Xiaochen Peng

Xiaochen Peng

Auto-generated profile, based on publicly available information

Xiaochen Peng

School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, USA

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
14 December 2020 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
Verified
Journal Article
Closed access
XPXiaochen Peng
Xiaochen Peng
SHShanshi Huang
Shanshi Huang
HJHongwu Jiang
Hongwu Jiang
ALAnni Lu
Anni Lu
SYShimeng Yu
Shimeng Yu
DNN+NeuroSim is an integrated framework to benchmark compute-in-memory (CIM) accelerators for deep neural networks, with hierarchical design options from device-level, to circuit level and up to algorithm level. A python wrapper is developed to interface NeuroSim with a popular machine learning platform: Pytorch, to support flexible network structures. The framework provides automatic algorithm-to-hardware mapping, and evaluates chip-level area, energy efficiency and throughput for training or inference, as well as training/inference accuracy with hardware constraints. Our prior inference version of DNN+NeuroSim framework available at https://github.com/neurosim/DNN_NeuroSim_V1.2 was developed to estimate the impact of reliability in synaptic devices, and analog-to-digital converter (ADC) quantization loss on the accuracy and hardware performance of an inference engine. In this work, we further investigated the impact of the “analog” emerging nonvolatile memory (eNVM)’s nonideal device properties for on-chip training. By introducing the nonlinearity, asymmetry, device-to-device and cycle-to-cycle variation of weight update into the python wrapper, and peripheral circuits for error/weight gradient computation in NeuroSim core, we benchmarked CIM accelerators based on state-of-the-art SRAM and eNVM devices for VGG-8 on CIFAR-10 dataset, revealing the crucial specs of synaptic devices for on-chip training. The latest training version of the DNN+NeuroSim framework is available at https://github.com/neurosim/DNN_NeuroSim_V2.1 .

16 April 2021 in IEEE Transactions on Circuits and Systems I: Regular Papers
Verified
Journal Article
Closed access
SYShimeng Yu
Shimeng Yu
WSWonbo Shim
Wonbo Shim
XPXiaochen Peng
Xiaochen Peng
YLYandong Luo
Yandong Luo
To efficiently deploy machine learning applications to the edge, compute-in-memory (CIM) based hardware accelerator is a promising solution with improved throughput and energy efficiency. Instant-on inference is further enabled by emerging non-volatile memory technologies such as resistive random access memory (RRAM). This paper reviews the recent progresses of the RRAM based CIM accelerator design. First, the multilevel states RRAM characteristics are measured from a test vehicle to examine the key device properties for inference. Second, a benchmark is performed to study the scalability of the RRAM CIM inference engine and the feasibility towards monolithic 3D integration that stacks RRAM arrays on top of advanced logic process node. Third, grand challenges associated with in-situ training are presented. To support accurate and fast in-situ training and enable subsequent inference in an integrated platform, a hybrid precision synapse that combines RRAM with volatile memory (e.g. capacitor) is designed and evaluated at system-level. Prospects and future research needs are discussed.

25 December 2019 in IEEE Transactions on Circuits and Systems I: Regular Papers
Verified
Journal Article
Closed access
XPXiaochen Peng
Xiaochen Peng
RLRui Liu
Rui Liu
SYShimeng Yu
Shimeng Yu
Recent state-of-the-art deep convolutional neural networks (CNNs) have shown remarkable success in current intelligent systems for various tasks, such as image/speech recognition and classification. A number of recent efforts have attempted to design custom inference engines based on processing-in-memory (PIM) architecture, where the memory array is used for weighted sum computation, thereby avoiding the frequent data transfer between buffers and computation units. Prior PIM designs typically unroll each 3D kernel of the convolutional layers into a vertical column of a large weight matrix, where the input data needs to be accessed multiple times. In this paper, in order to maximize both weight and input data reuse for PIM architecture, we propose a novel weight mapping method and the corresponding data flow which divides the kernels and assign the input data into different processing-elements (PEs) according to their spatial locations. As a case study, resistive random access memory (RRAM) based 8-bit PIM design at 32 nm is benchmarked. The proposed mapping method and data flow yields ~2.03x speed up and ~1.4x improvement in throughput and energy efficiency for ResNet-34, compared with the prior design based on the conventional mapping method. To further optimize the hardware performance and throughput, we propose an optimal pipeline architecture, with ~50% area overhead, it achieves overall 913x and 1.96x improvement in throughput and energy efficiency, which are 132476 FPS and 20.1 TOPS/W, respectively.

