Simulations and performance evaluation of Real-Time Multi-core Systems

The development of broad range of multi-core processors in desktop and server systems has lead to a definite need of an overall performance evaluation tool. This paper presents a new tool to analyze the performance of multi-core systems at early design phases. The proposed tool helps developers test different design options and choose the best solution for multi-core applications. Different design factors can be considered and evaluated to get the best core utilization of multi-core systems while achieving the best response time for the real-time applications. The paper explores the implementation of different algorithms at four different design stages: dependability analysis, task execution sequence, real-time scheduling and core mapping. As a proof of concept, a case study is presented to show the significance of changing one design parameter on the overall system performance. Experimental results show an increase of the CPU utilization by 31.25% when changing the number of cores from 3 to 2.