Neural networks generally require significant memory capacity/bandwidth to store/access a large number of synaptic weights. This paper presents design of an energy-efficient neural network inference engine based on adaptive weight compression using a JPEG image encoding algorithm. To maximize compression ratio with minimum accuracy loss, the quality factor of the JPEG encoder is adaptively controlled depending on the accuracy impact of each block. With 1% accuracy loss, the proposed approach achieves 63.4× compression for multilayer perceptron (MLP) and 31.3× for LeNet-5 with the MNIST dataset, and 15.3× for AlexNet and 10.2× for ResNet-50 with ImageNet. The reduced memory requirement leads to higher throughput and lower energy for neural network inference (3× effective memory bandwidth and 22× lower system energy for MLP).

This paper presents, DeepTrain, an embedded platform for high-performance and energy-efficient training of deep neural network (DNN). The key architectural concept of DeepTrain is to develop a spatially homogeneous computing (and memory) fabric with temporally heterogeneous programmable data flows to optimize memory mapping and data reuse during different phases of training operation.The DeepTrain is demonstrated as an in-memory accelerator integrated in the logic layer of a 3-D memory module. A programming model and supporting architecture utilizes the flexible data flow to efficiently accelerate training of various types of DNNs. The cycle level simulation and synthesized design in 15 nm FinFET shows power efficiency of 500 GFLOPS/W, and almost similar throughput for a wide range of DNNs, including convolutional, recurrent, and mixed (CNN+RNN) networks.

Neural networks generally require significant memory capacity/bandwidth to store/access a large number of synaptic weights. This paper presents an application of JPEG image encoding to compress the weights by exploiting the spatial locality and smoothness of the weight matrix. To minimize the loss of accuracy due to JPEG encoding, we propose to adaptively control the quantization factor of the JPEG algorithm depending on the error-sensitivity (gradient) of each weight. With the adaptive compression technique, the weight blocks with higher sensitivity are compressed less for higher accuracy. The adaptive compression reduces memory requirement, which in turn results in higher performance and lower energy of neural network hardware. The simulation for inference hardware for multilayer perceptron with the MNIST dataset shows up to 42X compression with less than 1% loss of recognition accuracy, resulting in 3X higher effective memory bandwidth and ~19X lower system energy.

The fast and energy-efficient simulation of dynamical systems defined by coupled ordinary/partial differential equations has emerged as an important problem. The accelerated simulation of coupled ODE/PDE is critical for analysis of physical systems as well as computing with dynamical systems. This paper presents a fast and programmable accelerator for simulating dynamical systems. The computing model of the proposed platform is based on multilayer cellular nonlinear network (CeNN) augmented with nonlinear function evaluation engines. The platform can be programmed to accelerate wide classes of ODEs/PDEs by modulating the connectivity within the multilayer CeNN engine. An innovative hardware architecture including data reuse, memory hierarchy, and near-memory processing is designed to accelerate the augmented multilayer CeNN. A dataflow model is presented which is supported by optimized memory hierarchy for efficient function evaluation. The proposed solver is designed and synthesized in 15nm technology for the hardware analysis. The performance is evaluated and compared to GPU nodes when solving wide classes of differential equations and the power consumption is analyzed to show orders of magnitude improvement in energy efficiency.

This paper presents a programmable and scalable digital neuromorphic architecture based on 3D high-density memory integrated with logic tier for efficient neural computing. The proposed architecture consists of clusters of processing engines, connected by 2D mesh network as a processing tier, which is integrated in 3D with multiple tiers of DRAM. The PE clusters access multiple memory channels (vaults) in parallel. The operating principle, referred to as the memory centric computing, embeds specialized state-machines within the vault controllers of HMC to drive data into the PE clusters. The paper presents the basic architecture of the Neurocube and an analysis of the logic tier synthesized in 28nm and 15nm process technologies. The performance of the Neurocube is evaluated and illustrated through the mapping of a Convolutional Neural Network and estimating the subsequent power and performance for both training and inference.

This paper presents methodology of feedback-controlled dynamic approximation to enable energy-accuracy trade-off in digital recurrent neural network (RNN). A low-power digital RNN engine is presented that employs the proposed dynamic approximation. The on-chip feedback controller is realized by utilizing hysteretic or proportional controller. The dynamic adaptation of bit-precisions during the RNN computation is selected as approximation approach. Considering various applications, the digital RNN engine designed in 28nm CMOS shows ~36% average energy saving compared to the baseline case, with only ~4% of accuracy degradation on average.

This paper presents the design of tier-to-tier interface circuits for 3-D-ICs, where different tiers may operate at different voltages and/or frequencies. The design and partitioning methodologies for the tier-to-tier interface circuit are discussed. The footprint, power, and performance of the interface are analyzed considering the effects of tier-to-tier process variations in 3-D-ICs and technology scaling. The simulation results show that dividing the interface circuit evenly between two tiers reduces footprint but increases power dissipation. For heterogeneous systems with different voltages for reading and writing tiers, dividing the interface between tiers provides better performance than the worst case scenario. On the other hand, placing the interface circuit in the reading tier maximizes throughput for a homogeneous system where both tiers operate at the same voltage. In advanced CMOS nodes, placing interface circuit in the reading tier is a better option due to high delay of the level shifters.

This paper proposes a power-aware digital feedforward neural network platform that utilizes the backpropagation algorithm during training to enable energy-quality trade-off. Given a quality constraint, the proposed approach identifies a set of synaptic weights for approximation in a neural network. The approach selects synapses with small impact on output error, estimated by the backpropagation algorithm, for approximation. The approximations are achieved by a coupled software (reduced bit-width) and hardware (approximate multiplication in the processing engine) based design approaches. The full-chip design in 130nm CMOS shows, compared to a baseline accurate design, the proposed approach reduces system power by ~38% with 0.4% lower recognition accuracy in a classification problem.

This paper studies the opportunities of energy-accuracy tradeoff in cellular neural network (CNN). Algorithmic characteristics of CNN is coupled with hardware-induced error distribution of a digital CNN cell to evaluate energy-accuracy tradeoff for simple image processing tasks as well as a complex application. The analysis shows that errors modulate the cell dynamics and propagate through the network degrading the output quality and increasing the convergence time. The error propagation is determined by the task being performed by the CNN, specifically, the strength of the feedback template. Controlling precision is observed to be a more effective approach for energy-accuracy tradeoff in CNN than voltage over scaling.

This paper studies charged device model electrostatic discharge (CDM-ESD) events in die stacking process of 3D-ICs and investigates CDM-ESD protection circuits for individual TSVs to prevent high voltage stress on transistor connected to TSV. The models for power, area, delay, and signal integrity of TSVs considering ESD protection are presented. The models are used to drive a methodology to design reliable 3D-ICs considering CDM-ESD while minimizing the overheads. We study the impact of ESD protection on die-to-die asynchronous interface circuit.

Flip-flop has long been used as a sequencing element of choice in ASIC design; commercial synthesis tools have also been developed in this context. This work has been motivated by a question of whether existing CAD tools can be employed from RTL to layout while pulsed latch replaces flip-flop as a sequencing element. Two important problems have been identified and their solutions are proposed: placement of pulse generators and latches for integrity of pulse shape, and design of special scan latches and their selective use to reduce hold violations. A reference design flow has also been set up using published documents, in order to assess the proposed one. In 40-nm technology, the proposed flow achieves 20% reduction in circuit area and 30% reduction in power consumption, on average of 12 test circuits.