UDC 004.75
PARALLEL MULTIPLICATION OF LARGE-DIMENSIONAL MATRICES ON MANY GIVEN PROCESSORS
V. M. Glushan, Dr. Sc. (Tech.), professor of the CAD Department of IKTIB SFU, Taganrog, Russia;
orcid.org/0000-0001-5822-9295, e-mail:This email address is being protected from spambots. You need JavaScript enabled to view it.
O. I. Krasyuk, Ogetto LLC, web developer, Taganrog, Russia;
orcid.org/0000-0003-3153-2651, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
A. Yu. Lozovoy, Ph.D. (Ped.), associate Professor of the Department of I&A IUPET SFU, Taganrog, Russia;
orcid.org/0000-0002-6701-3098, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
Matrix multiplication is applied to many practical problems. Small dimension matrix multi-plication Matrix multiplication is applied to many practical problems. Small dimension matrix multiplication causes no particular difficulties. They arise when you have to multiply matrices with thousands and millions of rows and columns. The object of the article is to study the validity of the previously developed by the authors distributed subsystem of client-server architecture for VLSI design engineering for the multiplication of highdimensional matrices. Studies of this subsystem have shown multiple acceleration of VLSI design time, so there was a natural desire to expand its functionality to solve other problems requiring parallelization. The article proposes a method to divide initial matrices into blocks for multiplying them on a given number of processors. Simulation modeling of the process of multiplying matrices with large dimensions and comparison of the results obtained with known solutions are carried out. The comparison showed the ability of the subsystem to parallelize the process of matrix multiplication. In this case, the achieved acceleration of multiplication process turns out to be no worse than in the known solutions, and in some cases it turns out to be even higher.
Key words: partitioning of matrices into blocks, block multiplication, parallelization, client-server architecture.