Benchmarking using the Community Atmospheric Model
Benchmarking using the Community Atmospheric Model
复制标题
使用社区大气模型进行基准测试
DOI：
--
复制
发表时间：
2006
期刊：
影响因子：
--
通讯作者：
P. Worley
中科院分区：
文献类型：
--
作者：
P. Worley
社区大气模式（Community Atmospheric Model，CAM）是为天气和气候研究社区开发的全球大气模式。CAM也是社区气候系统模式（CCSM）的大气组成部分。作为一个社区模型，CAM在不同的体系结构上高效运行，并且容易移植到新的平台上并在新的平台上进行优化是很重要的。当前版本的CAM包含许多性能可移植性功能-编译时或运行时参数，可用于针对给定平台、问题或处理器数量优化性能。大量的调优选项会使基准测试成为一项艰巨的任务。本文介绍了这些选项，以及如何优化管理，使其可行的早期系统的评估。本文还描述了所选平台对不同调优选项的一些性能敏感性。
The Community Atmospheric Model (CAM) is a global atmosphere model developed for the weather and climate research communities. CAM also serves as the atmospheric component of the Community Climate System Model (CCSM). As a community model, it is important that CAM run efficiently on different architectures, and that it be easily ported to and op- timized on new platforms. The current version of CAM contains a number of performance portability features - compile-time or runtime parameters that can be used to optimize performance for a given platform, problem or processor count. The large number of tuning options can make benchmarking an arduous task. The paper describes these options and how optimization is managed to make it feasible for evaluation of early systems. The paper also describes some of the performance sensitivities of selected platforms to the different tuning options.