基于Pandas+Matplotlib的显卡性能稳定性测试工具设计和应用  

Design and Application of the Performance Stability Testing Tool for Graphics Cards Based on Pandas and Matplotlib

作　　者：钱炎超 余泽超 吴凡 Qian Yanchao;Yu Zechao;Wu Fan(Zhejiang Supcon Technology Co.Ltd.,Hangzhou 310053)

机构地区：[1]浙江中控技术股份有限公司,杭州市310053

出　　处：《仪器仪表标准化与计量》2021年第2期17-20,共4页Instrument Standardization & Metrology

摘　　要：Windows性能计数器能实时采集、分析系统内的应用程序、服务、驱动程序等的性能数据,以此来分析系统的瓶颈、监视组件的表现,最终帮助用户进行系统的合理调配。对于WDDM版本低于2.0的操作系统,Windows性能计数器无法对显卡的性能数据进行监控。本文针对无法监控显卡性能数据情况,设计并实现一种基于Pandas+Matplotlib的显卡稳定性测试工具。本文以Python库函数为底层驱动,运用数据的获取与记录、文件的读取、图表的生成,实现了显卡性能的变化趋势记录,测试数据准确可靠,测试效率大大提升。Windows performance counter can collect and analyze the performance data of application program, service program and driver program in the system in real time, so as to analyze the bottleneck of the system, monitor the performance of the components, and finally help the user to make the reasonable deployment of the system. For Operating Systems Below Wddm version 2.0, the Windows performance counter can not monitor the performance data of the graphics card. In this paper, we design and implement a video card stability test tool based on Pandas + Matplotlib. In this paper, Python Library function as the underlying driver, the use of data acquisition and recording, file reading, Chart Generation, to achieve the change of the graphics card performance records, accurate and reliable test data, test efficiency is greatly improved.

关 键 词：GPU 性能测试 Matplotlib PANDAS 

分 类 号：TP334.7[自动化与计算机技术—计算机系统结构]