摘要
性能评测是查找嵌入式计算机系统性能瓶颈、指导设备选型、平衡相关部件、优化系统结构以及提高系统实际性能的重要方法。在介绍了机载嵌入式计算机系统性能评测的指标和基本方法之后,重点分析了使用基准测试程序的测试方法和特点,以及基准测试方法的对比测试原理、不同测试环境的构建以及主要测试过程;选取了SPEC CPU 2000基准测试程序集针对机载嵌入式环境进行裁剪和移植,对某国产机载嵌入式实时操作系统分别以不同的嵌入式操作系统、嵌入式文件系统和编译选项为变量参数进行对比测试,得出了国产操作系统与国外同类商用操作系统运行实际应用程序的性能基本相当的结论,另外也得出文件系统对计算类基准程序的结果影响较小以及编译优化后能够大幅提高嵌入式系统性能等结论。
Performance evaluation is an important method to find the performance bottleneck of embedded computer systems,guide equipment selection,balance related components,optimize the system structure and improve the system's actual performance.After the introduction of indicators and benchmarks of embedded computer systems performance evaluation,we analyze the characteristics of benchmark test methods and test procedures,build different test environments,as well as the comparative testing principle of benchmark method,and the main testing process.Then,select the SPEC CPU 2000 benchmarks set,cut and transplant them under the airborne embedded environment for comparison test on a domestic airborne embedded real-time operating system using different embedded operating systems,embedded file systems and compiler options as variable parameters.Conclusions show that the performance of a domestic real-time operating system for airborne embedded computer is roughly equal to similar foreign commercial operating system,file system has less effect on the benchmark results and compiler optimization can significantly improve the performance of embedded systems.
出处
《航空学报》
EI
CAS
CSCD
北大核心
2016年第4期1327-1335,共9页
Acta Aeronautica et Astronautica Sinica
基金
国家科技重大专项(2012ZX01041-006)
航空科学基金(2012ZC31006
2013ZC31005)
工信部民机专项基金(MJ-S-2012-05)~~