CURRENT DIRECTIONS OF DEVELOPMENT AND FEATURES OF COMPUTER ARCHITECTURE

Abstract

The article considers modern directions of development and features of the architecture of neurocomputers. This study is devoted to the analysis of issues related to the organization and functioning of modern computers and their main components. The main components are considered and the components of the structure and functioning of modern computers of the von Neumann type are analyzed. The structure of computer memory is analyzed, and the organization and procedure of interaction between its levels are studied, in particular, the features of segmental organization of memory and problems and ways to solve them regarding ensuring protection are considered. The main principles of organization of computer buses, operating devices and other functioning systems are substantiated. The leading trends in the architecture of modern processors have been investigated, the principles of construction and architectural features of parallel computer systems have been established, new approaches to the study of computer architecture and the features of their construction have been proposed. The conceptual apparatus for neurocomputers has been proposed and improved, and their architectural features have been determined. It is substantiated that modern computer architectures are based on the principles of the von Neumann architecture, which allows for the joint storage of commands and data in a single address space of memory. The significant structural components of the research system, which are necessary for the effective functioning and fulfillment of the tasks, are analyzed. The peculiarities of the functioning of components of modern architecture are investigated and the main requirements for it are determined. The necessity and feasibility of the requirements determined by the specifics of the functioning of information systems, which ensure high speed, reliability and universality of use, are substantiated. The requirements for the operation of high-performance technical means, considered according to the criterion of compliance with productivity and resource consumption, can be considered a method of their research and development of new logical systems for organizing activities and productivity, as well as improving existing principles of organizing computer calculations. It has been established that the most effective way to improve functioning, considering this problem through the prism of the above-mentioned features, is parallel processing of actions, mediated by the simultaneous execution of programs or their constituent components, procedures, subroutines, on independent devices. The structure of a modular nanocomputer has been improved, ensuring the execution of specified operations and realizing a number of advantages of multitasking. The ways of enhancing the functional capabilities of a nanocomputer designed to represent and process both real and complex input data, as well as when processing them in a ring of integers, are determined by neural network algorithms.