A static multiprocessor scheduling algorithm for arbitrary directed task graphs in uncertain environments
Publikation: Bidrag til bog/antologi/rapport › Konferencebidrag i proceedings › Forskning › fagfællebedømt
The objective of a static scheduling algorithm is to minimize the overall execution time of the program, represented by a directed task graph, by assigning the nodes to the processors. However, sometimes it is very difficult to estimate the execution time of several parts of a program and the communication delays under different circumstances. In this paper, an uncertain intelligent scheduling algorithm based on an expected value model and a genetic algorithm is presented to solve the multiprocessor scheduling problem in which the computation time and the communication time are given by stochastic variables. In simulation examples, it shows that the algorithm performs better than other algorithms in uncertain environments.
Originalsprog | Engelsk |
---|---|
Titel | Algorithms and Architectures for Parallel Processing - 8th International Conference, ICA3PP 2008, Proceedings |
Antal sider | 12 |
Forlag | Springer |
Publikationsdato | 2008 |
Sider | 18-29 |
ISBN (Trykt) | 9783540695004 |
DOI | |
Status | Udgivet - 2008 |
Eksternt udgivet | Ja |
Begivenhed | 8th International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2008 - , Cypern Varighed: 9 jun. 2008 → 11 jun. 2008 |
Konference
Konference | 8th International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2008 |
---|---|
Land | Cypern |
Periode | 09/06/2008 → 11/06/2008 |
Navn | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|
Vol/bind | 5022 LNCS |
ISSN | 0302-9743 |
ID: 362749356