Resource Estimation and Task Scheduling for Multithreaded Reconfigurable Architectures

  • Authors:
  • Arvind Sudarsanam;Mayur Srinivasan;Sethuraman Panchanathan

  • Affiliations:
  • Arizona State University;Arizona State University;Arizona State University

  • Venue:
  • ICPADS '04 Proceedings of the Parallel and Distributed Systems, Tenth International Conference
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

Reconfigurable computing is an emerging paradigm of research that offers cost-effective solutions for computationally intensive applications through hardware reuse. There is agrowing need in this domain for techniques to exploit parallelisminherent in the target application and to schedule the parallelizedapplication. This paper proposes a method to estimate theoptimal number of resources through critical path analysis whilekeeping resource utilization near optimal. We also propose anovel algorithm to optimally schedule the parallel threads ofexecution in linear time. Our algorithm is based on the ideaof enhanced Partial Critical Path (ePCP) and handles memorylatencies and reconfiguration overheads. Results obtained showthe effectiveness of our approach over other critical path based methods.