Memristor-Based Computing Architecture: Design Methodologies and Circuit Techniques.

摘要

Scientists have dreamed of information systems with cognitive human- like skills for years. However, constrained by the device characteristics and rapidly increasing design complexity under the traditional processing technology, little progress has been made in hardware implementation. The recently popularized memristor offers a potential break-through for neuromorphic computing because of its unique properties including nonvolatility extremely high fabrication density, and sensitivity to historic voltage/current behavior. In this project, we first investigated the memristor-based synapse design and the corresponding training scheme. Then, the design optimization and its implementation in multi-synapse systems were analyzed too. With the aid of a sharing training circuit and self-training mode, the performance and energy can be significantly improved. At last, a case study of an arithmetic logic unit (ALU) was designed to demonstrate the hardware implementation of a reconfigurable system built based on memristor synapses. All the circuit design, simulation, layout, and functionality verifications have been completed.