Research On Associated Task Scheduling Algorithm Based On Multi-core Processor

The multi-core processor improves the parallelism of task scheduling effectively,but the inter core communication overhead also becomes the main factor affecting the performance of the system.When tasks with data association are scheduled in parallel on multi-core processors,associated data needs to be transmitted through inter core communication network,and the communication overhead generated during this period has a negative impact on the scheduling performance of the system.At present,the challenge in the research of multi-core processor task scheduling is how to schedule computing tasks and communication tasks cooperatively and optimize scheduling performance.Because the scheduling of communication tasks is closely related to the inter core communication network architecture,this paper focuses on the scheduling problem of multi-core processor in bus architecture and multi-core processor in full connection architecture.Based on the scheduling characteristics of multi-core processors with associated tasks in the bus architecture,this paper proposes a scheduling optimization algorithm for associated tasks.In the algorithm design,considering the impact of communication tasks,a task scheduling priority allocation method is proposed,and according to the earliest completion time and correlation degree of tasks in the kernel,an optimal matching function between tasks and execution kernel is proposed.The processing kernel is selected according to the evaluation function value,so as to avoid a large amount of inter kernel communication overhead caused by task parallel scheduling.Based on the scheduling characteristics of associated tasks in multi-core processors of full connection architecture,this paper improves the existing scheduling algorithm so that it can schedule associated tasks in multi-core processors of full connection architecture.In addition,for the associated tasks with time constraints,this paper proposes a scheduling algorithm.By calculating the deadline of tasks in each kernel,the tasks are allocated in the same kernel as far as possible without exceeding the deadline,so as to reduce the communication overhead between cores.In this paper,the performance of the algorithm is evaluated by task scheduling length,processor utilization,generated traffic,number of active cores,task acceptance rate and so on.The experimental results show that the algorithm proposed in this paper can effectively reduce the inter core communication overhead and optimize the system scheduling performance.