DLSS (Deep Learning Super Sampling)
NVIDIA's AI-powered upscaling technology that renders games at lower resolution and uses neural networks to reconstruct near-native image quality at much higher frame rates.
What is DLSS?
DLSS (Deep Learning Super Sampling) is NVIDIA's AI upscaling technology. It renders a game at a lower internal resolution — say 1080p — and uses a trained neural network running on dedicated Tensor Cores to reconstruct a higher-resolution output — like 4K. The result is near-native image quality at dramatically higher frame rates.
DLSS 4, available on RTX 50 series GPUs, adds Multi Frame Generation: the AI generates up to three additional frames for every traditionally rendered frame, effectively multiplying your frame rate by up to 4×.
DLSS versions compared
DLSS 2.x — AI upscaling only. Renders at lower resolution, reconstructs to target resolution. Available on all RTX GPUs (20/30/40/50 series).
DLSS 3 — Added Frame Generation (1 AI-generated frame per rendered frame). RTX 40 series and newer only.
DLSS 3.5 — Added Ray Reconstruction, which uses AI to denoise ray-traced lighting more efficiently than traditional denoising. All RTX GPUs for SR, 40+ for FG.
DLSS 4 — Multi Frame Generation (up to 3 AI frames per rendered frame). RTX 50 series only. The biggest leap in the technology's history.
Frequently Asked Questions
Does DLSS work on AMD GPUs?
No. DLSS requires NVIDIA RTX hardware with Tensor Cores. AMD GPUs use FSR (FidelityFX Super Resolution) instead, which offers comparable but generally slightly lower quality upscaling.
Does DLSS add input lag?
DLSS Super Resolution adds negligible latency (<1ms). Frame Generation does add latency because frames are synthesized after rendering. NVIDIA Reflex mitigates this — in practice, the perceived smoothness outweighs the added lag for most players.