Multidimensional Exploration of Software Implementationsfor DSP Algorithms

  • Authors:
  • Eckart Zitzler;Jürgen Teich;Shuvra S. Bhattacharyya

  • Affiliations:
  • Computer Engineering and Networks Laboratory (TIK), ETH Zurich, Switzerland;Computer Engineering (DATE), University of Paderborn, Germany;Department of Electrical and Computer Engineering, and Institute for Advanced Computer Studies, University of Maryland, College Park, USA

  • Venue:
  • Journal of VLSI Signal Processing Systems - Special issue on recent advances in the design and implementation of signal processing systems
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

When implementing software for programmable digital signal processors(PDSPs), the design space is defined by a complex range of constraintsand optimization objectives. Three implementation metrics that arecrucial in many PDSP applications are the program memoryrequirement (code size), data memory requirement,and execution time. This paper addresses the problem of exploring the3-dimensional space of trade-offs that is defined by these crucialmetrics. Given a software library for a target PDSP, and a dataflow-basedblock diagram specification of a DSP application in terms of this library,our objective in this paper is to compute a full range ofPareto-optimal solutions. For solving this multi-objectiveoptimizationproblem, an evolutionary algorithm based approach is applied. Weillustrate our techniques by analyzing the trade-off fronts of apractical application for a number of well-known, commercialPDSPs.