Open-source ZLUDA v6 brings PhysX performance boost to AMD Radeon GPUs

The open-source ZLUDA v6 library has introduced support for Nvidia's 32-bit PhysX CUDA ecosystem on non-Nvidia hardware, delivering a significant performance increase for AMD Radeon GPUs running legacy PhysX titles.

According to the project's latest blog post (via Wccftech), the update allows graphics cards without built-in PhysX functionality to process 32-bit PhysX effects natively. Testing ZLUDA v6 in Mafia II on an AMD Radeon platform showed a performance increase from 26.2FPS to 80.2FPS when using the library. This integration also restores complex visual effects, such as debris and flames, on unsupported hardware. Those who want to try it out can check the project's GitHub and the Quickstart guide.

The developer notes that the tool is currently experiencing a halt in commercial support, meaning future development will proceed at a slower pace as a weekend project. Current optimisation limits mean users may encounter glitches in fluid simulation and difficulty loading the software into Steam games. However, the roadmap shows developers aim to fine-tune Windows support, improve ROCm library error reporting, and deliver better machine learning capabilities alongside newly added texture support.

For those who weren't in the PC gaming space at the time, PhysX was Nvidia's old solution to bringing better physics to games, including more detailed debris, cloth, particles and more. Nowadays, PhysX has been largely replaced in new games by updated game engine tools, like Chaos in Unreal Engine, or the widely used Havok physics engine.

KitGuru says: Seeing community projects like ZLUDA breathe new life into legacy proprietary features on competing hardware is impressive, even if development is slowing down.

Become a Patron!