期刊文献+

一种面向结点的加速比模型

The concept of node-oriented speedup on SMP cluster
在线阅读 下载PDF
导出
摘要 SMP机群系统是一种新型的系统结构。由于依赖于处理器的传统加速比模型不再适用于SMP机群系统,文中提出了面向结点的加速比模型,它将传统的串行与并行的对比关系扩展为个体与整体之间的对比关系。通过对基于处理器的传统加速比概念的研究,推导出了基于SMP机群系统的面向结点加速比的计算公式。面向结点加速比模型充分考虑了实际运行时间的可测性,反映出了系统在使用了多个结点之后性能的改变。事实上,传统的加速比模型是面向结点加速比模型的一个特例。 SMP cluster is a kind of novel parallel architecture. However , the traditional speedup model based on processor cannot be applied t o SMP cluster. In this paper the node-oriented speedup model is studie d. This model extends the relationship between the serial and the para llel program to that between the individual and the integration. We di scuss the original speedup model, and conclude the formulation of the node-oriented speedup on SMP cluster. The node-oriented speedup is mea surable and reflects the performance improvement after more nodes are included. In fact, the traditional speedup model is a special case of the node-oriented one.
出处 《计算机工程与设计》 CSCD 北大核心 2000年第5期16-21,共6页 Computer Engineering and Design
关键词 SMP机群系统 结点 工作负载 加速比模型 SMP cluster speedup node the degree of parallelism workload
  • 相关文献

参考文献9

  • 1Thomas E. Anderson et al.A Case for Now[J]. IEEE Macro,1995,(2):54-64.
  • 2Jean Papadopoulo. Evolution in High-End Parallel Computing: SMP, Clusters, and MPP Revisited[N]. Http://www-frec.bull.fr/OSBU2_0/wp_hep.html
  • 3Xian-He Sun and John L. Gustafson, Toward a better parallel performance metric[J].Parallel Computing,1991, 17(12):1093-1109.
  • 4Xian-He Sun and Lionel M. Ni, Another View on Parallel Speedup[C].The Proceedings of Supercomputing '90 ( Amsterdam ), New York: ACM Press, 1991.
  • 5Kai Hwang, Advanced Computer Architecture[M].McGraw-Hill, Inc., 1993.
  • 6Sartaj Sahni and Venkat Thanvantr, Performance Metrics: Keeping the Focus on Runtime[J]. IEEE Parallel & Distributed Technology,1996.43-56.
  • 7Xian-He Sun and Jianping Zhu, Performance Prediction-A Case Study Using a Scalable Shared-Virtual-Memory Machine[J].IEEE Parallel & Dis tributed Technology,1996.36-48.
  • 8Daniel Nussbaum and Anant Agarwal, Scalability of Parallel Machines[J].Communication of the ACM, 1991, 34(3):57-61.
  • 9Vishwani D. Agrawal and Srimat T. Chakradhar, Performance Estimation in a Massively Parallel System[C].In Proceedings of Supercomputing'90 ( Amsterdam ), New York: ACM Press, 1991.

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部