Late last month, NGOHQ posted a screenshot of PhysX software apparently running on an AMD Radeon HD 3850 graphics card. Nvidia began porting PhysX to its GeForce cards using its CUDA application programming interface (API) after acquiring Ageia, but there’s no official way to run PhysX on Radeons just yet.
Many initially dismissed the screenshot as a hoax, but NGOHQ maintains that it’s actively porting the PhysX API to run on Radeons. In fact, the site claims it has received support from Nvidia to do so, and it invites readers to sign up for a beta test program.
The folks at TG Daily have contacted Nvidia to verify that story, and surprisingly, the company admitted to working with NGOHQ editor-in-chief Eran Badit. "Eran and I have been talking via email and we have invited him to join NVIDIA’s registered developer program. We are delighted at his interest in CUDA and in GPU accelerated physics using PhysX," stated Nvidia Developer Relations VP Roy Taylor. Nvidia PR chief Derek Perez chimed in, "We’ll help any and all developers are using CUDA. That includes tools…documentation…and hands on help."
AMD has its own general-purpose APIs for GPUs, and it announced last month that it was working with middleware developer Havok to bring physics to its Radeon graphics cards. Perhaps for those reasons, NGOHQ claims AMD has refused to send it a Radeon HD 4800-series card sample. A successful port might happen without AMD’s backing, though. In an interview with Bit-Tech back in April, Nvidia Chief Scientist David Kirk mentioned, "We do take every opportunity to discuss the ability to run CUDA with anyone who’s interested. It’s not exactly an open standard, but there’s really not very much that is proprietary about it. Really, it’s just C and there are these functions for distributing it."