Domestic CPU Performance Test Model Based on Convergence Strategy
MA Yue,CHAI An-ying,YIN Zhen-yu,LI Ming-shi,WANG Chun-xiao,LI Suo
（University of Chinese Academy of Sciences,Beijing 100049,China）（Shenyang Institute of Computing Technology,Chinese Academy of Sciences,Shenyang 110168,China）
Abstract：Testing domestic high-performance processors and fully understanding the performance of domestic processors,plays an extremely important role in exerting the core role of domestic CPU in the important industrial control field in China.Domestic CPU software library support is lacking,and various benchmark test sets can not achieve more accurate performance evaluation of domestic processors.This paper relies on the " core electronic devices,high-end generic chips and basic software " national science and technology major project,and proposes a domestic CPU performance test model based on convergence strategy.This model design the evaluation module of initial operation scale,and uses three modules to comprehensively evaluate the performance of the processor floating-point operation to improve the accuracy of the test indicators.Using the good portability of the model and the convergence of the test method,This model implements the comprehensive evaluation of various functions such as floating-point arithmetic and integer arithmetic in the two domestic processor platforms.This paper selects the Loongson 2H1000 and MPRC PKUnity-3-HD65 for model verification.The results show that the model evaluates performance more comprehensively,especially in the evaluation of floating point performance.Experimental results can objectively reflect the performance of the two processors.This model ensures the predictability of embedded processors in project applications.This work is of great significance for the localization of CNC and industrial control devices.
马跃,柴安颖,尹震宇,李明时,王春晓,李锁. 一种基于收敛策略的国产CPU性能测试模型[J]. 小型微型计算机系统, 2020, 41(1): 92-97.
MA Yue,CHAI An-ying,YIN Zhen-yu,LI Ming-shi,WANG Chun-xiao,LI Suo. Domestic CPU Performance Test Model Based on Convergence Strategy. Journal of Chinese Computer Systems, 2020, 41(1): 92-97.
2020, Vol. 41 
