Please use this identifier to cite or link to this item:
Title: Robust scheduling for heterogeneous architectures
Authors: Wong, Yi Wen.
Keywords: DRNTU::Engineering::Computer science and engineering::Computer systems organization::Processor architectures
Issue Date: 2010
Abstract: In this report, the robust scheduling problem for heterogeneous architectures is de ned and discussed. This problem is compared to the classic makespan minimization prob- lem. Two di erent methods, a static heuristic and a dynamic rescheduling algorithm are proposed to solve this problem. The static heuristic is based on the minimizing the longest path through a scheduled node. The results show that the static heuristic is not only e ective in solving the robust scheduling problem but also the makespan minimization problem. To our knowledge, no heuristic in literature has been designed to address the robust scheduling problem, although studies have been done to compare the robustness of current heuristics. The dynamic rescheduling algorithm improves on any existing static list-scheduling heuris- tic by utilizing additional information obtained during runtime to improve on the robust- ness of the schedule. A number of schemes controlling the rescheduling are proposed. The results show that dynamic rescheduling is able to signi cantly improve the robustness of a static heuristic. The di erent schemes for rescheduling are also compared. These algorithms are compared against existing solutions in literature using both ran- domly generated graphs and modi ed standard benchmark graphs for the homogeneous scheduling problem.
Schools: School of Computer Engineering 
Organisations: A*STAR Institute of High Performance Computing (IHPC)
Rights: Nanyang Technological University
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:SCSE Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
  Restricted Access
4.87 MBAdobe PDFView/Open

Page view(s) 50

Updated on Jun 15, 2024


Updated on Jun 15, 2024

Google ScholarTM


Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.