13 March 2020 in IEEE Transactions on Computers
Verified
Journal Article
Closed access
HJHongwu Jiang
Hongwu Jiang
XPXiaochen Peng
Xiaochen Peng
SHShanshi Huang
Shanshi Huang
SYShimeng Yu
Shimeng Yu
Rapid development in deep neural networks (DNNs) is enabling many intelligent applications. However, on-chip training is challenging due to the extensive computation and memory bandwidth requirements. To overcome the memory wall, compute-in-memory (CIM) approach exploits the analog computation along the bitline of the memory array thus significantly speeds up the vector-matrix multiplications (VMM). So far, most CIM based architectures target at implementing inference engine for offline training only. In this work, we propose CIMAT, a CIM Architecture for Training. We design 7T and 8T transpose SRAM to implement bi-directional VMM that is needed for feedforward (FF) and backprogpagation (BP). Moreover, we design the periphery circuity, the data flow and optimized pipeline to support entire on-chip training process based on CIM. We utilize 7nm technology to design CIMAT architecture and explore the 8-bit training performance of ImageNet on ResNet-18, showing that 7T based design can achieve 3.38× energy efficiency (∼6.02 TOPS/W), 4.34× frame rate (∼4,020 fps) and only 50% chip size compared to the baseline architecture with conventional 6T SRAM array that supports row-by-row read only. The even better performance is obtained with 8T based architecture, which can reach ∼10.79 TOPS/W and ∼48,335 fps with 74% chip area compared the baseline.

26 March 2021 in IEEE Transactions on Circuits and Systems II: Express Briefs
Verified
Journal Article
Closed access
JMJian Meng
Jian Meng
LYLi Yang
Li Yang
XPXiaochen Peng
Xiaochen Peng
SYShimeng Yu
Shimeng Yu
DFDeliang Fan
Deliang Fan
JSJae-Sun Seo
Jae-Sun Seo
The high computational complexity and a large number of parameters of deep neural networks (DNNs) become the most intensive burden of deep learning hardware design, limiting efficient storage and deployment. With the advantage of high-density storage, non-volatility, and low energy consumption, resistive RAM (RRAM) crossbar based in-memory computing (IMC) has emerged as a promising technique for DNN acceleration. To fully exploit crossbar-based IMC efficiency, a systematic compression design that considers both hardware and algorithm is necessary. In this brief, we present a system-level design considering the low precision weight and activation, structured pruning, and RRAM crossbar mapping. The proposed multi-group Lasso algorithm and hardware implementations have been evaluated on ResNet/VGG models for CIFAR-10/ImageNet datasets. With the fully quantized 4-bit ResNet-18 for CIFAR-10, we achieve up to 65.4× compression compared to full-precision software baseline, and 7× energy reduction compared to the 4-bit unpruned RRAM IMC hardware with 1.1% accuracy loss. For the fully quantized 4-bit ResNet-18 model for ImageNet dataset, we achieve up to 10.9× structured compression with 1.9% accuracy degradation.

19 March 2021 in IEEE Transactions on Electron Devices
Verified
Journal Article
Closed access
JNJinhyun Noh
Jinhyun Noh
Hagyoul BaeHagyoul Bae
Hagyoul Bae
JLJunkang Li
Junkang Li
YLYandong Luo
Yandong Luo
YQYiming Qu
Yiming Qu
PYPeide D. Ye
Peide D. Ye
We have experimentally demonstrated robust beta-gallium oxide ( $\beta $ -Ga 2 O 3 ) ferroelectric (FE) field-effect transistors (FeFETs) on a sapphire substrate operated up to 400 °C. Atomic layer deposited (ALD) Hf 0.5 Zr 0.5 O 2 [hafnium zirconium oxide (HZO)] is used as the FE dielectric. The HZO/ $\beta $ -Ga 2 O 3 FeFETs are studied for their synaptic behavior applications at elevated temperatures. The devices show distinguishable polarization switching operation with the output conductance quasi-linearly controlled by the number of input pulses on the FE gate. In a simulation, on-chip learning accuracy reaches 94% at elevated temperatures using the Modified National Institute of Standards and Technology (MNIST) data set with a simple two-layer multilayer perceptron (MLP) network. These ultra wide bandgap semiconductor devices have the potential to fill the need for harsh environment neuromorphic applications.