NVIDIA and Havok in Bed
NVIDIA is partnering with Havok, arguably the most successful physics engine API for software development, to bring physics processing acceleration to NVIDIA GPUs. Does this spell the end for AGEIA?
It was way back in May of last year that we first took a detailed look at a small company named AGEIA. A startup with a new idea in the industry was showcasing their new hardware solution to computer-based physics calculations. I highly recommend you read my preview of AGEIA’s PPU (physics processing unit) that we published just after E3 last year.
AGEIA’s reasoning behind their new product was the increasingly complex physics being integrated into games and the CPU’s inability to parse through it all quickly enough. The problems that AGEIA faced then, and still face, are fairly daunting, with the biggest issue being one of software integration. They addressed this by giving away their custom physics API called Novodex, hoping that when the games developed using it were ready, they would build a market for AGEIA’s custom PPU hardware.
NVIDIA Partners with Havok
Today, right before the official start of GDC (game developers conference), NVIDIA and Havok are announcing a technology partnership that will allow physics processing to be done on NVIDIA GPUs using Havok’s upcoming Havok FX API.
Though the Havok FX API is still in development, Havok as a company has a very distinguished list of clients that have used their physics simulators in games previously.
You can see that many of the biggest game developers in the world have worked with Havok’s products before. This doesn’t necessarily indicate that Havok’s new Havok FX engine will be as successful, but it does bode well for NVIDIA’s partnership with them.
The new API will be able to directly access NVIDIA GPUs in order to perform physics calculations by working through DirectX. The Havok FX engine will send the data to the GPU to be worked on and the results will be passed on back through DirectX to the Havok API.
Havok FX will have support for both a single NVIDIA GPU as well as two or more GPUs running in SLI mode. An SLI configuration will allow for a user to dedicate an entire GPU to graphics and one to physics. If the Havok FX engine is not utilizing the GPU, NVIDIA claims that the SLI configuration can still be used for increased graphics performance.
Page 2 – NVIDIA’s Advantages