A task scheduling and allocation method for multi-core processors

Abstract

The invention relates to the field of processor task allocation, in particular to a multi-core processor-oriented task scheduling and allocation method. When the task arrives in the task queue, all the available cores of the current multi-core processor are converted to the distribution state, and the task is received from the task queue; after receiving the task, a task scheduling selection algorithm is executed to select a schedulable core; the received task is released Enter the task queue corresponding to the selected schedulable core; when the task arrives in the task queue, all the available cores of the current multi-core processor are converted to the computing state, and the task is taken from the task queue and calculated. The invention can balance the utilization rate of all cores of the multi-core processor to the greatest extent without the waste caused by the unbalanced capacity of the scheduling core and computing core; by making all the cores of the multi-core processor run in two modes, the scheduling mode and computing mode to improve the success rate of task scheduling and reduce spin loss.

Description

technical field [0001] The invention relates to the field of processor task allocation, in particular to a multi-core processor-oriented task scheduling and allocation method. Background technique [0002] Since the existing multi-core processor scheduling and allocation models are generally divided into two types: the first is global scheduling, that is, only one global task queue is maintained in the entire system, tasks are dynamically allocated, and tasks are allowed to migrate between cores. The second is local scheduling. The system maintains a local task queue for each core. Generally speaking, tasks in local scheduling are not allowed to be migrated. For global scheduling, generally multiple computing cores compete for tasks in the task queue, and it is easier to achieve load balancing and make each core fully loaded. However, for a scheduler that supports global scheduling, the calculation overhead is obviously too large, and frequent migration of tasks will reduce...

AI Technical Summary

Problems solved by technology

If the computing power of the scheduling core is insufficient, the global task queue will continue to pile up tasks, resulting in too many tasks that cannot be processed, and the task core cannot process the task; if the computing power of the computing core is insufficient, it will lead to the corresponding The task queue has a backlog of tasks, but the global task queue has no tasks, and the scheduling core is idle

Both cases will lead to a decrease in the success rate of task scheduling and an increase in spin loss

Images