File(s) not publicly available

Meta task graph for volunteer computing

conference contribution
posted on 06.12.2017, 00:00 by Wei LiWei Li
This paper presents the Meta Task Graph (MTG) as an intermediate model of effective expressing parallelism between various problems and efficient algorithms for solving these problems for volunteer computing. One of the goals of this work is to develop effective techniques to deal with the heterogeneities of potential applications and system architectures. Another goal of MTG is to provide a tuple-matching mechanism, which supports fast synchronization of data-dependence and crash tolerance. Ongoing state collection of the underlying system is also a goal. For scheduling to be optimal, various pieces of information are needed. This feature provides full optimisation potential for the design of task assignment and scheduling for load balance to deal with dynamic environment of volunteer computing. To demonstrate MTG's applicability and flexibility, this paper presents a real-world problem and its MTG implementation on our simulation platform GNet. The future research on MTG is presented at the end of this paper.

Funding

Category 1 - Australian Competitive Grants (this includes ARC, NHMRC)

History

Parent Title

Complex 2004 : Proceedings of the 7th Asia-Pacific Complex Systems Conference, Cairns Convention Centre, Cairns, Australia, 6-10 December 2004.

Start Page

419

End Page

432

Number of Pages

14

Start Date

01/01/2004

ISBN-10

1876674962

Location

Cairns, Australia

Publisher

Central Queensland University

Place of Publication

Rockhampton

Peer Reviewed

Yes

Open Access

No

External Author Affiliations

Faculty of Informatics and Communication;

Era Eligible

Yes

Name of Conference

Asia-Pacific Conference on Complex Systems