Algorithms minimizing mean flow time: schedule-length properties

E. G. Coffman, Ravi Sethi

Research output: Contribution to journalArticle

19 Citations (Scopus)

Abstract

The mean flow time of a schedule provides a measure of the average time that a task spends within a computer system, and also the average number of unfinished tasks in the system. The mean flow time of a schedule is defined to be the sum of the finishing times of all tasks in the system. On a system of identical processors O(nlog n) algorithms exist for determining minimal mean flow time schedules for n independent tasks. In general, there will be a large class C of schedules, of widely differing lengths, that all minimize mean flow time. The problem of finding the shortest schedule in C is NP-complete. We give heuristics that find schedules in C that are no more than 25% longer than the shortest schedule in C. The advantage of a short schedule is that processor utilization is high.

Original languageEnglish (US)
Pages (from-to)1-14
Number of pages14
JournalActa Informatica
Volume6
Issue number1
DOIs
StatePublished - Mar 1976
Externally publishedYes

Fingerprint

Computer systems

ASJC Scopus subject areas

  • Information Systems

Cite this

Algorithms minimizing mean flow time : schedule-length properties. / Coffman, E. G.; Sethi, Ravi.

In: Acta Informatica, Vol. 6, No. 1, 03.1976, p. 1-14.

Research output: Contribution to journalArticle

@article{c2521c606c4c43959121d113e04ce11c,
title = "Algorithms minimizing mean flow time: schedule-length properties",
abstract = "The mean flow time of a schedule provides a measure of the average time that a task spends within a computer system, and also the average number of unfinished tasks in the system. The mean flow time of a schedule is defined to be the sum of the finishing times of all tasks in the system. On a system of identical processors O(nlog n) algorithms exist for determining minimal mean flow time schedules for n independent tasks. In general, there will be a large class C of schedules, of widely differing lengths, that all minimize mean flow time. The problem of finding the shortest schedule in C is NP-complete. We give heuristics that find schedules in C that are no more than 25{\%} longer than the shortest schedule in C. The advantage of a short schedule is that processor utilization is high.",
author = "Coffman, {E. G.} and Ravi Sethi",
year = "1976",
month = "3",
doi = "10.1007/BF00263740",
language = "English (US)",
volume = "6",
pages = "1--14",
journal = "Acta Informatica",
issn = "0001-5903",
publisher = "Springer New York",
number = "1",

}

TY - JOUR

T1 - Algorithms minimizing mean flow time

T2 - schedule-length properties

AU - Coffman, E. G.

AU - Sethi, Ravi

PY - 1976/3

Y1 - 1976/3

N2 - The mean flow time of a schedule provides a measure of the average time that a task spends within a computer system, and also the average number of unfinished tasks in the system. The mean flow time of a schedule is defined to be the sum of the finishing times of all tasks in the system. On a system of identical processors O(nlog n) algorithms exist for determining minimal mean flow time schedules for n independent tasks. In general, there will be a large class C of schedules, of widely differing lengths, that all minimize mean flow time. The problem of finding the shortest schedule in C is NP-complete. We give heuristics that find schedules in C that are no more than 25% longer than the shortest schedule in C. The advantage of a short schedule is that processor utilization is high.

AB - The mean flow time of a schedule provides a measure of the average time that a task spends within a computer system, and also the average number of unfinished tasks in the system. The mean flow time of a schedule is defined to be the sum of the finishing times of all tasks in the system. On a system of identical processors O(nlog n) algorithms exist for determining minimal mean flow time schedules for n independent tasks. In general, there will be a large class C of schedules, of widely differing lengths, that all minimize mean flow time. The problem of finding the shortest schedule in C is NP-complete. We give heuristics that find schedules in C that are no more than 25% longer than the shortest schedule in C. The advantage of a short schedule is that processor utilization is high.

UR - http://www.scopus.com/inward/record.url?scp=0016873384&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0016873384&partnerID=8YFLogxK

U2 - 10.1007/BF00263740

DO - 10.1007/BF00263740

M3 - Article

AN - SCOPUS:0016873384

VL - 6

SP - 1

EP - 14

JO - Acta Informatica

JF - Acta Informatica

SN - 0001-5903

IS - 1

ER -