Authors
Rasika Amarasiri and Binh Do (Monash University)
Abstract
MATLAB is widely used in scientific applications due to its rich set of tools and efficient processing of matrices. MATLAB has introduced a parallel toolbox that allows parts of MATLAB code to execute on distributed computers achieving higher throughput. The distributed processing elements require worker licenses for each processor and a distributed computing server to manage them. This paper presents an alternative approach that utilizes the MATLAB compiler to produce parallel jobs that can be scheduled and executed using existing grid architechtures. A case study illustrating the use of this method in solving a computationally intensive finance application is also provided.