Magnetic Nanoelectronics for Brain-Inspired Computing (MN-BRIC): From Materials to Circuit Models

Abstract

This technical report summarizes the R and D efforts for the AFRL project "Magnetic Nanoelectronics for Brain-Inspired Computing (MN-BRIC): From Materials to Circuit Models". This effort focused on spintronics technology, wherein they exploited the magnetization dynamics of magnetic materials, most prominently antiferromagnets (AFMs), to implement brain-inspired circuits and architectures. The project focused on developing physics-based models at various levels of the design hierarchy to fully quantify the potential and limits of spintronics for brain-inspired computing. This effort leveraged RPIs complementary expertise in neuromorphic hardware simulator (NeMo) that can simulate neuromorphic architectures of arbitrary dimensions, allowing for novel architecture performance benchmarking. The major outcomes of the effort include (i) materials and device models of AFM neurons and ferromagnetic (FM) synapses (ii), (circuit-compatible SPICE models of these neurons and synapses, (iii) full architecture level benchmarking, including power, throughput, and area assessment of a neuromorphic architecture leveraging magnetic neurons and synapses and (iv) quantification of the impact of interconnects on architecture performance. The PI's results show that although spintronics hardware offers significant energy and latency advantages, for larger neuromorphic cores, the performance is dominated by interconnection networks. This limitation is overcome by architectural changes to the network or by using new interconnect materials that offer lower resistance and capacitance compared to copper/low-k interconnects, currently used in CMOS chips. Significantly, the work under this grant has also generated new concepts of brain inspired computing at the thermodynamic limits based on diffusive and stochastic phenomena that can be obtained in magnetic devices.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 2023
Accession Number
AD1200264

Entities

People

  • Shaloo Rakheja

Organizations

  • University of Illinois Urbana–Champaign

Tags

Communities of Interest

  • Advanced Electronics

DTIC Thesaurus Topics

  • Air Force
  • Air Force Research Laboratories
  • Artificial Intelligence
  • Artificial Intelligence Software
  • Complementary Metal-Oxide Semiconductors
  • Computers
  • Computing System Architectures
  • Current Density
  • Domain Walls
  • Electric Current
  • Electronics Laboratories
  • Energy Consumption
  • Energy Efficiency
  • Magnetic Devices
  • Magnetic Materials
  • Network Science
  • Neural Networks
  • Physics
  • Simulations
  • Simulators
  • Spintronics
  • United States

Readers

  • Computational Fluid Dynamics (CFD)
  • Integrated Circuit Design and Technology.
  • Materials Science and Engineering.

Technology Areas

  • Microelectronics