1 July 2021 in IEEE Transactions on Circuits and Systems I: Regular Papers
In this paper, we present a novel approximate computing scheme suitable for realizing the energy-efficient multiply-accumulate (MAC) processing. In contrast to the prior works that suffer from the error accumulation limiting the approximate range, we utilize different approximate multipliers in an interleaved way to compensate errors in the opposite direction during accumulate operations. For the balanced error accumulation, we first design the approximate 4-2 compressors generating errors in the opposite direction while minimizing the computational costs. Based on the probabilistic analysis, positive and negative multipliers are then carefully developed to provide a similar error distance. Simulation results on various practical applications reveal that the proposed MAC processing offers the energy-efficient computing scenario by extending the range of approximate parts. Even compared to the state-of-the-art solutions, for example, the proposed interleaving scheme relaxes the core-level energy consumption of the recent CNN accelerator by more than 35% without degrading the recognition accuracy.
6 March 2023 in IEEE Transactions on Computers
When training deep neural networks (DNNs), expensive floating point arithmetic units are used in GPUs or custom neural processing units (NPUs). To reduce the burden of floating point arithmetic, community has started exploring the use of more efficient data representations, e.g., block floating point (BFP). The BFP format allows a group of values to share an exponent, which effectively reduces the memory footprint and enables cheaper fixed point arithmetic for multiply-accumulate (MAC) operations. However, existing BFP-based DNN accelerators are targeted for a specific precision, making them less versatile. In this paper, we present FlexBlock, a DNN training accelerator with three BFP modes, possibly different among activation, weight, and gradient tensors. By configuring FlexBlock to a lower BFP precision, the number of MACs handled by the core increases by up to 4× in 8-bit mode or 16× in 4-bit mode compared to 16-bit mode. To reach this theoretical upper bound, FlexBlock maximizes the core utilization at various precision levels or layer types, and allows dynamic precision control to keep throughput at its peak without sacrificing training accuracy. We evaluate the effectiveness of FlexBlock using representative DNNs on CIFAR, ImageNet and WMT14 datasets. As a result, training in FlexBlock significantly improves training speed by 1.5 $\sim 5.3\times$ and energy efficiency by 2.4 $\sim 7.0\times$ compared to other training accelerators.
27 October 2022 in IEEE Transactions on Circuits and Systems II: Express Briefs
In this brief, we present a novel design methodology of cost-effective approximate radix-4 Booth multipliers, which can significantly reduce the power consumption of error-resilient signal processing tasks. In contrast that the prior studies only focus on the approximation of either the partial product generation with encoders or the partial product reductions with compressors, the proposed method considers two major processing steps jointly by forcing the generated error directions to be opposite to each other. As the internal errors are naturally balanced to have zero mean, as a result, the proposed approximate Booth multiplier can minimize the required processing energy under the same number of approximate bits compared to the previous designs. Simulation results on FIR filtering and image classification applications reveal that the proposed approximate Booth multiplier shows the most attractive energy-performance trade-offs, achieving 28% and 34% of energy reduction compared to the exact Booth multiplier, respectively, with negligible accuracy loss.
8 June 2017 in Journal of Signal Processing Systems
16 April 2024 in IEEE Transactions on Emerging Topics in Computing
As spiking neural networks (SNNs) are event-driven, energy efficiency is higher than conventional artificial neural networks (ANNs). Since SNN delivers data through discrete spikes, it is difficult to use gradient methods for training, limiting its accuracy. To keep the accuracy of SNNs similar to ANN counterparts, pre-trained ANNs are converted to SNNs (ANN-to-SNN conversion). During the conversion, encoding activations of ANNs to a set of spikes in SNNs is crucial for minimizing the conversion loss. In this work, we propose a single-spike phase coding as an encoding scheme that minimizes the number of spikes to transfer data between SNN layers. To minimize the encoding error due to single-spike approximation in phase coding, threshold shift and base manipulation are proposed. Without any additional retraining or architectural constraints on ANNs, the proposed conversion method does not lose inference accuracy (0.58% on average) verified on three convolutional neural networks (CNNs) with CIFAR and ImageNet datasets. In addition, graph convolutional networks (GCNs) are converted to SNNs successfully with an average accuracy loss of 0.90%. Most importantly, the energy efficiency of our SNN improves by 4.6 $\sim 17.3\times$ compared to the ANN baseline.