为改善并行应用程序运行效率,对性能数据的采集和分析两个方面进行了研究,分别提出多线程同时采集多种类型的性能数据,以及性能分析应该面向Web的两种思想,并提出了一种对负载均衡程度进行量化的算法,实现了面向Web的性能评价工具。它可以在应用程序运行的同时收集系统负载信息和记录应用程序的行为。用户可以藉此在Internet上任何一台计算机上通过浏览器对性能数据进行多视图、多角度的观察分析,找出性能瓶颈,最终提高应用程序的性能。 In order to improve the running efficiency of parallel applications, this paper studied two aspects of performance data collection and analysis, such as multi-thread simultaneous collection of multiple types of performance data and the two ideas that performance analysis should be oriented to Web, and proposed a This algorithm quantifies the degree of load balancing and implements Web-oriented performance evaluation tools. It collects system load information and records application behavior while the application is running. Users can take this on any computer on the Internet through the browser on the performance data multi-view, multi-angle observation and analysis to identify performance bottlenecks, and ultimately improve the performance of applications.