High-Throughput Pattern Matching With CMOL FPGA Circuits: Case for Logic-in-Memory Computing

Abstract

In this paper, we propose a novel CMOS+ MOLecular (CMOL) field-programmable gate array (FPGA) circuit architecture to perform massively parallel, high-throughput computations, which is especially useful for pattern matching tasks and multidimensional associative searches. In the new architecture, patterns are stored as resistive states of emerging nonvolatile memory nanodevices, while the analyzed data are streamed via CMOS subsystem. The main improvements over prior work offered by the proposed circuits are increased nanodevice utilization and, as a result, substantially higher throughput, which is demonstrated by a detailed analysis of the implementation of pattern matching task on the new architecture. For example, our estimates show that the proposed CMOL FPGA circuits based on the 22-nm CMOS technology and one crossbar layer with 22-nm nanowire half-pitch allows up to 12.5% average nanodevice utilization, i.e., the fraction of the devices turned to the high conductive state, as compared to a typical ∼0.1% of the original CMOL FPGA circuits. This in turn enables throughput close to 7.1 × 10 16 bits/s/cm 2 at ∼ 1 fJ/bit energy efficiency, for matching of ∼ 10 7 250-bit patterns stored locally on a 1 cm 2 chip. These numbers are at least 2 orders of magnitude better throughput as compared to that of other state-of-the-art FPGA methods, and begin to approach ternary content-addressable memory-like performance at similar CMOS technology nodes. More generally, we argue that the proposed concept combines the versatility of reconfigurable architectures and density of the associative memories. It can be viewed as a very tight symbiotic integration of memory and logic functions for high-performance logic-in-memory computing. Index Terms-CMOS+MOLecular (CMOL), fieldprogrammable gate array (FPGA), hybrid circuits, logicin-memory computing, memristor, pattern matching, ReRAM, resistive switching, ternary content-addressable memory (TCAM). I. INTRODUCTION R ECONFIGURABLE circuits (RCs) are very efficient for information processing tasks [1], such as image and signal processing (e.g., filtering, edge detection, coding, and

Scalable Method to Find the Shortest Path in a Graph with Circuits of Memristors

“…31, has been shown to give high densities of connections, which could be useful here. 32,33 Building one physical system capable of implementing any graph problem is unrealistic. However, it would be possible to have specialized chips for types of graph problems.…”

Associative memories using complex-valued Hopfield networks based on spin-torque oscillator arrays

“…Additional parasitics in the time delay network, such the capacitance of the memristor, would increase the energy cost of producing delays while also affecting the range of achievable delays. However, typical memristor parasitic capacitances are a fraction of the gate capacitances used in this study [58,59] and we estimate that the achievable delays per RC block of the time delay network decrease by less than fifteen percent. Such reductions in the achievable delay ranges could be readily compensated by the addition of additional delay blocks or by allowing a larger resistance values for each memristor.…”

Resistive switching memory for high density storage and computing*

“…There are various challenges in 3D passive array designing, such as the sneaking current, disturbance, IR drop, thermal coupling, etc, which greatly retard the progress towards commercialization. Apart from data storage, the passive network also has other potential applications, such as neuromorphic computing, [23] 3D CMOS/molecular circuits (CMOL), [24] etc., with promising prospect of low cost and high energy efficiency. In this review article, we will firstly address the associative problems in passive array and 3D architectures (Section 2), and then review the states of the art of various selector devices and selfselective cells (Section 3).…”