ABSTRACT
As seen in Chapter 10, “Image Processing,” the compute shader can be used to implement new types of algorithms on the GPU in a very intuitive way. However, these new algorithms are certainly not limited to the image processing domain. Many different algorithms lend themselves to being executed on a massively parallel architecture, and this chapter aims to introduce several such applications. In general, as the GPU has continued to become more powerful, there has been a general trend to move more of the workload off of the CPU and onto the GPU. With this trend comes the guideline to minimize the interaction needed between the CPU and GPU during any given frame. These sample programs were designed with this concept in mind-minimal input is required by the CPU to both execute a simulation and to render it.
