Load Balancing of Multi-Core Heterogeneous CPU Based on BP Neural Network and Weighted Round-Robin Algorithm

Abstract

The development of GPU has launched a huge challenge to the current traditional computing architecture. In recent years, CPUs have developed from single-core to multi-core to make up for the shortcomings of CPUs that are not good at large-scale calculations. How to balance the computational load of multiple CPUs has become an urgent problem in the academic community in recent years. This paper uses BP neural network to predict the current weight of the smooth weighted round-robin algorithm by extracting the characteristics of the computer system and task graph characteristics, and then improves the smooth weighted round-robin algorithm. In the end, the load of the multi-core heterogeneous processor system when processing large-scale tasks is significantly reduced.