widely used in 2020. This algorithm will bring a balance between graphics and frame rate. Many major games have been announced to support DLSS, demonstrating the importance of this technology.
What is DLSS?
Deep Learning Super Sampling (DLSS) is an AI-based rendering technology that improves graphics performance using the AI processing power of the dedicated Tensor cores on NVIDIA GeForce RTX graphics cards. DLSS is the result of combining Supersampling to make graphics in games smoother and Deep Learning to apply general improvements in games.
DLSS is a combination of:
- Deep Learning: The process of collecting, learning, and analyzing AI data through multiple analyzes of large-scale data. From there, images can be processed better.
- Supersampling (SSAA): An anti-aliasing technology that helps display game images at higher resolutions than the screen can display, making the images appear smoother.
NVIDIA runs Supersampling on a specific game and iterates, on the graphics card in NVIDIA’s large data center. Deep Learning will intervene in these loops by repeating images, calculating the best way to apply Supersampling to the game.
From there, the system learns as much as possible about how to display the game, learns how to make the images sharper and more beautiful after each iteration.
How does DLSS work?
DLSS 1.0 version (launched in February 2018)
This version uses AI algorithms and the Neural Graphics Framework (NGX) to improve the quality of existing images. After anti-aliasing, it will compare and “learn” how to render images that are most similar in higher quality without significantly affecting the frame rate.
The AI network repeats this process and adjusts the algorithm to produce images with 64 samples/pixel from the input frame with just 1 sample/pixel. At the same time, it avoids issues related to TAA.
Techniques such as feedback coupling and motion vectors are also applied to DLSS to ensure the sharpness of images in games consistently on each frame and at higher resolutions.
DLSS 2.0 version (launched in March 2020)
NVIDIA helps improve the optimization of Tensor Cores by updating the neural AI network. DLSS 2.0 only needs to display images at a resolution of 25 – 50% (11% if Ultra Performance is selected) and combine with temporary feedback to provide clearer and more stable quality than DLSS 1.0.
The AI network in DLSS 2.0 has been modified, and NGX will provide a set of low-resolution images with motion vectors and anti-aliasing to serve the temporary feedback technique. AI will compare, find the differences, and learn how to render the best way.
NGX repeats this process continuously on tens of thousands or even millions of provided reference images, creating an AI network that can upgrade images to the desired quality and resolution.
Once completed, the AI network will be sent to NVIDIA RTX 20 – 30 Series graphics cards through the GeForce Game Ready driver. This allows your graphics card to leverage the AI of Tensor Core to run DLSS 2.0 in supported games.
DLSS 3.0 version (launched in October 2022)
The latest version, DLSS 3.0, operates like previous generations but with even more significant improvements, maintaining graphic quality and smooth response with higher performance.
One of the new features is Optical Multi Frame Generation, allowing the AI algorithm to predict and outline additional frames based on available data. This helps the machine not need a CPU to render each frame, limiting CPU and GPU bottlenecks, allowing you to optimize the maximum power of RTX 4090 with just a mid-range CPU.
DLSS 3.0 provides 4 times better performance and 2 times faster response than the original resolution, bringing a better experience to gamers on all virtual battlefields. Furthermore, DLSS 3.0 is integrated into the RTX 40 series and currently supports up to 35 popular games.
Benefits of DLSS
High-quality stability with high performance
AI-rendered graphics provide image quality equivalent to or higher than the native resolution. Reducing heat and power consumption on the graphics card while improving graphic quality, frame rate, and resolution.
Increased performance with AI
AI uses specialized Tensor technology to double computing speed compared to the past, thereby maximizing frame rate and resolution up to 8K.
Image quality customization
Easily adjust image quality, including Quality, Balanced, and Performance settings, to control the resolution rendering process in the game. In Performance mode, the system can increase the resolution 4 times from 1080p to 4K.
Boost computer performance with Tensor cores
DLSS harnesses the power of supercomputers. This means you can harness the power of the supercomputer network to improve performance and resolution.
DLSS-supported GPUs
NVIDIA GEFORCE RTX 3000 series:
- GEFORCE RTX 3090
- GEFORCE RTX 3060 Ti
- GEFORCE RTX 3060
- GEFORCE RTX 3080
- GEFORCE RTX 3070
NVIDIA GEFORCE RTX 2000 series:
- GEFORCE RTX 2080 Ti
- GEFORCE RTX 2080
- GEFORCE RTX 2070
- GEFORCE RTX 2060
- GEFORCE RTX 2080 SUPER
- GEFORCE RTX 2070 SUPER
Conclusion
The article above from FPT Shop has introduced what DLSS is, its applications, and the benefits of this technology. Hopefully, you will have a better gaming experience with this graphics card.
- NVIDIA DLSS technology is now supported for VR
- What is Ray Tracing? Which GPUs and games support this feature?