Dense crossbar arrays of non-volatile memory (NVM) devices represent one possible path for implementing massively-parallel and highly energy-efficient neuromorphic computing systems. We first review recent advances in the application of NVM devices to three computing paradigms: spiking neural networks (SNNs), deep neural networks (DNNs), and 'Memcomputing'. In SNNs, NVM synaptic connections are updated by a local learning rule such as spike-timing-dependent-plasticity, a computational approach directly inspired by biology. For DNNs, NVM arrays can represent matrices of synaptic weights, implementing the matrix-vector multiplication needed for algorithms such as backpropagation in an analog yet massively-parallel fashion. This approach could provide significant improvements in power and speed compared to GPU-based DNN training, for applications of commercial significance. We then survey recent research in which different types of NVM devices-including phase change memory, conductive-bridging RAM, filamentary and nonfilamentary RRAM, and other NVMs-have been proposed, either as a synapse or as a neuron, for use within a neuromorphic computing application. The relevant virtues and limitations of these devices are assessed, in terms of properties such as conductance dynamic range, (non)linearity and (a)symmetry of conductance response, retention, endurance, required switching power, and device variability.

Neuromorphic computing has emerged as a promising avenue towards building the next generation of intelligent computing systems. It has been proposed that memristive devices, which exhibit history-dependent conductivity modulation, could efficiently represent the synaptic weights in artificial neural networks. However, precise modulation of the device conductance over a wide dynamic range, necessary to maintain high network accuracy, is proving to be challenging. To address this, we present a multi-memristive synaptic architecture with an efficient global counter-based arbitration scheme. We focus on phase change memory devices, develop a comprehensive model and demonstrate via simulations the effectiveness of the concept for both spiking and non-spiking neural networks. Moreover, we present experimental results involving over a million phase change memory devices for unsupervised learning of temporal correlations using a spiking neural network. The work presents a significant step towards the realization of large-scale and energy-efficient neuromorphic computing systems.

We fabricated and characterized new ambipolar silicon nanowire (SiNW) FET transistors featuring two independent gate-all-around electrodes and vertically stacked SiNW channels. One gate electrode enables dynamic configuration of the device polarity (n or p-type), while the other switches on/off the device. Measurement results on silicon show I on /I off > 10 6 and S 64mV/dec (70mV/dec) for p(n)-type operation in the same device. We show that XOR operation is embedded in the device characteristic, and we demonstrate for the first time a fully functional 2-transistor XOR gate.

The presence of a direct band gap and an ultrathin form factor has caused a considerable interest in two-dimensional (2D) semiconductors from the transition metal dichalcogenides (TMD) family with molybdenum disulfide (MoS2) being the most studied representative of this family of materials. While diverse electronic elements, logic circuits, and optoelectronic devices have been demonstrated using ultrathin MoS2, very little is known about their performance at high frequencies where commercial devices are expected to function. Here, we report on top-gated MoS2 transistors operating in the gigahertz range of frequencies. Our devices show cutoff frequencies reaching 6 GHz. The presence of a band gap also gives rise to current saturation, allowing power and voltage gain, all in the gigahertz range. This shows that MoS2 could be an interesting material for realizing high-speed amplifiers and logic circuits with device scaling expected to result in further improvement of performance. Our work represents the first step in the realization of high-frequency analog and digital circuits based on 2D semiconductors.

No abstract

Abstract-Technology scaling has caused the feature sizes to shrink continuously, whereas interconnects, unlike transistors, have not followed the same trend. Designing 3D stack architectures is a recently proposed approach to overcome the power consumption and delay problems associated with the interconnects by reducing the length of the wires going across the chip. However, 3D integration introduces serious thermal challenges due to the high power density resulting from placing computational units on top of each other. In this work, we first investigate how the existing thermal management, power management and job scheduling policies affect the thermal behavior in 3D chips. We then propose a dynamic thermally-aware job scheduling technique for 3D systems to reduce the thermal problems at very low performance cost. Our approach can also be integrated with power management policies to reduce energy consumption while avoiding the thermal hot spots and large temperature variations.

Abstract-This paper presents a novel approach for implementing ultra-low-power digital components and systems using source-coupled logic (SCL) circuit topology, operating in weak inversion (subthreshold) regime. Minimum size pMOS transistors with shorted drain-substrate contacts are used as gate-controlled, very high resistivity load devices. Based on the proposed approach, the power consumption and the operation frequency of logic circuits can be scaled down linearly by changing the tail bias current of SCL gates over a very wide range spanning several orders of magnitude, which is not achievable in subthreshold CMOS circuits. Measurements in conventional 0.18 m CMOS technology show that the tail bias current of each gate can be set as low as 10 pA, with a supply voltage of 300 mV, resulting in a power-delay product of less than 1 fJ. Fundamental circuits such as ring oscillators and frequency dividers, as well as more complex digital blocks such as parallel multipliers designed by using the STSCL topology have been experimentally characterized.

No abstract