Optimization of processor devices based on the maximum indicators of self-correction

Abstract

It is proposed to characterize the state of digital processing devices’ outputs with a system of probabilities, where for any binary processing results the probability of getting the same real result is set, if it is theoretically possible. Since the outputs of some devices in their compositions are the inputs of other devices, than such a probability system is universal, that allows taking into account faults of hardware means. For the original inputs the models of probabilistic information converters (data distortion on the initial inputs system) are used, on the outputs of which the indicated probabilities system exists, allowing for distortion of the input data only. Heuristic algorithm for optimal encoding (as per the maximal possible self-correction) of digital machines (as the models of processing devices) is proposed for their structural synthesis. The obtained results enable reaching optimize results on the design stage, that are directed to increase veracity of computer systems processor devices’ functioning.