Binary Information Press
Modern embedded devices need to support a large number of new coming applications. For users' satisfactions of potential use-cases, MPSoC (Multi-Processor Systems-on-Chip) system is one of the promising solutions. But it becomes a challenge to predict the performance and analyze the design, especially when applications may be dynamically scheduled in the system. This paper presents a probabilistic technique to predict the performance of multiple applications on MPSoC system. And probabilistic computing and analysis is used to estimate the time of tasks executions in the contention with resources, and thereby predicting the performance of applications.