NVIDIA has announced improvements to its frame generation technology on GeForce RTX 40 GPUs, enabling better Deep Learning Super Sampling (DLSS) performance and lower memory consumption.
The company’s Frame Generation technique was first introduced with DLSS 3.0 for the RTX 40 series GPUs. Since then, NVIDIA has refined the technology, enhancing upscaling quality and performance. The new update will be applied to both the RTX 40 and RTX 50 series GPUs.
The upgraded frame generation technique will utilize a better AI model, resulting in lower latency and reduced VRAM requirements. This improvement will benefit users with 8 GB GPUs, as well as those with larger memory capacities.
A shared screenshot by NVIDIA’s Jacob Freeman demonstrates the benefits of the new DLSS version, showcasing improved performance and reduced VRAM consumption. The update introduces Enhanced Ray Reconstruction, Super Resolution, and Deep Learning Anti-Aliasing (DLAA) for RTX 40 series GPUs.
While Multi Frame Generation will not be available on RTX 40 GPUs, users can still enjoy a boost in performance with the new frame generation technique. However, analysis of input latency is still needed to fully understand its impact. NVIDIA claims that DLSS 4.0 reduces latency, but a comparison between DLSS 2.0 and 4.0 suggests no significant difference.
Source: https://wccftech.com/nvidia-dlss-frame-generation-rtx-40-gpus-new-ai-model-faster-lower-vram-requirement