On on-chip intelligence paradigms

Abstract

Hardware Intelligence is an attractive concept that has the potential to make electronic, industrial and reliability-oriented applications efficiently optimized. This work investigates different frameworks and analyzes their challenges. Hardware automation faces different challenges like scalability, overhead, testability as well as goal definition. For better classification, hardware systems are mapped to different layers with different granularities of reconfiguration, monitoring and decision making. An Intelligent Hardware Stack (IHW) is proposed to differentiate these layers showing which techniques can be mapped to which layer. IHW stack presented in this paper showcase different substrates essential for the process like monitoring, reconfiguration and decision making.