PhysX Processor – Dedicated Physics Processor
I have not dug too far into the documentation, but AGEIA has released their PhysX SDK which allows game developers to take advantage of the PhysX Processor for more realistic physics simulations in games. Computer games have been pushing the envelop of computing since the beginning of the electronics age and many of the advances that make game play better eventually translate into better computing for non-game applications. The PhysX Processor is a dedicated processor for handling only the physics calculations – generally done in software – of a game (or other applications that leverage the SDK and processor). This offloading and specialization is part of the trend seen in graphics cards that have become ever more powerful and removing the graphics processing tasks from the general purpose CPU in a machine. AGEIA says that the “PhysX Processor completes the triangle of game function, graphics and interactive real-time environments from physics computing, balancing the load of these processing tasks and enabling incredible realism in tomorrow’s games.”
The trend to find areas of opportunity for optimization – such as Graphics Processors (GPUs) and now the PhysX Processor – most likely will not just benefit the game developer and consumer. Not having dug deep into the SDK documentation, I can only imagine that there are opportunities to leverage the processing power of a dedicated physics processor for calculations of heavy duty stock market problems. Of course, the same could be said about leveraging the awesome power of graphics processors. The wonderful thing about general purpose processors is that you can throw anything at them and they will do what you need. Specialized processors like the GPU and physics processor have a much narrower view of the world and are effective on certain types of problems. But when you find a problem or calculation that can benefit from these processors, they can have a major impact on speed. And who wouldn’t want to solve some of their most complex problems faster?