Challenges and Trends of Nonvolatile In-Memory-Computation Circuits for AI Edge Devices
Je-Min Hung, Chuan-Jia Jhang, Ping-Chun Wu, Yen-Cheng Chiu, Meng‐Fan Chang
Abstract
Nonvolatile memory (NVM)-based computing-in-memory (nvCIM) is a promising candidate for artificial intelligence (AI) edge devices to overcome the latency and energy consumption imposed by the movement of data between memory and processors under the von Neumann architecture. This paper explores the background and basic approaches to nvCIM implementation, including input methodologies, weight formation and placement, and readout and quantization methods. This paper outlines the major challenges in the further development of nvCIM macros and reviews trends in recent silicon-verified devices.
Topics & Concepts
Non-volatile memoryComputer scienceVon Neumann architectureComputer architectureMacroSemiconductor memoryNon-volatile random-access memoryEmbedded systemComputer hardwareComputer memoryMemory refreshOperating systemProgramming languageAdvanced Memory and Neural ComputingFerroelectric and Negative Capacitance DevicesSemiconductor materials and devices