摘要
为了解决高性能科学计算程序设计当中存在的开发难度大,开发周期长以及对开发人员要求高等问题,人们已经开始将软件构件技术引入该领域。由美国能源部、犹他州大学、印弟安那大学等联合提出的CCA便是研究高性能科学计算构件技术的项目之一。本文主要介绍了CCA以及CCA框架下的语言互操作工具—Babel的相关情况,并且通过NPB基准测试程序IS详细描述了Babel的使用,分析了基于Babel的程序设计对程序性能的影响。初步实验表明Babel能够有效解决语言的互操作问题,在面向科学计算的构件程序设计环境中能够发挥关键作用。
In recent years, component technology has been imported to high performance scientific computing field in order to solve problems of complexity, reuse, and interoperability for high performance scientific computing software. The CCA, brought forward by various national laboratories U.S. Department of Energy and collaborating academic institutions, is a project to develop a common framework that will allow scientists to reuse computer codes on today's high-performance computing systems, and to allow interoperability between those codes. This paper primarily introduces CCA and Babel, a language interoperability tool in CCA framework. We also demonstrate the component based programming using Babel analyze the performance costs associated with Babel. Our preliminary experiment indicates that Babel do well in language interoperability and it suits the high-performance scientific computing.
出处
《计算机科学》
CSCD
北大核心
2006年第12期235-237,267,共4页
Computer Science