A High Energy-Efficiency Multi-core Neuromorphic Architecture for Deep SNN Training
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909443361341440 |
|---|---|
| author | Li, Mingjing Zhou, Huihui Xu, Xiaofeng Zhong, Zhiwei Quan, Puli Zhu, Xueke Lin, Yanyu Lin, Wenjie Guo, Hongyu Zhang, Junchao Ma, Yunhao Wang, Wei Meng, Qingyan Ma, Zhengyu Li, Guoqi Cui, Xiaoxin Tian, Yonghong |
| author_facet | Li, Mingjing Zhou, Huihui Xu, Xiaofeng Zhong, Zhiwei Quan, Puli Zhu, Xueke Lin, Yanyu Lin, Wenjie Guo, Hongyu Zhang, Junchao Ma, Yunhao Wang, Wei Meng, Qingyan Ma, Zhengyu Li, Guoqi Cui, Xiaoxin Tian, Yonghong |
| contents | There is a growing necessity for edge training to adapt to dynamically changing environment. Neuromorphic computing represents a significant pathway for high-efficiency intelligent computation in energy-constrained edges, but existing neuromorphic architectures lack the ability of directly training spiking neural networks (SNNs) based on backpropagation. We develop a multi-core neuromorphic architecture with Feedforward-Propagation, Back-Propagation, and Weight-Gradient engines in each core, supporting high efficient parallel computing at both the engine and core levels. It combines various data flows and sparse computation optimization by fully leveraging the sparsity in SNN training, obtaining a high energy efficiency of 1.05TFLOPS/W@ FP16 @ 28nm, 55 ~ 85% reduction of DRAM access compared to A100 GPU in SNN trainings, and a 20-core deep SNN training and a 5-worker federated learning on FPGAs. Our study develops the first multi-core neuromorphic architecture supporting the direct SNN training, facilitating the neuromorphic computing in edge-learnable applications. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_05302 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | A High Energy-Efficiency Multi-core Neuromorphic Architecture for Deep SNN Training Li, Mingjing Zhou, Huihui Xu, Xiaofeng Zhong, Zhiwei Quan, Puli Zhu, Xueke Lin, Yanyu Lin, Wenjie Guo, Hongyu Zhang, Junchao Ma, Yunhao Wang, Wei Meng, Qingyan Ma, Zhengyu Li, Guoqi Cui, Xiaoxin Tian, Yonghong Hardware Architecture Distributed, Parallel, and Cluster Computing Machine Learning There is a growing necessity for edge training to adapt to dynamically changing environment. Neuromorphic computing represents a significant pathway for high-efficiency intelligent computation in energy-constrained edges, but existing neuromorphic architectures lack the ability of directly training spiking neural networks (SNNs) based on backpropagation. We develop a multi-core neuromorphic architecture with Feedforward-Propagation, Back-Propagation, and Weight-Gradient engines in each core, supporting high efficient parallel computing at both the engine and core levels. It combines various data flows and sparse computation optimization by fully leveraging the sparsity in SNN training, obtaining a high energy efficiency of 1.05TFLOPS/W@ FP16 @ 28nm, 55 ~ 85% reduction of DRAM access compared to A100 GPU in SNN trainings, and a 20-core deep SNN training and a 5-worker federated learning on FPGAs. Our study develops the first multi-core neuromorphic architecture supporting the direct SNN training, facilitating the neuromorphic computing in edge-learnable applications. |
| title | A High Energy-Efficiency Multi-core Neuromorphic Architecture for Deep SNN Training |
| topic | Hardware Architecture Distributed, Parallel, and Cluster Computing Machine Learning |
| url | https://arxiv.org/abs/2412.05302 |