RePIM: Joint Exploitation of Activation and Weight Repetitions for In-ReRAM DNN Acceleration

Chen Yang Tsai, Chin Fu Nien, Tz Ching Yu, Hung Yu Yeh, Hsiang Yun Cheng

研究成果: Conference contribution同行評審

18 引文 斯高帕斯(Scopus)

摘要

Eliminating redundant computations is a common approach to improve the performance of ReRAM-based DNN accelerators. While existing practical ReRAM-based accelerators eliminate part of the redundant computations by exploiting sparsity in inputs and weights or utilizing weight patterns of DNN models, they fail to identify all the redundancy, resulting in many unnecessary computations. Thus, we propose a practical design, RePIM, that is the first to jointly exploit the repetition of both inputs and weights. Our evaluation shows that RePIM is effective in eliminating unnecessary computations, achieving an average of 15.24× speedup and 96.07% energy savings over the state-of-the-art practical ReRAM-based accelerator.

原文English
主出版物標題2021 58th ACM/IEEE Design Automation Conference, DAC 2021
發行者Institute of Electrical and Electronics Engineers Inc.
頁面589-594
頁數6
ISBN(電子)9781665432740
DOIs
出版狀態Published - 5 12月 2021
事件58th ACM/IEEE Design Automation Conference, DAC 2021 - San Francisco, 美國
持續時間: 5 12月 20219 12月 2021

出版系列

名字Proceedings - Design Automation Conference
2021-December
ISSN(列印)0738-100X

Conference

Conference58th ACM/IEEE Design Automation Conference, DAC 2021
國家/地區美國
城市San Francisco
期間5/12/219/12/21

指紋

深入研究「RePIM: Joint Exploitation of Activation and Weight Repetitions for In-ReRAM DNN Acceleration」主題。共同形成了獨特的指紋。

引用此