Accelerating MATLAB Performance: 1001 tips to speed up MATLAB programs (Chapters 6 and 7: explicit parallelization)

Abstract

Chapter 6 and 7: explicit parallelization in MATLAB using MathWorks Toolboxes and other thirdparty libraries. In these chapters we make a detailed overview of Parallel Computing Toolbox, Distributed Computing Server and MATLAB-integrated Jacket GPU library by AccelerEyes. Besides that, we discuss MATLAB’s parallelization through MEX interfacing such parallel processing libraries as AccelerEyes ArrayFire, NVIDIA CUDA, GPUmat, OpenMP and POSIX threads. The efficiency of parallelization techniques are evaluated on a concrete running example: the Matched Filter algorithm. This is a computationally expensive algorithm which admits a large degree of parallelism yet not straightforwardly achievable.

Publication
CRC Press