NVIDIA DLSS Frame Generation expanded via new GPU mods

Modding developments are expanding the capabilities of NVIDIA’s DLSS Frame Generation technology across multiple GPU architectures. On Ada Lovelace-based GeForce RTX 40 series cards, projects such as Universal RTXMFG are bypassing the standard NVIDIA DLSS-G pipeline. These experimental modifications allow for frame multipliers of 5X to 6X, moving beyond the officially supported limits.

Simultaneously, developers have successfully brought DLSS Frame Generation to the older Ampere-based GeForce RTX 30 series. Unlike previous mods that substituted NVIDIA’s technology with AMD’s FSR 3, the “dlssg_for_sm86” mod implements a dedicated path for Ampere hardware. This allows for 2X, 3X, and 4X frame generation while remaining closer to NVIDIA’s original implementation.

Technical testing on an RTX 3080 Ti indicates that these processes incur specific performance costs. In 4K resolution, the frame generation pipeline requires approximately 4.8 milliseconds of computation time and an additional memory overhead of roughly 770 MiB. These hardware requirements may impact performance in memory-intensive games, particularly on models with limited VRAM.