Runze Han
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Runze Han has a total of 49 co-authors. Here are the top co-authors by number of shared publications:
- Jinfeng Kang
Peking University
12 shared publications Xiaoyan Liu
School of Integrated Circuits Peking University Beijing 100871 China
12 shared publicationsZheng Zhou
School of Integrated Circuits, Peking University, Beijing, China
9 shared publicationsPeng Huang
School of Integrated Circuits Peking University Beijing 100871 China
8 shared publicationsLifeng Liu
Institute of Microelectronics, Peking University, Beijing 100871, China
7 shared publications- Peng Huang
Peking University
6 shared publications
Latest Publications
- Open Access
- 8 November 2022
- Published by Wiley in Advanced Intelligent Systems
- Vol. 4 (12)
Floating Gate Transistor‐Based Accurate Digital In‐Memory Computing for Deep Neural Networks
To improve the computing speed and energy efficiency of deep neural network (DNN) applications, in‐memory computing with nonvolatile memory (NVM) is proposed to address the time‐consuming and energy‐hungry data shuttling issue. Herein, a digital in‐memory computing method for convolution computing, which holds the key to DNNs, is proposed. Based on the proposed method, a floating gate transistor‐based in‐memory computing chip for accurate convolution computing with high parallelism is created. The proposed digital in‐memory computing method can achieve the central processing unit (CPU)‐equivalent precision with the same neural network architecture and parameters, different from the analogue or digital–analogue‐mixed in‐memory computing techniques. Based on the fabricated floating gate transistor‐based in‐memory computing chip, a hardware LeNet‐5 neural network is built. The chip achieves 96.25% accuracy on the full Modified National Institute of Standards and Technology database, which is the same as the result computed by the CPU with the same neural network architecture and parameters.
- 8 November 2022
- Published by Wiley in Advanced Intelligent Systems
- Vol. 4 (12)
Open AccessFloating Gate Transistor‐Based Accurate Digital In‐Memory Computing for Deep Neural Networks
To improve the computing speed and energy efficiency of deep neural network (DNN) applications, in‐memory computing with nonvolatile memory (NVM) is proposed to address the time‐consuming and energy‐hungry data shuttling issue. Herein, a digital in‐memory computing method for convolution computing, which holds the key to DNNs, is proposed. Based on the proposed method, a floating gate transistor‐based in‐memory computing chip for accurate convolution computing with high parallelism is created. The proposed digital in‐memory computing method can achieve the central processing unit (CPU)‐equivalent precision with the same neural network architecture and parameters, different from the analogue or digital–analogue‐mixed in‐memory computing techniques. Based on the fabricated floating gate transistor‐based in‐memory computing chip, a hardware LeNet‐5 neural network is built. The chip achieves 96.25% accuracy on the full Modified National Institute of Standards and Technology database, which is the same as the result computed by the CPU with the same neural network architecture and parameters.
- 2 June 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 69 (7) , 3708-3715
Mixed-Precision Partial Differential Equation Solver Design Based on Nonvolatile Memory
In this study, a nonvolatile memory (NVM)-based mixed-precision partial differential equation (PDE) solver is developed to achieve both high computational accuracy and high energy efficiency in a non-Von Neumann manner. The mixed-precision architecture consists of an outer solver utilizing CPU/GPU for a small amount of high-precision computing and an inner solver using NVM array for a large number of approximate iterations by in-memory computing. The impacts of inner iteration number and inner precision on time and energy cost are analyzed to configure the optimum algorithm. Based on this, we investigate the impacts of device reliability properties on the performance of mixed-precision computing system, including device variation, read noise, ON/OFF ratio, and read disturb. The results indicate that the computational accuracy is highly tolerant to the device variation. However, the time and energy cost are increased due to the device reliability, especially the ON/OFF ratio. To suppress the impact of ON/OFF ratio, we present a circuit compensation method, which subtracts the current of reference string with all devices OFF before quantization. To improve the hardware utilization and reduce the number of iterations, a matrix slicing method combined with hybrid layer update is proposed. By employing the two optimization methods, a$57.3\times $ reduction at energy cost with the matrix size of$1600\times1600$ is achieved.- 2 June 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 69 (7) , 3708-3715
Mixed-Precision Partial Differential Equation Solver Design Based on Nonvolatile Memory
