Dynamic Loading Simulation Method for Large-scale Spiking Neural Network

Authors

Jiawei Shen, College of Computer Science and Technology, Zhejiang University, Hangzhou 310013, China
Daye Cai, College of Computer Science and Technology, Zhejiang University, Hangzhou 310013, China
Guoqing Yang, College of Computer Science and Technology, Zhejiang University, Hangzhou 310013, China
Pan Lü, College of Computer Science and Technology, Zhejiang University, Hangzhou 310013, China
Hong Li, College of Computer Science and Technology, Zhejiang University, Hangzhou 310013, China

Abstract

Abstract: To address the problem of high GPU memory requirements in large-scale spiking neural network simulation, a dynamic loading simulation method for large-scale spiking neural networks is proposed. This method uses data movement at the sub-network granularity and utilizes the host memory as a larger memory pool to reduce the limitation of GPU memory on the model simulation scale, enabling large-scale spiking neural network simulation on a single GPU computer. The pipeline acceleration technique is adopted to reduce the impact of data movement on simulation speed. The simulation of a million-scale neural network is achieved in a single GPU experimental environment, which solves the problem of insufficient memory during spiking neural network simulation.

Recommended Citation

Shen, Jiawei; Cai, Daye; Yang, Guoqing; Lü, Pan; and Li, Hong (2025) "Dynamic Loading Simulation Method for Large-scale Spiking Neural Network," Journal of System Simulation: Vol. 37: Iss. 2, Article 20.
DOI: 10.16182/j.issn1004731x.joss.23-1220
Available at: https://dc-china-simulation.researchcommons.org/journal/vol37/iss2/20

First Page

541

Last Page

550

CLC

TP391

Recommended Citation

Shen Jiawei, Cai Daye, Yang Guoqing, et al. Dynamic Loading Simulation Method for Large-scale Spiking Neural Network[J]. Journal of System Simulation, 2025, 37(2): 541-550.

Corresponding Author

Yang Guoqing

DOI

10.16182/j.issn1004731x.joss.23-1220