UDC 681.3: 004.031
MULTI-PIPELINE SCHEMES FOR UNITARY CODES PROCESSING IN HIGH-PERFORMANCE COMPUTING SYSTEMS
E. A. Titenko, Ph.D. in technical sciences, associate professor, associate professor, department of computer engineering, Kursk, Russia;
orcid.org/0000-0002-5659-4747 e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
E. I. Vatutin, Dr. in technical sciences, full professor, associate professor, department of computer engineer ing, SWSU, Kursk, Russia;
orcid.org/ 0000-0002-7362-7387, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
I. E. Chernetskaya, Dr. in technical sciences, associate professor, head of the department of computer engineering, SWSU, Kursk, Russia;
orcid.org/0009-0009-8254-9606), e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
D. A. Trokoz, Dr. in technical sciences, associate professor, leading developer in R7; Nizhniy Novgorod, Russia; orcid.org/ 0000-0003-0989-3928, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
M. A. Titenko, postgraduate student, SWSU, Kursk, Russia;
orcid.org/0009-0004-0819-1707, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
I. I. Puzanov, postgraduate student, SWSU, Kursk, Russia;
orcid.org/0009-0004-8193-7520, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The article considers the problem of ensuring the operation of computing systems associated with the implementation of parallel calculations over bit streams (streams of unitary codes). The aim of the work is to increase specific performance of pipeline circuits for processing unitary codes. Such codes are carriers of both numerical and symbolic information. Data dualism is the starting point while organizing multi-threaded computations based on iterative networks. The authors show that iterative networks are the basis for creat ing various pipeline schemes. They can accept and process several input streams based on multi functionali ty embedded in network cells. The comparison of obtained circuit solutions in mono- and multi-pipeline schemes has shown the time gain of the latter with insignificant increase in hardware costs.
Key words: : iterative network, starting point, multifunctionality, computational cell.