What is DLSS?
NVIDIA DLSS (Deep Learning Super Sampling) is an AI-powered upscaling technology that boosts frame rates and generates sharp images in supported games. It renders frames at a lower resolution and then uses AI to intelligently reconstruct them to a higher resolution, like 1440p or 4K. This process significantly improves performance without a noticeable drop in visual quality.
How DLSS Works: AI Upscaling
DLSS leverages dedicated Tensor Cores found on NVIDIA RTX GPUs (GeForce RTX 20-series and newer). These Tensor Cores run a deep learning neural network that has been trained on thousands of high-resolution, super-sampled images. When a game is running, DLSS takes a lower-resolution frame, analyzes it with the AI model, and then intelligently reconstructs it to the target resolution, adding detail and anti-aliasing in the process. This allows your GPU to render fewer pixels, leading to higher frame rates.
DLSS Generations: From 1.0 to 3.5
DLSS has evolved through several generations. DLSS 1.0 (RTX 20-series) offered early upscaling but sometimes had visual artifacts. DLSS 2.0 (RTX 20-series and newer) brought significant improvements in image quality and broader game support. DLSS 3.0 (exclusive to RTX 40-series GPUs) introduced Frame Generation, creating entirely new, interpolated frames using AI, leading to massive performance gains. DLSS 3.5 enhances ray tracing quality with Ray Reconstruction, improving denoising for more realistic lighting.
Performance Boosts and Visual Quality
The primary benefit of DLSS is a substantial increase in frame rates, especially at higher resolutions like 4K or with demanding settings like ray tracing enabled. For example, an RTX 4070 Ti Super might see a 2x or more frame rate increase in a DLSS 3.0 game with Frame Generation enabled. While rendering at a lower internal resolution, DLSS's AI reconstruction often results in an image that looks as good as, or sometimes even better than, native resolution rendering due to superior anti-aliasing.
DLSS vs. FSR vs. XeSS
DLSS is NVIDIA's proprietary solution, requiring RTX GPUs. AMD offers FSR (FidelityFX Super Resolution), which is open-source and works on a wider range of GPUs, including older NVIDIA cards and integrated graphics, but generally relies on spatial upscaling without dedicated AI hardware. Intel's XeSS (Xe Super Sampling) also uses AI, but is optimized for Intel Arc GPUs while also supporting other vendors. DLSS typically offers the best image quality and performance, especially with Frame Generation.
Key takeaways
- •DLSS requires an NVIDIA GeForce RTX graphics card (20-series or newer).
- •It significantly boosts frame rates, especially at 1440p and 4K resolutions, and with ray tracing enabled.
- •DLSS 3.0 with Frame Generation offers the largest performance gains but is exclusive to RTX 40-series GPUs.
- •Look for games that explicitly support DLSS in their settings menus.
- •DLSS can make demanding games playable on mid-range RTX cards, like an RTX 4060 Ti, at higher resolutions.
What to do next
- →Browse our selection of NVIDIA GeForce RTX graphics cards.
- →Read our guide: 'RTX 40-series vs. RTX 30-series: Is the Upgrade Worth It?'
- →Check out our 'Best Gaming Monitors for DLSS' recommendations.
Ready to build?
Skip the research — tell our AI your goal and budget. It picks every part for you.