ALGORITHMIC SKELETONS FOR GENERAL SPARSE MATRICES ON MULTI-CORE PROCESSORS

摘要

We develop an extension of the Muenster Skeleton Library Muesli by a distributed data structure for general sparse matrices. The data structure supports data parallel algorithmic skeletons such as fold, map, and zip. Our implementation is highly flexible, object-oriented and makes use of the C++ template mechanism. As a result, the storable data type as well as the compression and distribution scheme can easily be changed and even be substituted by a user-defined one. As a unique feature, our implementation not only supports multi-processor architectures, but also efficiently makes use of current multi-core processors.