@inproceedings{10.1145/3673038.3673143,
author = {Wu, Tong and He, Shuibing and Zhu, Jianxin and Chen, Weijian and Yang, Siling and Chen, Ping and Yin, Yanlong and Zhang, Xuechen and Sun, Xian-He and Chen, Gang},
title = {AUTOHET: An Automated Heterogeneous ReRAM-Based Accelerator for DNN Inference},
year = {2024},
isbn = {9798400717932},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3673038.3673143},
doi = {10.1145/3673038.3673143},
abstract = {ReRAM-based accelerators have become prevalent in accelerating deep neural network inference owing to their in-situ computing capability of ReRAM crossbars. However, most existing ReRAM-based accelerators are designed with homogeneous crossbars, leading to either low resource utilization or sub-optimal energy efficiency. In this paper, we propose AutoHet, an automated heterogeneous ReRAM-based accelerator with varied-size crossbars for different DNN layers. To achieve both high crossbar utilization and energy efficiency, AutoHet uses a reinforcement learning algorithm to automatically determine the proper crossbar configuration for each DNN layer. Additionally, AutoHet introduces rectangle crossbars and a tile-shared crossbar allocation scheme to reduce crossbar wastage and energy consumption. Experiment results show that AutoHet effectively improves crossbar utilization by up to 3.1 \texttimes{} and reduces energy consumption by up to 94.6\%, compared to approaches with homogeneous ReRAM crossbars.},
booktitle = {Proceedings of the 53rd International Conference on Parallel Processing},
pages = {1052–1061},
numpages = {10},
keywords = {Heterogeneous architecture, Processing-in-memory, ReRAM-based accelerator, Reinforcement learning},
location = {Gotland, Sweden},
series = {ICPP '24}
}