基于Open VG云电子书系统的多级优化框架设计

Multiple-level Optimization Framework Design of Cloud Electronic Book System Based on Open VG

摘要:

针对电子书应用存在的文件格式、性能效率低下和图像失真等问题,设计了一种应用于云电子书系统的多级优化框架,优化框架主要体现在如下三个方面；第一,对向量图形类库的性能进行描述,并提出了一种优化算法,减少了类库的时间复杂度;第二,在嵌入式GPU上并行进行坐标系统的计算;利用GPU在并行计算方面的优势,云电子书在向量图形类库方面获取了显著的性能提升;第三,云电子书将文件转化功能转嫁给Hadoop云平台,节省了移动设备的能量消耗和计算时间。同时为了对Hadoop调度过程中的数据位置问题进行优化,将位置感知调度器运用到提出的系统;实验结果表明:云电子书系统与最初的Open VG类库相比,性能提升了约70%,而且云电子书系统与连续服务器平台相比,计算时间减小了约60%。

Abstract:

Concerning the problem of file format, low performance efficiency and image distortion in the application of e-books, a multiple-level optimization framework of cloud e-book is proposed, in which the optimization framework is mainly reflected in the following three aspects. Firstly, the performance of the vector graphics library is described, and an optimization algorithm is proposed to reduce the time complexity of the class library. Secondly, parallel coordinate system is used to calculate on an embedded GPU. Cloud electronic books in the vector graphics library has obtained a significant performance improvement by making use of the advantage of GPU in parallel computing. Thirdly, file conversion function of cloud e-books is passed on to the Hadoop cloud platform, by which cloud save energy consumption and computation time of the mobile equipment. And in order to optimize the Hadoop scheduler in the process of data location problem, location aware scheduler is applied to the proposed system. The experimental results show that, compared with the original VG Open class library, the performance of the cloud e-book system is improved by about 70%, and the computing time is reduced by about 60% compared with the continuous server platform.