Novel Hybrid Computing Architecture with Memristor-Based Processing-in-Memory for Data-Intensive Applications

In this work, a novel hybrid computing architecture with memristor-based processing-in-memory (MPiM) is proposed to resolve the memory-wall issue for data-intensive applications. The datapath and control logic are redesigned to incorporate MPiM. Thus, some data can be processed in the memory array or the specific logic near the memory array. So the data to be transferred over the bus can be reduced considerably, saving power as well as improving the performance. Evaluations on data-intensive applications show reductions of 80% in energy consumption and 75% reduction in processing latency.