МЕТОД РАСПАРАЛЛЕЛИВАНИЯ ПО БАЗОВЫМ МАКРООПЕРАЦИЯМ ДЛЯ ОБРАБОТКИ БОЛЬШИХ РАЗРЕЖЕННЫХ НЕСТРУКТУРИРОВАННЫХ МАТРИЦ НА РВС

Abstract

Analysis calculating large sparse unstructured matrices (LSU-matrices) methods and toolsfor cluster computing systems with a traditional architecture showed that for most tasks of processingmatrices with about 105 rows, performance compose reduced 5-7 times compared to thepeak performance. Meanwhile peak performance of computing systems is mainly estimated by theLINPAC test, which involves the execution of matrix operations. The main goal of the work is toincrease the efficiency processing LSU-matrices, for this purpose advisable to use reconfigurablecomputing systems (RSC) based on FPGAs as the main type of computing tools. For efficient processingLSU-matrices on RCS, a set method and approaches previously described in the papersare used, such as the structural organization of calculations, the format for representingLSU-matrices "row of lines", the paradigm of discrete-event organization of data flows, the method of parallelization by iterations. The article considers the method of parallelization by basicmacro-operations for solving the problem of processing LSU-matrices on RCS, which impliesobtaining a constant computational efficiency, regardless of the portrait of processed LSUmatrices.Using developed methods for processing LSU-matrices for reconfigurable computingsystems makes it possible to provide computational efficiency at the level of 50%, which is severaltimes superior to traditional parallelization methods