Adaptive scheduling method and device and computer readable storage medium

Abstract

The invention discloses an adaptive scheduling method and device, and a computer readable storage medium, and the method comprises the steps of initializing the calculation resources of each algorithmnode, and setting an adaptive adjustment time interval; counting the task load of each algorithm node in a time interval; and allocating new computing resources to each algorithm node according to the task load, the computing resources of each piece of programmable hardware and the computing load required for executing the algorithm program, and controlling each piece of programmable hardware toexecute the algorithm program corresponding to each algorithm node according to the new computing resources. By means of the mode, allocation of computing resources can be adjusted in an adaptive mode, and the processing speed of an algorithm program is increased.

Description

technical field [0001] The present application relates to the technical field of communications, and in particular to an adaptive scheduling method, device, and computer-readable storage medium. Background technique [0002] With the development of deep learning-related intelligent algorithms, the application of intelligent services is becoming wider and wider, and intelligent algorithms can perform more and more functions. Usually, a complex intelligent service function is completed by multiple different intelligent algorithms in sequence; in In the security application scenario, the intelligent services that need to be carried in the front-end equipment and the back-end equipment are becoming more and more complex, and the hardware units used for algorithm processing in a corresponding equipment also have different computing performance, and the processing performance usually pursues throughput (unit The number of pictures processed in time), under the premise of pursuing ...

AI Technical Summary

Problems solved by technology

[0002] With the development of deep learning-related intelligent algorithms, the application of intelligent services is becoming wider and wider, and intelligent algorithms can perform more and more functions. Usually, a complex intelligent service function is completed by multiple different intelligent algorithms in sequence; in In the security application scenario, the intelligent services that need to be carried in the front-end equipment and the back-end equipment are becoming more and more complex, and the hardware units used for algorithm processing in a corresponding equipment also have different computing performance, and the processing performance usually pursues throughput (unit The number of pictures processed in time), under the premise of pursuing throughput, different algorithm nodes process different frames in the same video channel or picture stream at the same time, and the hardware computing resources required by different algorithm nodes are usually statically configured according to empirical values. However, this cannot cope with the dynamic changes in the data processing volume of different algorithm nodes in actual scenarios.

Images