IFAC-PapersOnLine · 2022 · 19 citations · 7 references
ComputingComputer EngineeringComputer ScienceComputer ArchitectureOperating SystemsPerformance MeasurementJitter AnalysisReal-time Systems ArchitectureResponse Time AnalysisTiming AnalysisWorst-case Execution Time AnalysisContext Switching OverheadExecution JitterExecution Time AnalysisJitterMeasurement TechniquesMulti-core ComputingMulticore Timing AnalysisRescheduling
Real-Time Operating System (RTOS) executing on multi-core architectures still links to a lot of unresolved issues. The article presents potential problems and discusses measurement techniques indicating the performance and determinism. The advantages and disadvantages of individual measurement approaches are discussed, which cover a range from non-invasive techniques to techniques that require the cooperation of a monitored application. On top of that, the parameters indicating the performance and properties of RTOS running on multi-core are defined. Among them, the measuring of the semaphore taking, and task period jitter are described deeply. The operations are measured on ESP32-WROOM-32 development kit equipped with XTENSA dual-core processor running FreeRTOS. The results show unexpectedly high values of the switching context time and jitter when the rescheduling to another core was forced compared to time values measured on single core. Consequently, the unexpected rescheduling to other core increases the execution time of FreeRTOS operations. The work should facilitate the improvement of FreeRTOS implementation when running on multi-core architecture.
7
Calculating the maximum execution time of real-time programs
Peter Puschner, C. Koza · Real-Time Systems · 1989 · 582 citations
20th International Carpathian Control Conference
2019 · 68 citations · Full text
Measuring the Performance of Real-Time Systems
Wolfgang A. Halang, Roman Gumzej, Matjaž Colnarič et al. · Real-Time Systems · 2000 · 37 citations