GGPRF: AGPU-Based Generic Parallel Rendering Framework

Juan ZUO, Feng JIANG, Zhi-jie LIN, Hai-feng HUANG, Qin-fei ZHANG

Abstract


With the development of CUDA architecture and GPU technologies, parallel rendering technologys have been developed rapidly. There are already some parallel rendering frameworks, such as Chromium, MPK, VR Juggler, etc., but these systems have obvious flaws.In this paper,GGPRF: a GPU-Based Generic Parallel Rendering Framework is proposed. The framework contains three roles, Master, Server, and Slave, which are responsible for rendering task drivers, cluster management, and rendering. Server manages the entire parallel rendering system's organizational structure through the NTTree (Network Topological Tree) data structure. GGPRFhas improved the shortcomings of the existing universal parallel rendering framework, which has the advantages of universal, extensibility and ease of use. In this paper, the design of the frame is described in detail and the experiment is tested.

Keywords


GPU cluster, Parallel rendering framework


DOI
10.12783/dtcse/cmee2017/19966

Refbacks

  • There are currently no refbacks.