• Medientyp: E-Artikel; Elektronischer Konferenzbericht; Sonstige Veröffentlichung
  • Titel: Predictably and Efficiently Integrating COTS Cache Coherence in Real-Time Systems
  • Beteiligte: Hossam, Mohamed [VerfasserIn]; Hassan, Mohamed [VerfasserIn]
  • Erschienen: Schloss Dagstuhl – Leibniz-Zentrum für Informatik, 2022
  • Sprache: Englisch
  • DOI: https://doi.org/10.4230/LIPIcs.ECRTS.2022.17
  • Schlagwörter: Multi-Core ; Caches ; Shared Data ; Coherence ; Memory ; Real-Time
  • Entstehung:
  • Anmerkungen: Diese Datenquelle enthält auch Bestandsnachweise, die nicht zu einem Volltext führen.
  • Beschreibung: The adoption of multi-core platforms in embedded real-time systems mandates predictable system components. Such components must guarantee the satisfaction of the timing constraints of various applications running on the system. One of the components that can break the system predictability is cache coherence, which ensures the correctness of shared data. This paper proposes a solution towards the enablement of predictable cache coherent real-time systems. The solution uses existing COTS coherence protocols and proposes a methodology to integrate them with legacy real-time arbiters without imposing any required modification to either of them. Doing so, the paper also works as an exploratory study of the integration of various coherence protocols with various predictable arbitration schemes leading to a total of 12 different architecture configurations. Evaluation against four state-of-the-art predictable coherence solutions as well as COTS-based solutions show that the proposed approach achieves the tightest existing latency bounds among predictable solutions with minimal performance degradation over the COTS ones.
  • Zugangsstatus: Freier Zugang