Dynamic task allocation method of heterogeneous computing system

Abstract

The invention relates to a dynamic task allocation method of a heterogeneous computing system. The heterogeneous computing system completes the application task by cooperating with a group of heterogeneous processors together, decomposes the task into a group of parallel subtasks, and schedules the subtasks to all processors according to the execution sequence. The method realizes the dynamic allocation and the optimized allocation of the task in the heterogeneous computing system through the dynamic selection of the processors to a task to be processed. Compared with the prior art, the invention provides an optimizing scheduling method based on the heterogeneous computing system by considering the optimizing scheduling problem of the heterogeneous computing system from the dynamic angle and utilizing the swarm intelligence technique. The method can perform the dynamic task allocation according to the processing capacities of different processors in the heterogeneous computing system, and considers the possibly occurred competence problem. The invention dynamically considers the computing power of the processors and the loading condition of the processors so as to further quicken the execution time of the heterogeneous computing system.

Description

technical field [0001] The invention relates to an optimization scheduling technology of a heterogeneous computing system, in particular to a dynamic task allocation method of a heterogeneous computing system. Background technique [0002] The heterogeneous computing system HCS (Heterogeneous Computing System) is to use a group of heterogeneous computers to work together to complete a certain application task. Decompose the task into multiple parallel sub-tasks, and dispatch them to each processor of HCS according to the order of execution, which not only satisfies different types of applications, but also develops and utilizes the computing power of various machines in the system, so that the system has a higher performance. In the past, it was believed that adding more machines would improve the performance of the system. However, in many cases, using a larger number of machines cannot achieve the expected results, because even if a large number of machines are selected,...

AI Technical Summary

Problems solved by technology

However, in many cases, using a larger number of machines cannot achieve the expected results, because even if a large number of machines are selected, due to the different computing capabilities of various models, each subtask waits for each other during the calculation process. The execution time of the entire task is prolonged, thereby reducing the performance of the system. Therefore, people are considering how to allocate tasks reasonably so that the performance of heterogeneous computing systems can be well exerted.

Images