AI-rewritten: This is a summary of an article from Tom’s Hardware, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

---

Over recent weeks, modders Dashdogy and Uncle Burrito have unlocked features in Nvidia’s DLSS technology for older graphics cards. Dashdogy successfully enabled Multi Frame Generation (MFG) on RTX 4000 series hardware, which previously displayed a black screen when attempting to run Blackwell instructions. He achieved this by spoofing calls to Nvidia’s Streamline and Neural Graphics Framework (NGX) software in memory, tricking the GPU into believing it could generate more frames than its architecture normally allowed.

Uncle Burrito independently ported DLSS 5 Neural Rendering to Ada Lovelace hardware after reverse-engineering the underlying DLL code with the help of an AI model. Nvidia has since clarified that they intended to support this feature on RTX 4000 series GPUs, suggesting their initial implementation was incomplete rather than a hard limit. These foundational mods now serve as a basis for ports extending back to 2018’s Turing architecture and even AMD hardware.

Dashdogy further improved his MFG mod by analyzing kernel dumps from an RTX 6000 Pro Blackwell GPU using OpenAI’s Daybreak Blue AI model. He discovered that the original 40-series implementation compressed generated frames around a single interval, causing sluggishness and poor responsiveness. His fix spread these frames out at separate intervals to match the Blackwell reference behavior. This allows users to select higher frame generation multipliers without the previous performance issues, though latency can still vary depending on the specific game settings.