RRAM fabric for neuromorphic and reconfigurable compute-in-memory systems

We are on the cusp of a new computing era with increased reliance on artificial intelligence, cognitive processing, and big-data analysis. These tasks demand faster and more power-efficient hardware, which may, in turn, be driven by architecture and device innovations. RRAM is widely viewed as a promising candidate that can meet future storage and computing needs. In this paper, we discuss potential computing applications enabled by RRAM devices within both conventional and emerging computing paradigms. A concept of RRAM-based Memory Processing Unit (MPU), which allows the hardware resources to be dynamically allocated to optimally process a broad range of tasks, will be introduced.