AMD and a Swiss company software maker Pixelux set up a draft Open Physics. Its aim is to expand the current physics opensource engine Bullet Physics of OpenCL programming interface. Competitive engine supports Nvidia PhysX GPU accelerated 3D games physical effects only through the framework CUDA on GeForce cards. Physics is open while the work in accordance with OpenCL processors and graphics systems all producers, both on computers and consoles.

Additionally, AMD plans to link the Bullet Physics engine interface DirectCompute API supports DirectX 11 Thanks to this unit Compute Shader graphics chips can be used with calculation modules for DirectX 10 and newer. Examples of this type of construction are the cards from the Radeon HD 2000 series and GeForce cards from the eighth generation.

By contrast, cooperating with AMD Pixelux company announced that in future plans to integrate its middleware, Digital Molecular Matter (DMM) engine physical Bullet. DMM is used inter alia in the game Star Wars: The Force Unleashed. In addition to the game with Bullet Physics engine also benefit schemes for modeling three-dimensional objects, such as Blender and Cinema 4D.

Open Physics is a second approach AMD (namely ATI), which aims to transfer the calculation of physical effects in games on graphics chip. Already in 2006, ATI has worked with independent software vendor then Havok and even demonstrated the physical effects on the GPU accelerated Radeon cards. In late 2007, Havok was acquired by Intel, which had suspended work on the GPU assisted by Havok FX engine. Several months later, the company's competitive Nvidia bought Ageia (PhysX).

At this time, the largest market share of engines has PhysX physics (26.8 percent), and behind him followed by Havok (22.7 percent) and Bullet Physics (10.3 percent). While in June 2008, subject to the company Havok Intel to conclude some kind of alliance with AMD, it still reserves its NVIDIA PhysX only for GeForce cards. It came even to the fact that the 186 version driver for Nvidia disables the calculation of the physical on the GPU, once detected in the computer graphics chip different manufacturer. This prevents concurrent performing computations Radeon graphics card and the physical effects of the GeForce connected.