Efficient 16 Boolean logic and arithmetic based on bipolar oxide memristors
- PDF / 531,450 Bytes
- 8 Pages / 595 x 842 pts (A4) Page_size
- 11 Downloads / 180 Views
. RESEARCH PAPER .
October 2020, Vol. 63 202401:1–202401:8 https://doi.org/10.1007/s11432-020-2866-0
Efficient 16 Boolean logic and arithmetic based on bipolar oxide memristors Rui YUAN1 , Mingyuan MA2 , Liying XU1 , Zhenhua ZHU2 , Qingxi DUAN1 , Teng ZHANG1 , Yu ZHU2 , Yu WANG2 , Ru HUANG1,3,4* & Yuchao YANG1,3,4* 1
Key Laboratory of Microelectronic Devices and Circuits (MOE), Department of Micro/Nanoelectronics, Peking University, Beijing 100871, China; Department of Electronic Engineering, Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing 100084, China; 3 Center for Brain Inspired Chips, Academy for Artificial Intelligence, Peking University, Beijing 100871, China; 4 Frontiers Science Center for Nano-optoelectronics, Peking University, Beijing 100871, China
2
Received 15 March 2020/Revised 25 March 2020/Accepted 8 April 2020/Published online 26 August 2020
Abstract The physically separated memory and logic units in traditional von Neumann computers place essential limits on the performance and cause increased energy consumption, and hence in-memory computing is required to overcome this bottleneck. Here, a new bipolar memristor based in-memory logic approach is proposed, which is capable of achieving all 16 possible Boolean logic functions in a single device in less than 3 steps. This approach does not require initialization and can facilitate logic cascading, and the calculation taking place in-situ is showcased by 1-bit full adder and 2-bit multiplier with improved efficiency, thus showing a great prospect in future in-memory computing architecture. Keywords
memristor, nonvolatile logic, in-memory computing, full adder, multiplier
Citation Yuan R, Ma M Y, Xu L Y, et al. Efficient 16 Boolean logic and arithmetic based on bipolar oxide memristors. Sci China Inf Sci, 2020, 63(10): 202401, https://doi.org/10.1007/s11432-020-2866-0
1
Introduction
As the scaling of CMOS devices approaches their physical limits, Moore’s law is approaching its end [1]. The development of future computing technology requires innovative materials, devices, and architectures [2–5]. While the performance of traditional computers is strongly limited by the von Neumann bottleneck, in-memory logic can fundamentally address the overhead caused by data movements between logic and memory and thus achieve high efficiency [6]. Recently, memristor based in-memory computing has dawn extensive attention [7–10], due to its high speed, high scalability, and low energy consumption [11–15]. Memristor is a simple metal-insulator-metal sandwich structure, and their resistances can be switched between two digital states, low resistance state and high resistance state [16–18], laying the foundation for logical operations. Memristors have demonstrated rich logic and arithmetic functionalities in recent years [8, 19–34]. Memristor based nonvolatile logic can usually be divided into two classesstateful and non-stateful logic, according to the physical quantities of input and output variables. For s
Data Loading...
