Design And Implementation Of Parallel Task Scheduling For OpenMp

As a mainstream parallel programming framework,OpenMP framework is widely used in general computing and high-performance computing.In recent years,with the rapid development of multi-core embedded processors and complex parallel applications,parallel real-time systems supporting OpenMP programs have received extensive attention.However,the OpenMP runtime mechanism still lacks a mechanism related to real-time task scheduling,which fails to meet the real-time scheduling requirements of embedded real-time systems.Aiming at the OpenMP parallel programming framework,focusing on the task scheduling problem of OpenMP programs in real-time operating systems,this thesis studies the OpenMP task division method and runtime thread scheduling technology,and develops a corresponding operating system scheduling visualization simulation platform.The main research work of this thesis is as follows:（1）This thesis proposes a new optimization method for the task granularity division problem existing in the OpenMP parallel framework,which effectively improves the overall execution efficiency of the program.（2）This thesis combines the OpenMP parallel programming framework with the task scheduling algorithm in multi-core processors.Through the interface functions provided in Litmus-RT,the tasks existing in the OpenMP framework bound to the specific processor that executes the tasks,and the Partitioned Fixed-Priority（P-FP）scheduling algorithm is studied and implemented.At the same time,this thesis combines with the Global Fixed-Priority（G-FP）scheduling algorithm and the OpenMP framework to conduct related research and analysis.（3）For the two scheduling algorithms in this thesis,a visual simulation platform for task scheduling algorithms based on the OpenMP parallel framework is designed and implemented.By the simulation analysis tool,the process of the above scheduling algorithm is analyzed and simulated.