Optimum Retiming Of Multi-Dimensional Data Flow Graphs With Variable Processing Order

Ratnesh Verma, Dongming Peng

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Multidimensional Data flow graphs are used to represent multi-dimensional DSP applications for concurrent implementation on parallel hardware. Retiming is one of the techniques to achieve high speed implementation of these applications. In retiming memory issues are crucial, and processing order can tremendously affect the memory requirement for such applications. Despite its known importance, retiming with variable processing order has never been attempted. This paper introduces a novel approach of optimum retiming of acyclic MDFG with variable processing order, which not only gives the processing order but also the retiming factors with least memory requirement. The effectiveness of the technique is shown in the example section.

Original languageEnglish (US)
Title of host publication17th International Conference on Computer Applications in Industry and Engineering 2004, CAINE 2004
EditorsParimal A. Patel
PublisherInternational Society for Computers and Their Applications (ISCA)
Pages51-56
Number of pages6
ISBN (Electronic)9781618395382
StatePublished - 2004
Event17th International Conference on Computer Applications in Industry and Engineering, CAINE 2004 - Orlando, United States
Duration: Nov 17 2004Nov 19 2004

Publication series

Name17th International Conference on Computer Applications in Industry and Engineering 2004, CAINE 2004

Conference

Conference17th International Conference on Computer Applications in Industry and Engineering, CAINE 2004
Country/TerritoryUnited States
CityOrlando
Period11/17/0411/19/04

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Computer Science Applications
  • Information Systems
  • Software
  • General Engineering

Fingerprint

Dive into the research topics of 'Optimum Retiming Of Multi-Dimensional Data Flow Graphs With Variable Processing Order'. Together they form a unique fingerprint.

Cite this