The Japan Society of Applied Physics

9:00 AM - 9:30 AM

[J-5-01 (Invited)] Impact of Spintronics-Based Nonvolatile Hardware for AI Applications

Takahiro Hanyu1, Naoya Onizawa1, Daisuke Suzuki2, Masanori Natsui1 (1. Tohoku Univ. (Japan), 2. Univ. of Aizu (Japan))

https://doi.org/10.7567/SSDM.2023.J-5-01

Spintronics devices have potential advantages such as non-volatility, fast read/write, and high endurance under good back-end-of-the-line compatibility. They offer the possibility of not only replacing ordinary RAMs, but also realizing low-power and high-performance VLSI proces-sors including edge AI hardware. This is achieved through a new logic-circuit style called "nonvolatile (NV) logic-in-memory architecture," which merges nonvolatile storage elements into CMOS logic-gate circuits. This pa-per explains the advantages of employing spintronics de-vices combined with CMOS circuits, and how to contrib-ute this technology towards edge AI applications.