基于InfiniBand的并行绘制系统中的数据压缩  

Data compression in parallel rendering system based on InfiniBand

作　　者：彭世雄[1] 季玉龙[1] 吴志红[1] PENG Shixiong;JI Yulong;WU Zhihong(College of Computer,Sichuan University,Chengdu Sichuan 610065,China)

机构地区：[1]四川大学计算机学院,四川成都610065

出　　处：《太赫兹科学与电子信息学报》2017年第2期268-272,共5页Journal of Terahertz Science and Electronic Information Technology

摘　　要：为提高并行绘制系统中的帧率,从数据压缩的角度出发,搭建了基于InfiniBand的并行绘制系统。首先对并行绘制系统中数据进行分类,根据当前帧的数据量大小设置临界值,对达到临界值的几何数据进行无损压缩。本文研究并对比了LZ77,LZW,LZO三种数据压缩算法对系统的影响。实验表明,LZ77,LZW和LZO算法均对系统的帧率有所提高,而LZO算法对系统的帧率提高最明显且稳定性最好,使系统帧率提高近60%。无损数据压缩对提高并行绘制系统性能具有很明显的效果。From the perspective of data compression,parallel rendering system based on InfiniBand is set up in order to improve the frame rate.Data classification in parallel rendering system is performed;and the thresholds are set according to the data of the current frame;the lossless compression is applied to the geometric data which reach up to the critical value.The effects of three types of data compression algorithms,LZ77,Lempel-Ziv-Welch(LZW),Lempel-Ziv-Oberhumer(LZO),are researched and compared.The experiments results show that,LZ77,LZW and LZO algorithms can improve the frame rate of system;LZO algorithm increases the frame rate of system most significantly with the best stability,and the system frame rate is increased by nearly60%.Lossless data compression can improve the performance of parallel rendering system significantly.

关 键 词：并行绘制系统 数据压缩 InfiniBand架构 LZO算法 LZW算法 LZ77算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]