Performability Analysis of Mesh-Based NoCs Using Markov Reward Model

Abstract

Technology scaling makes it possible to implement systems with hundreds of processing cores, and thousands in the future. The communication in such systems is enabled by Networks-on-Chips (NoCs). A downside of technology scaling is the increased susceptibility to failures in NoC resources. Ensuring reliable operation despite such failures degrades NoC performance and may even invalidate the performance benefits expected from scaling. Thus, it is not enough to analyze performance and reliability in isolation, as usually done. Instead, we suggest treating both aspects together using the concept of performability and its analysis with Markov reward models. Our methodology is exemplified for mesh NoCs and transient faults but can be transferred to other topologies and fault models. We investigate how performability develops with scaling towards larger NoCs and explore the limits of scaling by determining the break-even failure rates under which scaling can achieve net performance increase.