集群计算技术及其在石油工业中的应用

Cluster computing technique and its application to the petroleum industry

中国石化石油勘探开发研究院南京石油物探研究所,南京210014

Institute of Geophysical Prospecting, SINOPEC, Nanjing 210014

对于高性能计算不断增长的需求和计算技术的持续发展,促进了计算机集群技术的发展和应用。计算机集群系统具有系统建设、维护和升级成本低,可扩展性好,易维护,易升级,计算性能强以及通用和跨平台的并行计算软件开发环境(PVM和MPI等)等优点。本文概述了计算机集群技术的概念、技术基础、并行软件开发环境和Beowulf集群;论述了石油工业对高性能计算的需求;介绍了集群技术在石油勘探中的应用现状;指出了集群技术将大大降低石油勘探开发成本,促进新技术的应用和推广。

The increasing demands for high performance computing (HPC) and the rapid development of computing technique, lead to advances and applications of cluster computing technique. Cluster computing has driven down costs in the construction, maintenaince and upgrade of HPC systems. Cluster systems are easy to maintain and upgrade, and have good scalability and high performance. In this paper, the authors outline the basic concepts and technical fundamentals of cluster computing, Beowulf cluster, and software development environments of parallel computing (MPI and PVM), which are independent of hardware plateforms. Need of the petroleum industry for HPC is discussed with the example of seismic exploration. Investigation shows that clusters already begin their applications and will become the most important HPC plateform in the petroleum industry. Preliminary results of performance benchmark for PC cluster in comparison with multiprocessor superserver are presented.