Optimizing communication is a key issue in compiling dataparallel languages for distributed memory architectures. We examine here the cyclic distribution, and we derive symbolic expressions for communication sets under the only assumption that the initial parallel loop is defined by afline expressions of the indices. This technique relys on unimodular changes of basis. Analysis of the properties of communications leads to a tiling of the local memory addresses that provides maximal message vectorization.
CITATION STYLE
Germain, C., & Delaplace, F. (1995). Automatic vectorization of communications for data-parallel programs. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 966, pp. 429–440). Springer Verlag. https://doi.org/10.1007/bfb0020483
Mendeley helps you to discover research relevant for your work.