The Study And Optimization Of The Architecture Of Distributed Rendering System

Today with the fast developing of information technology the Computer Graphics is becoming widely used in all fields and brings realistic visual experience to users. But the computing ability of the current systems becomes the bottleneck for CG rendering. Distributed rendering is one of the most effective solutions to address this problem and solve the bottleneck.In this article we design and implement a triple layered distributed rendering system architecture which consists of a center, multiple masters and multiple rendering clusters. It supports cross-platform and multi-render rendering. The Center handles requests of users and cooperates with masters to finish jobs. This architecture partially implements the MVC model to make the system more practical and expandable. This system effectively solves the bottleneck for CG rendering and brings up the CG productivity through distributed computing.The rendering pool is the aggregate of N computing nodes. In this article we present the concept of level of services of rendering pool to provide differentiated services to clients and studies the relative algorithms.In this article we also study the segmentation model of jobs and present the adaptive job segmentation model based on BP neutral networks. This model can be trained with historical sample of data and learn the optimized segmentation of jobs so as to reduce the rendering time.In this article we make full test of the system with focus on following issues: the rise of performance compared to normal single computer rendering and the performance and usability of the job segmentation model. The results of the tests show that the system effectively brings up the CG rendering efficiency and the job segmentation model is practicable.