Nvidia has taken a significant step towards community innovation by finally opening up the GPU simulation kernel source code for its PhysX SDK.
While the PhysX SDK has been released under the BSD-3 license since 2018, the crucial component—the GPU code—has remained locked until now. With this release, developers can now access over 500 CUDA kernels and drive features like rigid body dynamics, fluid simulation, and deformable objects.
PhysX is Nvidia's real-time physics simulation engine, originally developed by NovodeX. It is widely used in gaming and robotics to simulate lifelike movement and interactions. The newly released GPU code brings transparency to one of the most advanced CUDA-based simulation engines, enabling experimentation, education, and new projects in GPU-accelerated computing.
Nvidia has also open-sourced the GPU compute shader implementation for its Flow SDK, a real-time sparse grid-based fluid and gas simulation toolkit. Flow supports PhysX by handling fluid phenomena in high-performance applications, making it a crucial tool for developers who value realism in visual and simulation-based products.
Nvidia's senior director of simulation technologies, Adam Moravanszky, emphasized the significance of this release, stating, "We hope this release will be a valuable resource for learning, experimentation, and development across the community." His statement reinforces Nvidia's commitment to promoting innovation via transparency and community engagement.
Interestingly, PhysX has lately reappeared due to compatibility concerns with Nvidia's GeForce RTX 50 Series GPUs, which no longer support earlier PhysX-dependent titles like Batman: Arkham Asylum and Mirror's Edge. While Nvidia has yet to rectify the issue officially, this open-source release offers modders the tools they need to construct their remedies potentially.
This release marks a shift in Nvidia's approach to community participation. By making PhysX's GPU simulation kernels and Flow's shader code accessible on GitHub under the BSD-3 license, Nvidia enables academics, educators, and hobbyists to push the limits of real-time simulation.
The possibilities expanded significantly, whether for academic exploration or a retro game model.