摘要
现有化学计量学软件普遍采用的偏最小二乘(PLS)算法均以单线程方式计算,建模速度缓慢,给应用带来较大的不便。随着多核处理器的普及,采用多线程并行计算技术可显著提高算法执行速度。本文将多线程并行计算技术引入化学计量学软件开发,提出并实现了PLS算法的并行化计算。利用标准数据集进行了性能对比实验,结果表明在四核计算机中多线程并行计算比单线程计算有大约3.1倍的速度提升。
PLS which is commonly used in most Chemometrics software always calculates in a single-threaded way, which brings low modeling speed and inconvenience to apply.With the popularization of the multi-core processor, multithread parallel computing technology could significantly boost the execution speed.In this paper, the multithreaded parallel computing technology was introduced into the development of Chemometrics software.A kind of parallelizing calculation of PLS was proposed and actualized.The contrast experiment with two datasets demonstrated that multithread parallel computation on a quad-core computer achieved a speed of 3.1 times faster than that of a single-thread computation.
出处
《微计算机信息》
2010年第9期208-210,共3页
Control & Automation
基金
国家高技术产业化发展计划项目
基金申请人:罗国安
杨辉华
项目名称:中药注射剂质量标准化技术支撑体系建设示范项目
基金颁发部门:国家发展和改革委员会(发改办高技[2007]2490号)
关键词
化学计量学软件
偏最小二乘
多线程
并行计算
Chemometrics software
partial least square
multithread
parallel computing
