Nvidia’s new compact DLSS model delivers 90% speed boost in games lacking official support
Enthusiasts extracted Nvidia’s latest neural rendering model from NBA 2K27, showing massive speed gains over previous versions in titles without official support.
Key points
- Modders extracted DLSS version 310.9.16.0 from NBA 2K27 to test on unsupported titles.
- Performance gains of up to 90% were observed in The Witcher 3 compared to older models.
- The new AI model is 24% smaller and uses 20% less VRAM at 4K resolution.
Enthusiasts have successfully extracted and deployed Nvidia’s latest DLSS 5 neural rendering technology in games that do not officially support it, achieving significant performance improvements. According to Tom’s Hardware, modders obtained the new software version from the recent update for NBA 2K27. They are now testing this unofficial implementation across various titles. The results indicate substantial speed increases compared to previous iterations of the technology. This activity highlights the ongoing experimentation within the gaming community regarding graphics optimisation tools.
Renan Maniero extracted the specific build version 310.9.16.0, allowing other modders to utilise it in unsupported environments. Gaming Optimized on X compared the new neural rendering model against the older one. They reported performance improvements of up to 90% in The Witcher 3 when running DLSS 5 natively. This comparison was facilitated because the latest version still includes the older model as a preset. The new DLL contains two neural rendering presets: Preset A represents the old model, while Preset B is the newer model.
View the original post on X
View the original post on X
What we know
Running a full neural rendering model to modify game images is resource-intensive, similar to how large language models require substantial RAM. To improve performance, Nvidia made the AI model 24% smaller. This reduction allows it to use 20% less VRAM at 4K resolution. Nvidia states that the new model processes its neural rendering workload up to twice as fast. Typically, improving an AI model involves increasing its parameters and size. However, consumer GPUs have limited, non-upgradable VRAM. Therefore, Nvidia focused on efficiency rather than adding a larger model.
Modders injected DLSS into the graphics pipeline using RenoDX or OptiScaler. These tools allow users to tweak DLL settings for different results. The modding community also modified the runtime to support RTX 20-series GPUs. The OptiScaler GitHub README notes that this modification enables older hardware to run the technology. Nvidia plans to release DLSS 5 to RTX 40-series GPUs officially. There are no reports of it coming to RTX 30- or 20-series cards officially, but modders have bypassed these restrictions.
Defender's checklist
- Verify the source of any graphics driver or DLL modifications before installation.
- Check system stability after applying unofficial rendering presets or runtime modifications.
- Monitor VRAM usage closely when testing new neural rendering models at high resolutions.
- Ensure backups of original game files are created before injecting external code into pipelines.
The background in plain language
Neural rendering uses artificial intelligence to enhance image quality and frame rates. It works by analysing lower-resolution images and reconstructing them into higher-quality visuals. This process requires significant computing power, particularly video memory. When teams optimise these models, they aim to reduce the computational load while maintaining visual fidelity. Users should understand that unofficial implementations may not be tested for stability or security. Teams usually evaluate the trade-off between performance gains and potential system instability. Check whether the modified software introduces unexpected behaviour in your specific hardware configuration.
Background: Large language models
Large language models are zero-shot reasoners only in appearance. They predict the next word by matching patterns in training data, not by applying logical rules. This means they can solve novel problems if similar examples exist in their data, but they fail when true novel logic is required.
Read the full guide: Are Large Language Models Zero-Shot Reasoners? The Reality Check
What to do and how to stay safe: Nvidia
- Review the official support list for DLSS 5 before attempting unofficial installations on unsupported hardware.
- Monitor system temperatures and power consumption when running resource-intensive neural rendering tasks.
- Keep a record of original driver versions to facilitate rollback if instability occurs.
- Wait for official vendor updates before relying on modded versions for critical gaming sessions.
General security guidance from the TechWire Beat newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
Can I use DLSS 5 on my RTX 20-series GPU?
Modders have modified the runtime to support RTX 20-series GPUs, but Nvidia has not officially released DLSS 5 for this hardware.
How much faster is the new DLSS 5 model compared to the previous one?
Tests showed performance gains of up to 90% in The Witcher 3 and up to 40% in other comparisons against the previous neural rendering model.
What version of DLSS was extracted from NBA 2K27?
Renan Maniero extracted DLSS build version 310.9.16.0 from the NBA 2K27 update for testing purposes.
