Forget Nvidia: ZLUDA 6 Lets AMD Cards Run PhysX Games at Triple FPS
When corporate giants throw away brilliant technology, open-source developers pick up the trash and turn it into absolute gold. The legendary ZLUDA project is back, proving that hardware artificial walls are just a hilarious illusion.
Developer Andrzej Janik released the open-source ZLUDA 6 translation layer on GitHub, allowing unmodified CUDA-designed applications to run directly on AMD graphics cards. This project, written in the Rust programming language, has survived a chaotic history of being funded and then abruptly abandoned by both Intel and AMD.
The absolute highlight of this release is experimental support for Nvidia's proprietary PhysX physics engine. Gamers testing old classics like Mafia II on AMD hardware reported a massive performance jump from an unplayable 26 frames per second to a buttery-smooth 80 frames per second once ZLUDA took over the heavy lifting.
But it is not just about retro gaming. ZLUDA 6 introduces crucial texture support that enables the 3D modeling suite Blender to run its CUDA backend on AMD chips. Ironically, tests show that some Radeon cards actually render faster using this translated CUDA path than when running on AMD's native HIP rendering framework. The update also improves Windows compatibility and enhances support for running massive machine learning models like PyTorch.
What started as a corporate-funded experiment has evolved into a beautiful rebellion, showing that artificial software locks are no match for a single determined programmer. Corporate giants spend billions trying to lock users into proprietary ecosystems, only to watch their expensive garden walls get completely demolished by free code on the internet.
Source: GitHub
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.