Along with unveiling a new line-up of graphics cards at CES 2025, Nvidia has also announced new software technologies that use AI-based hardware in the new GPUs. Headlining the new features is the latest iteration of Nvidia’s popular DLSS.
Dubbed DLSS 4, the technology is powered by the Tensor cores present in RTX graphics cards, and along with its previous features like outputting visuals at higher resolutions, now also has Multi Frame Generation. An upgrade to Frame Generation, the new Multi Frame Generation can generate three additional frames for every single frame rendered by the system.
Nvidia claims that it can multiply a game’s frame rate by up to eight times over traditional rendering techniques by making use of Multi Frame Generation.
DLSS 4 also features upgrades to its suite of other features, including Ray Reconstruction, Super Resolution, and DLAA, all of which will now run on transformers – architecture that powers AI models like ChatGPT, Flux and Gemini. This will be the first real-time application of the technology for graphics rendering.
Along with DLSS 4, Nvidia has also announced Reflex 2, which the company claims can synchronise the work done on CPUs and GPUs to reduce in-game latency by up to 75 percent.
