DLSS 5 Runs on Unsupported Nvidia GPUs Thanks to Clever App Hack

A free app called NeuralScreen has been developed to run Nvidia's DLSS 5 neural renderer on GeForce RTX 30-series cards, which were not officially supported by Nvidia. According to the project's GitHub page, DLSS 5 can process games, videos, and photos in real time, with intensity controls and a before-and-after slider.

The app intercepts an Nvidia API call inside its own process and reports a Blackwell architecture to the runtime, which is required for DLSS 5. Modders had already demonstrated how to install DLSS 5 on almost any RTX GPU, but applying it across Windows makes the compatibility gap harder to ignore.

While NeuralScreen has no HDR support, caps the neural pass at 2560 x 1440, adds 40-60ms of latency, and can't overlay true fullscreen games. It also depends on a leaked Nvidia DLL rather than a finished public release. DLSS 5 is Nvidia's most computationally demanding model, and experimental implementations have cut performance roughly in half even on newer cards.