This book provides an interdisciplinary overview of spin-orbitronics-based magnetic racetrack memory (RTM). The authors provide an overview of RTM device physics, its evolution, modeling tools, strengths and challenges, and its application in the memory subsystem. They describe leading-edge developments in RTM physics and materials, data sensing elements, controlled and energy-efficient movement of the magnetic elements within the nanowire, and custom shifts-aware data and instruction placement solutions using optimal, near-optimal, and evolutionary algorithms. The book describes automatic software optimization methods that take an input program, analyze its memory access behavior, and, if possible, optimize it by reordering the memory access without any user intervention. Further, it also discusses state-of-the-art reliability schemes that can accurately detect and correct position errors in RTMs. Finally, this book demonstrates RTM-based computation-in-memory and its potential to break the memory and bandwidth walls. The authors explain how the RTM devices can morph into compute elements and perform in-situ computations using their properties. As a concrete use case, the book presents an entire hyperdimensional computing framework and explains the implementation of various operations.

Provides interdisciplinary coverage of state-of-the-art, covering physics, applications, optimization, and reliability Presents RTM architectures and device parameters for most RTM-specific and relevant components of the memory subsystem Includes hardware and software optimizations for RTM caches, scalar accesses, instruction streams, and array accesses

Autorentext

Asif Ali Khan is a post-doctoral researcher at the Chair for Compiler Construction in the Computer Science Department of the TU Dresden, Germany. His research interests include computer architecture, heterogeneous memories, and compiler support for the memory subsystem. Asif s research focuses on exploring the emerging nonvolatile memory technologies in the memory subsystems and their optimizations for various metrics. He is also investigating the potential of emerging NVMs in future non- von-Neumann heterogeneous systems employing custom domain-specific accelerators.

Jeronimo Castrillon is a professor in the Department of Computer Science at the TU Dresden, where he is also affiliated with the Center for Advancing Electronics Dresden (CfAED). He is the head ofthe Chair for Compiler Construction, with research focus on methodologies, languages, tools and algorithms for programming complex computing systems. He received the Electronics Engineering degree from the Pontificia Bolivariana University in Colombia in 2004, his masters degree from the ALaRI Institute in Switzerland in 2006 and his Ph.D. degree (Dr.-Ing.) with honors from the RWTH Aachen University in Germany in 2013. In 2014, Prof. Castrillon co-founded Silexica GmbH/Inc, a company that provides programming tools for embedded multicore architectures, which is now with AMD/Xilinx.

Alex K. Jones received the BS degree in 1998 in physics from the College of William and Mary in Williamsburg, VA, USA, and the MS and PhD degrees in 2000 and 2002, respectively, in ECE from Northwestern University, Evanston, IL, USA. He is a Professor of ECE and CS at the University of Pittsburgh, Pittsburgh, PA, USA.He is currently serving as a Program Director at the US NSF in the CNS Division of the CISE Directorate. Dr. Jones' research interests include compilation for configurable systems and architectures, scaled and emerging memory, reliability, fault tolerance, and sustainable computing. He is the author of more than 200 publications in these areas. His research is funded by the NSF, DARPA, NSA, and industry. Dr. Jones received a top 25 paper award from the first 20 years of FCCM. He is a senior member of the IEEE and the ACM.

Stuart S. P. Parkin is currently the Managing Director of the Max Planck Institute for Microstructure Physics, Halle, Germany, and an Alexander von Humboldt Professor with Martin-Luther-University Halle- Wittenberg, Halle. He also proposed magnetic random access memory based on magnetic tunnel junctions in 1995 which he and his colleagues at IBM Research Almaden, San Jose, CA, USA, demonstrated in 1999, and which has recently become available as a mainstream foundry technology. In 2002, he also proposed Racetrack Memory that is the subject of this article. His research interests include spintronic materials and devices for advanced sensor, memory, and logic applications, oxide thin-film heterostructures, topological metals, exotic superconductors, and cognitive devices. His discoveries in spintronics enabled a more than 10 000-fold increase in the storage capacity of magnetic disk drives. He has published approximately 570 articles and 118 issued patents and an impact factor H = 112 (Google Scholar).

Dr. Parkin is a Fellow/Member of the Royal Society, the Royal Academy of Engineering, the National Academy of Sciences, the National Academy of Engineering, the German National Academy of Science - Leopoldina, the Royal Society of Edinburgh, the Indian Academy of Sciences, and Third World Academy of Sciences (TWAS) Academy of Sciences for the developing world. For this work, which thereby enabled the big data world of today, he was awarded the Millennium Technology Award from the Technology Academy Finland in 2014.

Inhalt

Introduction.- Magnetic Racetrack Memories.- Design and applications.- RTM optimizations.- Position error correction codes for RTMs.- Co-design/synergetic solutions.- In-RTM and near-RTM computing.- Conclusions and Outlook.