¡¡Chinese Journal of Computers   Full Text
  TitleOn-Line Scheduling and Placement of Real-Time Tasks for Reconfigurable Computing System
  AuthorsZHOU Xue-Gong LIANG Liang HUANG Xun-Zhang PENG Cheng-Lian
  Address(Department of Computing and Information Technology, Fudan University, Shanghai 200433)
  Year2007
  IssueNo.11(1901¡ª1909)
  Abstract &
  Background
Abstract Efficient task scheduling and placement are very important for obtaining high performance in reconfigurable computing system. This paper focuses on the on-line scheduling of real-time tasks on a 2-dimensional reconfigurable device, and introduces the notion of recognition-complete for scheduling algorithms, that is the algorithm do not reject a task when a feasible schedule for this task exists. A new on-line scheduling and placement algorithm is proposed. By utilizing the temporal information of real-time tasks thoroughly, the proposed algorithm achieves recognition-complete. The simulation results show that the proposed algorithm gains a prominent improvement in scheduling performance over previous algorithms, while the run-time efficiency is preserved.

keywords reconfigurable computing; on-line scheduling; real-time; placement scheme; recognition-complete

background This paper is supported by the National Natural Science Foundation of China under grant No.60573105. The project aims at designing an operating system for reconfigurable embedded computers, and mainly researches on these fields: The implementation of hardware tasks, on-line schedule of both hardware and software tasks, management of dynamic reconfigurable resources, communication between hardware and software tasks, reduce system reconfigure time.
This paper focuses on the management of dynamic reconfigurable resources. The goal is to increase the utilization of reconfigurable resources. The first recognition-complete on-line scheduling algorithm is proposed in this paper. A simulation framework is constructed to experimentally evaluate the performance of the scheduling and placement algorithms. This provides an infrastructure for further researches.