Scheduling and Execution Time Analysis for Multi Core Architecture

Abstract

Multi-core processors are already widespread in general-purpose computing systems with many no of manufacturers currently offering up to a ten to twelve cores per processor. With this the systems adopting such multi-core enabled processors gain increased computational capacity, improved parallelism, and higher performance with respect to power consumption. However, using multicore processors in real-time applications also introduces new challenges and opportunities for development of efficient scheduling algorithms. In this report on scheduling issue in multi-core processor, we study this problem of scheduling, characterize the design space, and develop an analytical system which will provide the efficiency of few of the effective algorithms available till today. Exploiting the nature of processor cores, the general principle adopted in this report is to statically partition tasks among processor cores, co-allocate multiple synchronizing tasks when possible, and introduce limited intercore task migration when required and scheduling them when necessary for improving system utilization. We then analyze the overheads of inter-core task scheduling and synchronization and provide mechanisms to efficiently allocate synchronizing sequential tasks on multicores by co-locating such tasks on the basis of statics gathered over developed new system of scheduling. The results of this report contribute to a system that can efficiently utilize multi-core processors to predictably execute periodic real-time tasks with well-defined deadlines.