Computer Engineering and Applications ›› 2009, Vol. 45 ›› Issue (3): 198-201.DOI: 10.3778/j.issn.1002-8331.2009.03.059

• 工程与应用 • Previous Articles     Next Articles

Memory optimization for static scheduling of Synchronous Data Flow programs

LIU Guo-xin,TAN Guo-qiang,HE Ye-ping   

  1. National Engineering Research Center of Fundamental Software,Institute of Software,CAS,Beijing 100190,China
  • Received:2008-07-07 Revised:2008-09-27 Online:2009-01-21 Published:2009-01-21
  • Contact: LIU Guo-xin

同步数据流模型调度序列的空间优化

刘国鑫,谭国强,贺也平   

  1. 中国科学院 软件研究所 基础软件国家工程研究中心,北京 100190
  • 通讯作者: 刘国鑫

Abstract: A method for reducing memory requirements of embedded DSP systems is proposed.The method basing on SDF(Synchronous Data Flow) performs all of the scheduling at compile time by means of periodic schedules.Because the SAS(Single Appearance Schedules) scheduling sequence of optimized code memory is not always available for those SDF graph with feedback loop.This method designs a Non-SAS heuristic algorithm for optimizing memory of feedback loops,and then introduces a hierarchical optimized memory way,which combines the SAS scheduling sequence with Non-SAS,to solve the optimized memory problem for SDF graph with loops.Experimental results prove the validity of the proposed method.

Key words: embedded system, Synchronous Data Flow(SDF), scheduling sequence, memory optimization, feedback loop

摘要: 提出了一种嵌入式DSP系统的存储优化方法。该方法基于同步数据流模型SDF(Synchronous Data Flow)。针对其他优化算法不适用于存在反馈环的同步数据流模型的问题,该方法为反馈环的空间优化设计实现了启发式的调度算法,并提出了将SAS(Single Appearance Schedules)和Non-SAS类型调度序列相结合的层次化的空间优化方案,为同步数据流模型调度序列的空间优化提供一个通用的解决方案。实验结果证实了该方案的有效性。

关键词: 嵌入式系统, 同步数据流, 调度序列, 存储优化, 反馈环