Neuromorphic computing at scale

• 成果类型: Review
• 署名作者: Kudithipudi, Dhireesha; Schuman, Catherine; Vineyard, Craig M.; Pandit, Tej; Merkel, Cory; Kubendran, Rajkumar; Aimone, James B.; Orchard, Garrick; Mayr, Christian; Benosman, Ryad; Hays, Joe; Young, Cliff; Bartolozzi, Chiara; Majumdar, Amitava; Cardwell, Suma George; Payvand, Melika; Buckley, Sonia; Kulkarni, Shruti; Gonzalez, Hector A.; Cauwenberghs, Gert; Thakur, Chetan Singh; Subramoney, Anand; Furber, Steve
• 署名单位: University of Texas System; University of Texas at San Antonio; University of Tennessee System; University of Tennessee Knoxville; United States Department of Energy (DOE); Sandia National Laboratories; Rochester Institute of Technology; Pennsylvania Commonwealth System of Higher Education (PCSHE); University of Pittsburgh; Intel Corporation; Intel USA; Technische Universitat Dresden; United States Department of Defense; United States Navy; United States Naval Research Laboratory; Alphabet Inc.; DeepMind; Istituto Italiano di Tecnologia - IIT; University of California System; University of California San Diego; University of Zurich; Swiss Federal Institutes of Technology Domain; ETH Zurich; National Institute of Standards & Technology (NIST) - USA; United States Department of Energy (DOE); Oak Ridge National Laboratory; Indian Institute of Science (IISC) - Bangalore; University of London; Royal Holloway University London; University of Manchester
• 刊物名称: Nature
• ISSN/ISSBN: 0028-2385
• DOI: 10.1038/s41586-024-08253-8
• 发表日期: 2025-01-23
• 页码: 801-812
• 关键词: neural-networks; on-chip; SYSTEM; architecture; DESIGN; loihi

摘要

Neuromorphic computing is a brain-inspired approach to hardware and algorithm design that efficiently realizes artificial neural networks. Neuromorphic designers apply the principles of biointelligence discovered by neuroscientists to design efficient computational systems, often for applications with size, weight and power constraints. With this research field at a critical juncture, it is crucial to chart the course for the development of future large-scale neuromorphic systems. We describe approaches for creating scalable neuromorphic architectures and identify key features. We discuss potential applications that can benefit from scaling and the main challenges that need to be addressed. Furthermore, we examine a comprehensive ecosystem necessary to sustain growth and the new opportunities that lie ahead when scaling neuromorphic systems. Our work distils ideas from several computing sub-fields, providing guidance to researchers and practitioners of neuromorphic computing who aim to push the frontier forward.