论文部分内容阅读
基于分布存储的工作站机群系统由于具有很高的性能价格比,越来越受到高性能计算领域的重视。用户在使用机群系统时,迫切地需要一种性能监视、评估和预测的工具。THPT系统就是为满足这一需要而设计开发的。 THPT系统由两个部分组成:网络并行计算环境的实时管理和性能监测子系统MMS以及并行计算性能分析和预测子系统PAPS。本文主要对PAPS子系统进行论述。
Workstation cluster system based on distributed storage has attracted more and more attention in high performance computing due to its high performance-price ratio. When using fleet systems, users urgently need a tool for performance monitoring, assessment and forecasting. THPT system is designed to meet this need and development. The THPT system consists of two parts: a real-time management and performance monitoring subsystem MMS for network parallel computing environment and a parallel computing performance analysis and prediction subsystem PAPS. This article mainly discusses the PAPS subsystem.