A distributed divide and conquer skeleton

Authors:
Juan R. González;Coromoto León;Casiano Rodríguez
Affiliations:
Dpto. Estadítica, I.O. y Computación, Universidad de La Laguna, La Laguna, Tenerife, Spain;Dpto. Estadítica, I.O. y Computación, Universidad de La Laguna, La Laguna, Tenerife, Spain;Dpto. Estadítica, I.O. y Computación, Universidad de La Laguna, La Laguna, Tenerife, Spain
Venue:
PARA'04 Proceedings of the 7th international conference on Applied Parallel Computing: state of the Art in Scientific Computing
Year:
2004

Citing 5
Cited 1

MPI: The Complete Reference

MPI: The Complete Reference
Generating Parallel Programs from the Wavefront Design Pattern

IPDPS '02 Proceedings of the 16th International Parallel and Distributed Processing Symposium
A Skeleton Library

Euro-Par '02 Proceedings of the 8th International Euro-Par Conference on Parallel Processing
MALLBA: A Library of Skeletons for Combinatorial Optimisation (Research Note)

Euro-Par '02 Proceedings of the 8th International Euro-Par Conference on Parallel Processing
Satin: Efficient Parallel Divide-and-Conquer in Java

Euro-Par '00 Proceedings from the 6th International Euro-Par Conference on Parallel Processing

MALLBA: a software library to design efficient optimisation algorithms

International Journal of Innovative Computing and Applications

Quantified Score

Hi-index	0.00

Visualization

Abstract

The MaLLBa library provides skeletons to solve combinatorial optimization problems. Its main objective is to simplify the implementation of algorithms based on some commonly used techniques such as Branch and Bound, Dynamic Programming and Divide and Conquer. This work is focused on the MaLLBa::DnC skeleton, which solves problems that fit in the Divide and Conquer paradigm. The user has to provide functions particularized to the problem he wants to solve. Given that functions the skeleton encapsulates all remaining work and allows the problem to be solved either in a sequential or parallel way. In this work we will present a new mpi asynchronous peer-processor implementation of the MaLLBa::DnC skeleton where all processors are peers and behave the same way (except during the initialization phase) and where decisions are taken based only on local information. Results on a Linux cluster of PC for matrix and huge integer multiplication are presented.