An investigation of various algorithms suitable for 3D compute on heterogeneous platform

An investigation of various algorithms suitable for 3D compute on heterogeneous platform

This research explores the usage of heterogeneous computing, which is the current trend in the computing industry, for the 3D compute based on the collaborative algorithm. As a part of literature review the areas related to the heterogeneous computing is investigated, which includes OpenCL, General Purpose Graphical Processing Unit [ GPGPU ], collaborative algorithm. The novelty of this research work is the application of collaborative algorithms, which have been used in other areas to 3D compute. This exploration of 3D compute algorithm was part of the exercise in order to find the suitable 3D compute algorithm for the monocular image to binocular image conversation engine. Here the two algorithms, i.e., intelligent water drops and zombie crowd algorithm were implemented using OpenCL to run on heterogeneous platform. The performance of both the algorithm was measured on the experimental setup rig. Based on the above research exercise, it can be inferred that, the collaborative algorithm can indeed be suitable for 3D compute and it can be also noted that each algorithm is suitable for particular 3D compute. KEYWORDS: Heterogeneous Computing, GPGPU, OpenCL, Collaborative Algorithm

Swamy H. Raghavendra

Computer Science