A 22nm 3.5TOPS/W Flexible Micro-Robotic Vision SoC with 2MB eMRAM for Fully-on-Chip Intelligence

We present a highly flexible micro-robotic vision SoC featuring a hybrid Processing Element (PE) for efficient processing of both Convolutional Neural Network (CNN) and non-CNN vision tasks with 2MB embedded MRAM for retentive fully-on-chip weight storage. Fabricated in 22nm, the design achieves 0.2...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhang, Qirui, An, Hyochan, Fan, Zichen, Wang, Zhehong, Li, Ziyun, Wang, Guanru, Kim, Hun-Seok, Blaauw, David, Sylvester, Dennis
Format: Conference Proceeding
Language:English
Subjects:
Convolutional neural networks
Corner detection
Task analysis
Very large scale integration
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We present a highly flexible micro-robotic vision SoC featuring a hybrid Processing Element (PE) for efficient processing of both Convolutional Neural Network (CNN) and non-CNN vision tasks with 2MB embedded MRAM for retentive fully-on-chip weight storage. Fabricated in 22nm, the design achieves 0.22nJ/pix for Harris corner detection (a non-CNN vision task) and 3.5TOPS/W (INT16) for CNN, a 60% efficiency improvement over state-of-the-art NVM-based NN ASICs.
ISSN:2158-9682
DOI:10.1109/VLSITechnologyandCir46769.2022.9830340