Unicorn: A Multicore Neuromorphic Processor with Flexible Fan-In and Unconstrained Fan-Out for Neurons

被引：5

作者：

Yang, Zhijie ^{[1
]}

Wang, Lei ^{[1
]}

Wang, Yao ^{[1
]}

Peng, Linghui ^{[1
]}

Chen, Xiaofan ^{[1
]}

Xiao, Xun ^{[1
]}

Wang, Yaohua ^{[1
]}

Xu, Weixia ^{[1
]}

机构：

[1] Natl Univ Def Technol, Coll Comp Sci & Technol, Changsha, Hunan, Peoples R China

来源：

PROCEEDINGS OF THE 59TH ACM/IEEE DESIGN AUTOMATION CONFERENCE, DAC 2022 | 2022年

基金：

国家重点研发计划;

关键词：

multicore architecture; neuromorphic processor; spiking neural network; dynamic vision sensor; hardware accelerator; LARGE-SCALE MODEL;

D O I：

10.1145/3489517.3530563

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Neuromorphic processor is popular due to its high energy efficiency for spatio-temporal applications. However, when running the spiking neural network (SNN) topologies with the ever-growing scale, existing neuromorphic architectures face challenges due to their restrictions on neuron fan-in and fan-out. This paper proposes Unicorn, a multicore neuromorphic processor with a spike train sliding multicasting mechanism (STSM) and neuron merging mechanism (NMM) to support unconstrained fan-out and flexible fan-in of neurons. Unicorn supports 36K neurons and 45M synapses and thus supports a variety of neuromorphic applications. The peak performance and energy efficiency of Unicorn reach 36TSOPS and 424GSOPS/W respectively. Experimental results show that Unicorn can achieve 2x-5.5x energy reduction over the state-of-the-art neuromorphic processor when running an SNN with a relatively large fan-out and fan-in.

引用

页码：943 / 948

页数：6

共 50 条

[31] From Fan-out Wafer to Fan-out Panel Level Packaging
Braun, T.
Becker, K. -F.
Raatz, S.
Bader, V.
Bauer, J.
Aschenbrenner, R.
Voges, S.
Thomas, T.
Kahle, R.
Lang, K. -D.
2015 EUROPEAN CONFERENCE ON CIRCUIT THEORY AND DESIGN (ECCTD), 2015, : 29 - 32
[32] Stacked polymer waveguide type fan-in/fan-out device for dense multi-core fibre
Watanabe, Tatsuhiko
Kokubun, Yasuo
IET OPTOELECTRONICS, 2015, 9 (04) : 158 - 162
[33] Fan-in/Fan-out for Heterogeneous 19-core Fibers Based on Metasurfaces with Nonuniform Phase Plates
Wang, Yang
Wang, Xutao
Yang, Zhiqun
Liu, Yaping
Huang, Zhanhua
Zhang, Lin
2023 OPTICAL FIBER COMMUNICATIONS CONFERENCE AND EXHIBITION, OFC, 2023,
[34] Magnetization dynamics, throughput and energy dissipation in a universal multiferroic nanomagnetic logic gate with fan-in and fan-out
Fashami, Mohammad Salehi
Atulasimha, Jayasimha
Bandyopadhyay, Supriyo
NANOTECHNOLOGY, 2012, 23 (10)
[35] Fan-in/fan-out for heterogeneous 19-core fibers based on metasurfaces with nonuniform phase plates
Wang, Yang
Wang, Xutao
Li, Chunshu
He, Yichen
Huang, Zhanhua
Liu, Yaping
Yang, Zhiqun
Zhang, Lin
OPTICS LETTERS, 2024, 49 (01) : 5 - 8
[36] Fused Taper Type Fan-in/Fan-out Device for 12 Core Multi-Core Fiber
Uemura, H.
Omichi, K.
Takenaga, K.
Matsuo, S.
Saitoh, K.
Koshiba, M.
2014 OPTOELECTRONICS AND COMMUNICATIONS CONFERENCE AND AUSTRALIAN CONFERENCE ON OPTICAL FIBRE TECHNOLOGY (OECC/ACOFT 2014), 2014, : 49 - +
[37] WORST-CASE DESIGNED TUNNEL DIODE LOGIC NETWORK WITH ARBITRARY FAN-IN AND FAN-OUT CAPABILITY
BELL, LS
SEMICONDUCTOR PRODUCTS AND SOLID STATE TECHNOLOGY, 1965, 8 (09): : 11 - &
[38] FAN-OUT FILTERS
FRITZEMEYER, RD
ELECTRO-TECHNOLOGY, 1966, 77 (04): : 36 - +
[39] COMPUTER-AIDED SYNTHESIS OF MULTIPLE-OUTPUT MULTILEVEL NAND NETWORKS WITH FAN-IN AND FAN-OUT CONSTRAINTS
SU, SYH
NAM, CW
IEEE TRANSACTIONS ON COMPUTERS, 1971, C 20 (12) : 1445 - +
[40] Multi Functionality Demonstration for Multi Core Fiber Fan-in/Fan-out Devices using Free Space Optics
Tottori, Yusaku
Tsuboya, Hiroshi
Kobayashi, Tetsuya
2014 OPTICAL FIBER COMMUNICATIONS CONFERENCE AND EXHIBITION (OFC), 2014,

← 1 2 3 4 5 →