The Improvement And Implement Of Dual-priority Scheduling Algorithm Based On Multi-processor

With the development of multi-processor technology, MPSoC architecture which is composed by multiprocessor unit and storage system in chip has become the standard in embedded system. Multiprocessor system has taken high calculate density, more powerful parallel computing performance and less clock frequency. Also it largely relieves the high processor consumption and heat output. Not only MPSoC is more and more popular in high-end embedded system design, but also MPSoC has become the main trend platform in multi-core research field.Real-time systems with high reliability and instantaneity, has been widely used in many fields. In current real-time control system, task scheduling in Real-time systems is a research hot spot, Multiprocessor task scheduling is a NP-hard problem, dual priority algorithm is a simple and efficient multi-task scheduling algorithm. Dual priority algorithm divides the priority to three pieces so that it can modify the whole system to more impartial and high-efficiency. This paper has analysis the problems in multi-task scheduling algorithm based on the current research situation in here and abroad. Combining the development of multiprocessor, this paper improves the dual priority algorithm so that it can adapt to multiprocessor, also this paper propose two kinds of task dispatching strategy.Aiming to verify the algorithm performance, this paper has designed two kinds of testing platform. The first one is a simulator based on C language developing. Inputting system task using ratio, creating the task sequence by random functions, it can simulate the whole implementation and get the soft real-time periodic task loss rate. The second platform is realized by Virtex-5. It can achieve the real task scheduling in multiprocessor system by multi-core number configuration, inter-core communication and share mechanism. At last, this paper tests the algorithm from different platform, different task dispatching strategy and different number of multi-core. A primary conclusion has come out.From the paper, we can find out the MPSoC testing platform construct lays foundations for the subsequent research in multiprocessor scheduling. The improving of dual priority scheduling algorithm take some purpose to multiprocessor scheduling field.