计算机科学 ›› 2012, Vol. 39 ›› Issue (7): 1-6.

• 综述 •    下一篇

基于横向局部性的多核计算模型

袁 良,张云泉   

  1. (中国科学院软件研究所并行软件与计算科学实验室 北京100190);(中国科学院计算机科学国家重点实验室 北京100190);(中国科学院研究生院 北京100049)
  • 出版日期:2018-11-16 发布日期:2018-11-16

Multi-core Parallel Computational Model Based on Horizontal Locality

  • Online:2018-11-16 Published:2018-11-16

摘要: 片内多核已成为延长摩尔定律的方式,并行算法设计、编程模型、编译器和运行时系统都需要利用计算模型进行分析。现有多核模型对线程间共享缓存等资源的竞争已有较精确的模型,但是对于线程间数据共享考虑较少。提出线程间共享缓存的横向局部性和任务共享率概念,基于此扩展串行存储层次模型RAM ( h),提出考虑任务共享率的多核并行计算模型MRAM(h)。

关键词: 多核,并行计算模型,共享缓存,横向局部性

Abstract: Almost all modern CPUs are multi-cores with shared cache on chip. A number of models have been proposed for predicting the shared cache contention, but few of them consider the influence of shared cache sharing. This paper proposed horizontal locality and task shared ratio, and then proposed a parallel computational model for multi-core architecture, which can be used by parallel algorithm, compiler, parallel programming model and runtime system.

Key words: Multi-core, Parallel computational model, Shared cache, Horizontal locality

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!