Opportunities and Challenges of Resistive RAM for Neuromorphic Applications

摘要

Many technological concepts of Resistive RAM (RRAM) have been demonstrated in recent years. Simultaneously, a paradigm shift towards integrating computational functions into memory has been observed, with neuromorphic networks as one example. We discuss where and how RRAM might come in as a potential functional device for developing hardware accelerators for neuromorphic applications. We also discuss the opportunities for using RRAM as a true computational element in an integrated in-memory computing concept. Understanding the dynamic statistical properties of RRAM is the key challenge to unlock the potential of these devices for on-chip artificial intelligence.