Fault-tolerance models for AI chips of intellectual systems

Abstract

The fault tolerance of a system is its ability to perform its functions in the presence of failures of some elements. In a fault-tolerant system, the failure of one or more elements leads to only a partial decrease in the quality of work. Fault tolerance is ensured by the use of methods and means of introducing redundancy (for example, using backup methods, parallel computing, etc.). The most rational use of hardware and software in creating fault-tolerant systems is achieved through the application of methods and tools of system diagnostics.