Core dispatch methods and terminals

Abstract

An embodiment of the present invention provides a core scheduling method and related equipment. The method includes: obtaining target model parameters, which are used to represent the calculation density of a convolutional neural network model; according to the target model parameters, from the preset first The core weight values ​​of at least two cores are determined in a corresponding relationship, the core weight values ​​of at least two cores correspond to the target model parameters, at least two cores are heterogeneous cores on the terminal, and the first corresponding relationship includes the target model parameters and at least Correspondence between the core weight values ​​of the two cores, the core weight value is used to indicate the priority degree of the core being selected to run the convolutional neural network model; according to the core weight values ​​of at least two cores, the running volume is determined from at least two cores The core of the product neural network model. Through the core weight values ​​of different cores, an adapted core can be determined to run a convolutional neural network model with a specific computational density.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of chip systems, and in particular, to a core scheduling method and a terminal. Background technique [0002] Convolutional neural network (CNN) is a feed-forward neural network, its artificial neurons can respond to surrounding units within a part of the coverage, and it has excellent performance for large-scale image processing. There are more and more applications of convolutional neural networks on the terminal, such as image classification, feature extraction, and face clustering using convolutional neural networks. [0003] In order to improve the computing capability of the terminal, the system chip on the terminal often includes multiple heterogeneous cores, so that different cores can be used to perform different services. In the existing solutions, there is no effective core scheduling mechanism for the business processing performed by running the convolutional neural network...

AI Technical Summary

Problems solved by technology

[0004] Because different convolutional neural network models often have different characteristics, different cores also have different characteristics. In the existing scheme, the characteristics of different cores are not used to execute specific convolutional neural network models on the adapted core. This makes it inefficient to run specific convolutional neural network models and wastes computing resources on the terminal

Images