DAM SRAM CORE: An Efficient High-Speed and Low-Power CIM SRAM CORE Design for Feature Extraction Convolutional Layers in Binary Neural Networks
Micromachines · 30.04.2024 · 15 (5) · 617
DOI: 10.3390/mi15050617 ↗Kurz gefasst
Fachartikel in der Zeitschrift Micromachines, online erschienen am 30.04.2024. Verfasst von Ruiyong Zhao und 3 weiteren Autor:innen. Beteiligte Einrichtung: Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, Shanghai 200031, China. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 11 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Science and Technology Commission of Shanghai Municipality, Shanghai Zhangjiang Laboratory, Zhangjiang National Laboratory.
Abstract (Originalsprache)
This article proposes a novel design for an in-memory computing SRAM, the DAM SRAM CORE, which integrates storage and computational functionality within a unified 11T SRAM cell and enables the performance of large-scale parallel Multiply–Accumulate (MAC) operations within the SRAM array. This design not only improves the area efficiency of the individual cells but also realizes a compact layout. A key highlight of this design is its employment of a dynamic aXNOR-based computation mode, which significantly reduces the consumption of both dynamic and static power during the computational process within the array. Additionally, the design innovatively incorporates a self-stabilizing voltage gradient quantization circuit, which enhances the computational accuracy of the overall system. The 64 × 64 bit DAM SRAM CORE in-memory computing core was fabricated using the 55 nm CMOS logic process and validated via simulations. The experimental results show that this core can deliver 5-bit output results with 1-bit input feature data and 1-bit weight data, while maintaining a static power consumption of 0.48 mW/mm2 and a computational power consumption of 11.367 mW/mm2. This showcases its excellent low-power characteristics. Furthermore, the core achieves a data throughput of 109.75 GOPS and exhibits an impressive energy efficiency of 21.95 TOPS/W, which robustly validate the effectiveness and advanced nature of the proposed in-memory computing core design.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, Shanghai 200031, China.
- University of Chinese Academy of Sciences, Beijing 100049, China.
Förderung
- Science and Technology Commission of Shanghai Municipality
- Shanghai Zhangjiang Laboratory
- Zhangjiang National Laboratory
Zitieren
APA
Zhao, R., Gong, Z., Liu, Y., & Chen, J. (2024). DAM SRAM CORE: An Efficient High-Speed and Low-Power CIM SRAM CORE Design for Feature Extraction Convolutional Layers in Binary Neural Networks. Micromachines, 15(5), 617. https://doi.org/10.3390/mi15050617
Vancouver
Zhao R, Gong Z, Liu Y, Chen J. DAM SRAM CORE: An Efficient High-Speed and Low-Power CIM SRAM CORE Design for Feature Extraction Convolutional Layers in Binary Neural Networks. Micromachines. 2024;15(5):617. doi:10.3390/mi15050617
BibTeX
@article{Zhao2024,
author = {Zhao, Ruiyong and Gong, Zhenghui and Liu, Yulan and Chen, Jing},
title = {DAM SRAM CORE: An Efficient High-Speed and Low-Power CIM SRAM CORE Design for Feature Extraction Convolutional Layers in Binary Neural Networks},
journal = {Micromachines},
year = {2024},
volume = {15},
number = {5},
pages = {617},
doi = {10.3390/mi15050617}
}
EUVIMED