[11p-W810-1~17] 31.1 Focused Session "AI Electornics"

▼ [11p-W810-2] Characterization and Modeling of FTJ Memristive Devices for in-Memory Computing

〇(P)Radu M Berdan¹, Takao Marukame¹, Yoshifumi Nishi¹, Kensuke Ota², Masumi Saito², Shosuke Fujii² (1.TOSHIBA, 2.TOSHIBA MEMORY CORPORATION)

Keywords:memristor, neural networks, FTJ

In the context of in-memory computing, emerging non-volatile memristive devices can be used as the memory computing element, storing learned features of incoming data while also performing computation at the location of the memory, drastically decreasing power consumption promoted by the intrinsic non-von-Neumann architecture of the system. We present a pulsing study on HfSiO FTJ nano-scale memristive devices showing promising behavior suitable for novel in-memory computing systems, expanded to hardware friendly neural networks.