Racetrack Memory

Racetrack memory (RTM) is a novel memory technology aiming to improve the energy efficiency and speed of computation, particularly for artificial intelligence applications. Current research focuses on optimizing RTM's use in processing-in-memory (PIM) architectures for convolutional neural networks and other machine learning models, including exploring its application in hyperdimensional computing. This work addresses challenges in areas such as minimizing data transfer overhead, improving accuracy, and reducing energy consumption compared to traditional von Neumann architectures, with promising results demonstrated in various applications like autonomous racing and image recognition. The potential impact lies in enabling more sustainable and powerful AI systems at the edge and in other resource-constrained environments.