ABSTRACT

With the increasing performance and parallelism of graphic processors and the introduction of accelerated general compute systems such as CUDA or the DirectX 11 Compute Shaders, the appeal of transferring complex and CPU hungry tasks to the GPU more compelling than ever. The latest advances in GPU technologies and APIs makes it easier than ever to implement a broad range of algorithms ranging from offline shading, to AI and physics using GPU acceleration. This section will thus cover articles that present techniques that go beyond pixels and triangles and by taking advantage of hardware acceleration. This section includes the following articles:

The first article, “Parallelization Implementation of Universal Visual Computer,” is written by Tze-Yui Ho, Ping-Man Lam, and Chi-Sing Leung. This article explores a framework for the generalization of image processing algorithms that can run a wide range of filters using a single set of shaders. The article exposes a few practical examples and compares the performance of such algorithm compared to a CPU based implementation.