Dependency-Based Task Optimal Deployment Method

Abstract

The present invention provides a correlation-based task optimization deployment method. The symmetric multi-processor system adopted includes multiple CPUs, and each CPU has a dedicated affinity queue, including: if a new task is ready, judge the new Whether the task has affinity; put the new task with affinity into the affinity queue, and judge whether the task running on the affinity CPU has CPU affinity, if the task running on the CPU does not have CPU affinity When the affinity is high, it will preempt the CPU and exit at the same time. If the task running on the CPU has CPU affinity, it will exit; put the new task without affinity into the global ready queue, and look for the lowest priority task to run CPU, if it finds a CPU running on a task with a lower priority than the new task, it will preempt the CPU, otherwise it will exit directly. The invention reasonably distributes the task scheduling of the operating system, balances the loads of multiple CPUs, and makes full use of hardware resources.

Description

technical field [0001] The invention relates to a symmetric multiprocessor system, in particular to a correlation-based task optimization deployment method. Background technique [0002] With the advancement of science and technology, people have higher and higher requirements for data processing speed. For example, biologists are trying to understand the meaning of human genes, astronomers are understanding the universe, and gamers are pursuing the sense of accomplishment brought by smooth and high-definition games. Unlike Shock, these require more CPU (processor) cycles. [0003] Single-processor technology has been pushed to the extreme, and one way to achieve higher processing speeds is to use parallel computing on a large scale. Shared-memory multiprocessors emerged as the times require, in which two or more CPU cores all share access to a common RAM (memory). [0004] Multiprocessor operation requires a multiprocessor operating system. At present, there are two type...

AI Technical Summary

Problems solved by technology

The first method uses only one processor, and the performance of the processor is equivalent to that of a single processor, but the power consumption, cost, and volume of the system increase greatly; the second method uses other processors as computing resources of the system, Assigning tasks to each computing resource is more efficient than the first method, but since the operating system runs on one of the processors, the parallel computing capabilities of multi-processors have not been fully developed

Images