MikhbarMIKHBAR
Artificial Intelligence

Official DLSS 5 Hits Unsupported Games With Massive Gains

Modders have successfully extracted the official DLSS 5 release from NBA 2K27 and deployed it on unsupported games, achieving dramatic performance improvements over previous versions.

Official DLSS 5 Hits Unsupported Games With Massive Gains

Official DLSS 5 Extraction and Early Testing

Following the release of NBA 2K27, modders quickly extracted a new version of DLSS 5 and began running it on unsupported games. Initial performance comparisons between older and newer Neural Rendering (NR) models showed immediate speed improvements, with some testing indicating the new framework runs significantly faster than its predecessor.

While many experimenters were still evaluating leaked and first-release builds, community members such as Renan Maniero extracted the 310.9.16.0 DLSS NR build. This allowed other developers and enthusiasts to integrate it into unsupported software and evaluate performance differentials firsthand.

Performance Gains Up to 90 Percent

Detailed benchmarking shared on social media by Gaming Optimized demonstrated massive enhancements when testing the official release. Running native DLSS 5 in titles like The Witcher 3 revealed performance gains reaching up to 90% compared to the older iteration, though the exact magnitude of the boost varies depending on the specific game tested.

These comparisons were streamlined because the latest DLL file contains both the old and new models as selectable presets. Preset A corresponds to the legacy model, while Preset B activates the newer iteration, allowing direct side-by-side performance checks without constant file swapping.

Efficiency Improvements and Reduced VRAM Footprint

Operating a full Neural Rendering model to alter a game's rendered image is inherently resource-intensive, much like how large language models demand substantial memory resources. To address hardware constraints, Nvidia optimized its approach since consumer graphics cards feature limited, non-upgradable VRAM.

To counter hardware limitations, Nvidia engineered the underlying AI model to be 24% smaller. According to official company statements, this optimization results in 20% lower VRAM consumption at 4K resolution and enables the system to process neural rendering workloads up to two times faster than before.

Modder Adaptations for Older RTX Hardware

Experiments involving DLSS 5 typically rely on custom injection techniques into the graphics pipeline using tools like RenoDX or OptiScaler. These utilities give users the flexibility to tweak dynamic library settings and experiment with different visual and performance outcomes.

Official support paths from Nvidia focus primarily on bringing DLSS 5 to newer hardware generations, such as RTX 40-series GPUs, with no official announcements targeting RTX 30- or RTX 20-series cards. Despite this official limitation, the modding community successfully modified the runtime environment to support older RTX 20-series hardware, as documented in the OptiScaler GitHub repository.

Sources

  • Tom's HardwareOfficial DLSS 5 running on unsupported games delivers up to 90% better performance than the leaked version

Continue chronologically

Related entity coverage