View article

For the last decades, industries from the safety-critical domain have been using Commercial Off-The-Shelf (COTS) architectures despite their inherent runtime variability. To guarantee hard real-time constraints in such systems, designers massively relied on resource over-provisioning and disabling the features responsible for runtime variability. The recent shift to multi-core architectures in the embedded COTS market worsened the runtime variability problem as contention on shared hardware resources brought new variability sources. Additionally, hiding this variability in additional safety margins as performed in the past will offset most if not all the multi-core performance gains. To enable the use of multi-cores in this domain, it has become essential to finely characterize at system level the application workload, as well as the possible contention on shared hardware resources. In this paper, we introduce measurement techniques based on a set of dedicated stressing benchmarks and architecture hardware monitors to characterize (1) the architecture, by identifying the shared hardware resources and their associated contention mechanisms. (2) the application, by identifying which shared hardware resources it is sensitive to. Such information would guide us toward identifying which applications can run smoothly together without endangering individual worst-case execution times.