Bounds for Multifit Scheduling on Uniform Processors

We examine the nonpreemptive assignment of $N$ independent tasks to a system of $M$ uniform processors with the objective of reducing the makespan, or the time required from the start of execution until all tasks are completed. Since the problem of finding a minimal makespan has been shown to be NP-...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:SIAM J. Comput.; (United States) 1983-02, Vol.12 (1), p.60-70
Hauptverfasser: Friesen, D. K., Langston, M. A.
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:We examine the nonpreemptive assignment of $N$ independent tasks to a system of $M$ uniform processors with the objective of reducing the makespan, or the time required from the start of execution until all tasks are completed. Since the problem of finding a minimal makespan has been shown to be NP-hard, and hence unlikely to permit an efficient solution procedure, near-optimal heuristic algorithms have been studied. It is known that LPT (longest processing time first) schedules are within twice the length of the optimum. We analyze a variation of the MULTIFIT algorithm derived from bin packing, and prove that its worst-case performance bound is within 1.4 of the optimum.
ISSN:0097-5397
1095-7111
DOI:10.1137/0212004