| Popis: |
Reducing the time for multiplying matrices of large dimensions is an urgent task, since it finds application in many application areas. One of the ways to solve it is distributed computing based on the use of client-server architectures. The article proposes to expand the functional capabilities of an earlier developed subsystem for distributed design of VLSI, also built using a client-server architecture (CSA). Based on the formulated fundamental principles of dividing matrices into blocks depending on a given number of processors, a partitioning algorithm has been developed and an estimate of its complexity has been given. The results of simulation modeling of multiplication of matrices of large dimensions are presented, which confirmed the possibility of expanding the KSA functional for distributed design of VLSI. |
