Memristive devices for computation-in-memory

Abstract

CMOS technology and its continuous scaling have made electronics and computers accessible and affordable for almost everyone on the globe; in addition, they have enabled the solutions of a wide range of societal problems and applications. Today, however, both the technology and the computer architectures are facing severe challenges/walls making them incapable of providing the demanded computing power with tight constraints. This motivates the need for the exploration of novel architectures based on new device technologies; not only to sustain the financial benefit of technology scaling, but also to develop solutions for extremely demanding emerging applications. This paper presents two computation-in-memory based accelerators making use of emerging memristive devices; they are Memristive Vector Processor and RRAM Automata Processor. The preliminary results of these two accelerators show significant improvement in terms of latency, energy and area as compared to today's architectures and design.

Time-division Multiplexing Automata Processor

“…MAP is one of the most successful hardware implementations of Non-deterministic Finite Automata (NFA), whose high efficiency has been proved by many researches [2][3][4][5][6][7]12]. Recent works such as Cache Automaton [16] and RRAM-AP [17] intend to improve MAP by using different memory technologies while maintaining its basic structure. These designs have the following features in common:…”

“…Therefore, this pipeline stage is the bottleneck that limits Cache Automaton's speed and throughput. RRAM-AP concept [17] shows the potential of building an automata accelerator using Resistive Random Access Memory (RRAM) arrays, which are even faster than SRAM arrays. However, a complete design was not given.…”

“…In this paper, we further improve the throughput of automata accelerators. The main contributions of this paper are: Figure 1: General architecture of Automata Processors [17].…”

Architectures and automation for beyond-CMOS technologies

“…Also, efficient neuromorphic circuits have been designed using ReRAM devices [69][70][71][72], that use analog operations on crossbar arrays. Analog matrix operations have also been realized using ReRAM crossbar arrays [73,74]. Furthermore, it has been demonstrated that traditional applications such as image processing can also benefit from logic-inmemory computations using ReRAM [75].…”

“…Furthermore, it has been demonstrated that traditional applications such as image processing can also benefit from logic-inmemory computations using ReRAM [75]. Novel application specific processor for automata processing have also been realized using crossbar arrays [74]. realize material implication and suggested their use for logic-in-memory operations, but no architecture was proposed [79].…”