In-memory computing using FLASH memory

FLASH is a type of electrically erasable read-only memory (EEPROM), where the program code is usually stored. It requires high programming voltages at the control gate in order to help electrons tunnel and get trapped in the floating gate and hence, store a binary bit. The cell structure of a floating gate is illustrated in Fig. 1. Flash memory technology has two cell structures: NAND, and NOR. NOR flash is faster to read than NAND flash, but it's also more expensive and it takes longer to erase and write new data. NAND has a higher memory capacity than NOR. As a consequence, NOR is utilized to store executable code for embedded computing devices. While NAND memory devices are usually used for data storage (Nand vs nor flash memory: an embedded developer's guide to choosing (2022) [Online]. Available: https://www.tuxera.com/blog/nand-nor-flash-embedded-developers-guide/). Cell storage types can go from single-level cell (SLC) to penta-level cell (PLC). The NAND flash memory market was valued at USD 46.62 billion in 2020, and it is expected to reach USD 85.36 billion by 2026, registering a CAGR of 11.27% during the forecast period (2021-2026). The current macro trends of AI and machine learning, mobility, and connectivity are favorable to the NAND markets and expected to increase its share in the semiconductor market (NAND flash memory market - growth, trends, covid-19 impact, and forecasts (2022-2027) (2021). [Online]. Available: https://www.mordorintelligence.com/industry-reports/nand-flash-memory-market). This Chapter discusses the potential applications utilizing Flash-based IMC applications, challenges and market availability.