Fundamentals of Self-Sensing Processing Systems Slides

Abstract: High-performance systems are challenged by the stringent computational, reliability and availability requirements of emerging cloud-native applications. Unfortunately, efficiency gains through scaling alone have slowed, even as susceptibility to variation-induced system failures have increased, thus necessitating further innovations in energy efficient and reliable processor and system design. This tutorial addresses the following key aspects: how do sources of variations impact design margins and system reliability?; how do self-monitoring systems use sensors to measure ambient environment?; how is environment adaptation actuated in high-volume production systems using a combination of power-delivery and clocking techniques?; what design and analysis techniques can mitigate transient soft errors and hard errors due to transistor aging and interconnect failures?