摘要
随着芯片开发技术在嵌入式系统中的发展,单位面积上晶体管数量不再无限制的增长,为追求嵌入式系统整体性能的提高,代码的执行效率逐渐被重视起来。多核处理器在嵌入式系统中应用越来越广泛,要进一步大规模地提高计算效率,更好的发挥出多核处理的优势,必须引入并行开发技术。文章采用openmp3.1作为并行框架,支持在嵌入式应用中进行c/c++语言开发,通过生成识别openmp指令的GCC4.8.1编译器,来解决openmp应用代码的编译问题,将openmp框架适配到嵌入式系统中,以可配置代码库的方式融入到操作系统应用层,深入分析openmp框架代码,提出了四个方面的优化策略,最后以FTD2000为目标机,多种算法作为测试用例,取得了优异的并行效果。
With the development of chip development technology in embedded systems,the number of transistors per unit area is no longer growing indefinitely,and in order to pursuc the improvement of the overall performance of cmbedded systems,the execution efficicncy of code has gradually been valued.Multi-core processors are more and more widely used in embedded systems,and in order to further improve computing efficiency on a large scale and btter play the advantages of multi-core processing,parallel development technology must be introduced.This paper uses OpenMP3.1 as a parallel framework to support C/C++language development in embedded applications.By generating a GCC4.8.1 compiler that recognizes openmp instruc-tions,it solves the compilation problem of openmp application code,adapts the openmp frame-work to the cmbedded system,integrates it into the operating system application layer in the form of a configurable code basc,deeply analyzes the openmp framcwork code,proposcs four optimization strategies,and finally takes the FTD2000 as the targct machine and a varicty of al-gorithms as test cases,and achieves excellent parallel effects..
作者
林卓
陈聪
王宏伟
LIN Zhuo;CHEN Chong;WANG Hongwei(Xi'an Aeronautics Computing Technique Research Institute,AVIC,Xi'an 710000,China)
出处
《长江信息通信》
2024年第10期120-122,共3页
Changjiang Information & Communications
基金
国家重点研发计划资助项目(2017YFB1301100)。