Dynamic load balancing with MatlabMPI

Authors:
Ricolindo L. Cariño;Ioana Banicescu;Wenzhong Gao
Affiliations:
Center for Computational Sciences ERC;Center for Computational Sciences ERC;Department of Electrical and Computer Engineering, Mississippi State University, Mississippi State, MS
Venue:
ICCS'06 Proceedings of the 6th international conference on Computational Science - Volume Part II
Year:
2006

Citing 7
Cited 0

Guided self-scheduling: A practical scheduling scheme for parallel supercomputers

IEEE Transactions on Computers
Factoring: a method for scheduling parallel loops

Communications of the ACM
Dynamic Scheduling Parallel Loops with Variable Iterate Execution Times

IPDPS '02 Proceedings of the 16th International Parallel and Distributed Processing Symposium
Load Balancing Highly Irregular Computations with the Adaptive Factoring

IPDPS '02 Proceedings of the 16th International Parallel and Distributed Processing Symposium
Performance of Scheduling Scientific Applications with Adaptive Weighted Factoring

IPDPS '01 Proceedings of the 15th International Parallel & Distributed Processing Symposium
On the Scalability of Dynamic Scheduling Scientific Applications with Adaptive Weighted Factoring

Cluster Computing
A Load Balancing Tool for Distributed Parallel Loops

Cluster Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

A number of packages have recently been developed to enable the execution of MATLAB programs on parallel processors. For many applications, an integrated load balancing functionality is also necessary to realize the full benefits of parallelization. This paper describes a toolkit based on MatlabMPI to ease the parallelization and integration of load balancing into MATLAB applications that contain computationally-intensive loops with independent iterations. Modifications to existing code to incorporate the toolkit are minimal. Performance tests of two nontrivial MATLAB programs with the toolkit on a general-purpose Linux cluster indicate that significant speedups are achievable.