Open Physics

1. Open Physics Project.

   Open Physics Project, English Open Physics. It is a physical simulation computing platform built by AMD for its 3A platform. It is based on OpenCL and is completed by the joint calculation of CPU+GPU. The so-called “open” means that the physics engines participating in this project are all open source and distributed to game developers for free. The Open Physics Project is not exclusive to the AMD platform. Any hardware that supports OpenCL and Directcompute can run. This is in sharp contrast to Nvidia’s closed attitude towards PhysX. Because any closed technology has no way out.

   After Intel acquired Havok and Nvidia acquired AGEIA, a dramatic scene occurred in the physics engine industry:

   1) Intel did not have independent GPU production capacity, so it tried its best Promote the physics simulation engine to be calculated by a multi-threaded CPU. The Havok engine is optimized for CPU multi-threading. The original GPU computing part of the Havok engine is shelved.
   2) Since Nvidia does not have CPU production capacity, it strongly promotes the physics simulation engine to be calculated by the GPU, that is, the graphics card. And the PhysX engine itself is not very suitable for CPU computing.
   Actually technology, neither CPU nor GPU can support complex physics simulation calculations alone. Only the joint calculation of CPU and GPU can improve the calculation efficiency to the greatest extent. AMD has the full platform design and production capabilities of CPU, chipset, and GPU, but it does not have its own physical simulation engine and is always under the control of others.
   Fortunately, OpenCL appeared at this time, laying the foundation for AMD’s open physics project.
   physical plan is also “currently only GPU + CPU general accelerated physics solution submitted to Khronos Group”.
Reference: https://baike.baidu.com/item/%E5%BC%80% E6%94%BE%E7%89%A9%E7%90%86%E8%AE%A1%E5%88%92
2. Deformable bodies. (deformable objects)
  As hardware capabilities improve and more-efficient algorithms are developed, physics engines are beginning to provide support for deformable bodies. DMM< /span> is an excellent example of such an engine.
  DMM is < strong>pixelux The company’s products have been used in the film industry.

  1) Because Intel does not have independent GPU production capacity, it strongly promotes the physical simulation engine to be calculated by multi-threaded CPU. The Havok engine is optimized for CPU multi-threading. The original GPU computing part of the Havok engine Was shelved.

   2) Since Nvidia does not have CPU production capacity, it strongly promotes the physics simulation engine to be calculated by the GPU, that is, the graphics card. And the PhysX engine itself is not very suitable for CPU computing.

   Actually technology, neither CPU nor GPU can support complex physics simulation calculations alone. Only the joint calculation of CPU and GPU can improve the calculation efficiency to the greatest extent. AMD has the full platform design and production capabilities of CPU, chipset, and GPU, but it does not have its own physical simulation engine and is always under the control of others.

   Fortunately, OpenCL appeared at this time, laying the foundation for AMD’s open physics project.

  Physics plan is also “currently the only GPU submitted to Khronos Group + CPU general acceleration physical solution”.

Reference: https://baike.baidu.com/item/%E5%BC%80%E6%94%BE%E7%89%A9%E7%90 %86%E8%AE%A1%E5%88%92

2, Deformable bodies. (deformable objects)

  As hardware capabilities improve and more-efficient algorithms are developed, physics engines are beginning to provide support for deformable bodies. DMM is an excellent example of such an engine.

  DMM is pixelux The company’s products have been used in the film industry.

Leave a Comment

Your email address will not be published.