Search

Search Results

• DISCRETE-EVENT METHOD COMPUTATIONS ORGANIZING FOR PROCESSING LARGE SPARSE UNSTRUCTURED MATRIXES ON RCS

  А.v. Podoprigora

  189-197

  2021-10-05

  Abstract ▼

  Increasing models complexity objects and processes study, in different sphere of science and technology, set up plenty issues to necessary to use high-performance computing systems. Arrays matrix processing by cluster multiprocessor computing systems in conjunction special methods aimed at organizing parallel computations, basically obtain computing performance system is quite high. However, that computational efficiency is not observed for all types of matrices. Matrix structure be in a position contain large amount of insignificant elements, large dimension and unstructured portrait. Calculation execute for described kind of matrices on cluster multiprocessor computing system couldn't achieve close peak performance. Considering that processing methods leave out the complex structure of the matrix being processed. As a result, the performance of the system is significantly reduced. The development of cluster MCS methods doesn't allow for full ensure high performance for class of problems processing of large sparse unstructured matrices. Rigid architecture of processor commutation net doesn’t take into account the peculiarities of such matrices, and lead to non-uniformity loading processor. To achieve performance close the peak for tasks large sparse unstructured matrices processing necessary to use reconfigurable compu-ting systems. RCS architecture allows adapting computation structure to the problem solved. This makes it possible to organize pipeline processing, such a way that computational resource RCS used only for informational significant operations. In addition using generally accepted methods for structural organization of high-performance computing for RCS, it is necessary to develop a format for storing and transferring large sparse unstructured matrices, to determine the principles of constructing basic matrix macro-operations and the possibility of organizing composite dis-crete-event matrix functions for solving applied problems. Сconsequently method founding laid allows organizing computations operands, which are large sparse unstructured matrices. The application this method for organizing computations can significantly increase productivity, and provide an increase in the efficiency of such a system