D.R. Pikulin
Research Subjects
D.R. Pikulin has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Boris Kostrov
Ryazan State Radio Engineering University, Ryazan, Russian Federation
1 shared publication D.V. Grigorenko
1 shared publication- E.V. Ruchkina1 shared publication
- V.N. Ruchkin1 shared publication
Latest Publications
- 1 January 2025
- Published by Publishing house "Radiotekhnika" in Dynamics of Complex Systems - XXI century
- Vol. 19 (2) , 30-39
Hardware-software co-design of intelligent multi-core cyber-physical systems-on-chip with universality, scalability and energy-efficiency properties
The possibilities of intelligent multi-core Cyber-Physical System on Chip (CPSoN) for providing high-quality properties of universality, scalability and energy-efficiency based on modern VLSI K1879ВМ8Я and software Neuro Matrix are shown. A methodology for intelligent management of neuroprocessor resources is proposed for the purpose of joint design of hardware and software (Co-Design) based on Intellectual Multiprocessing and set-theoretical equivalence of a group of algorithms, a group of cores and a group of robots on a chip. Mathematical formalization of the analysis of various structures is carried out for the purpose of an optimal choice allowing us to design software-reconfigurable multi-core neuroprocessor architecture for processing large data flows. Modeling multi-criteria control for choosing the best structure according to various technical characteristics is carried out with emphasis on significant speed up the work of neural network tools, accelerators and servers.
- 1 January 2025
- Published by Publishing house "Radiotekhnika" in Dynamics of Complex Systems - XXI century
- Vol. 19 (2) , 30-39
Hardware-software co-design of intelligent multi-core cyber-physical systems-on-chip with universality, scalability and energy-efficiency properties
The possibilities of intelligent multi-core Cyber-Physical System on Chip (CPSoN) for providing high-quality properties of universality, scalability and energy-efficiency based on modern VLSI K1879ВМ8Я and software Neuro Matrix are shown. A methodology for intelligent management of neuroprocessor resources is proposed for the purpose of joint design of hardware and software (Co-Design) based on Intellectual Multiprocessing and set-theoretical equivalence of a group of algorithms, a group of cores and a group of robots on a chip. Mathematical formalization of the analysis of various structures is carried out for the purpose of an optimal choice allowing us to design software-reconfigurable multi-core neuroprocessor architecture for processing large data flows. Modeling multi-criteria control for choosing the best structure according to various technical characteristics is carried out with emphasis on significant speed up the work of neural network tools, accelerators and servers.