In this study, a nonvolatile memory (NVM)-based mixed-precision partial differential equation (PDE) solver is developed to achieve both high computational accuracy and high energy efficiency in a non-Von Neumann manner. The mixed-precision architecture consists of an outer solver utilizing CPU/GPU for a small amount of high-precision computing and an inner solver using NVM array for a large number of approximate iterations by in-memory computing. The impacts of inner iteration number and inner precision on time and energy cost are analyzed to configure the optimum algorithm. Based on this, we investigate the impacts of device reliability properties on the performance of mixed-precision computing system, including device variation, read noise, ON/OFF ratio, and read disturb. The results indicate that the computational accuracy is highly tolerant to the device variation. However, the time and energy cost are increased due to the device reliability, especially the ON/OFF ratio. To suppress the impact of ON/OFF ratio, we present a circuit compensation method, which subtracts the current of reference string with all devices OFF before quantization. To improve the hardware utilization and reduce the number of iterations, a matrix slicing method combined with hybrid layer update is proposed. By employing the two optimization methods, a$57.3\times $ reduction at energy cost with the matrix size of$1600\times1600$ is achieved.- Open Access
- 29 April 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 67 (6) , 2329-2335
Efficient and Robust Spike-Driven Deep Convolutional Neural Networks Based on NOR Flash Computing Array
In this article, we propose an efficient and robust spike-driven convolutional neural network (SCNN) based on the nor flash computing array (NFCA), which is mapped by the pretrained convolutional neural network with the same structure. The spike-driven system eliminates the additional analog-to-digital/digital-to-analog (AD/DA) conversion in the NFCA-based CNN. To study the performance of the hardware implementation, an NFCA-based SCNN for the recognition of the Mixed National Institute of Standards and Technology (MNIST) data set is simulated. Simulation results illustrate that the system achieves 97.94% accuracy with the computing speed of 1 x 10⁶ frame per second (fps). Compared with the typical mixed-signal NFCA-based CNN, the NFCA-based SCNN saves 97% area and 56% energy consumption. Moreover, the NFCA-based SCNN demonstrates great robustness to 30% image noise with less than 2% accuracy loss. The impact of random telegraph noise (RTN) is also greatly reduced in which less than 1% accuracy decrease can be achieved at the 32-nm technology node.
- 29 April 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 67 (6) , 2329-2335
Open AccessEfficient and Robust Spike-Driven Deep Convolutional Neural Networks Based on NOR Flash Computing Array
In this article, we propose an efficient and robust spike-driven convolutional neural network (SCNN) based on the nor flash computing array (NFCA), which is mapped by the pretrained convolutional neural network with the same structure. The spike-driven system eliminates the additional analog-to-digital/digital-to-analog (AD/DA) conversion in the NFCA-based CNN. To study the performance of the hardware implementation, an NFCA-based SCNN for the recognition of the Mixed National Institute of Standards and Technology (MNIST) data set is simulated. Simulation results illustrate that the system achieves 97.94% accuracy with the computing speed of 1 x 10⁶ frame per second (fps). Compared with the typical mixed-signal NFCA-based CNN, the NFCA-based SCNN saves 97% area and 56% energy consumption. Moreover, the NFCA-based SCNN demonstrates great robustness to 30% image noise with less than 2% accuracy loss. The impact of random telegraph noise (RTN) is also greatly reduced in which less than 1% accuracy decrease can be achieved at the 32-nm technology node.
- 27 December 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 67 (2) , 469-473
An Improved RRAM-Based Binarized Neural Network With High Variation-Tolerated Forward/Backward Propagation Module
Binarized neural network (BNN) enables resistive switching random access memory (RRAM) with high nonlinearity and nonsymmetry to realize online training, using an RRAM & comparator structure. In this work, a new hardware implementation approach is proposed to improve the efficiency of BNN. In the approach, an 1T1R array-based propagation module is introduced and designed to realize the computing acceleration of fully parallel vector-matrix multiplication (VMM) in both forward and backward propagations. Using the 1T1R-based propagation module, high computing efficiency is achieved in both training and inference tasks, improving by 50x and 177x, respectively. To solve the computation error caused by device variation, a novel operation scheme with low gate voltage is proposed. With the operation scheme, the RRAM variation is dramatically suppressed by 74.8% for cycle-to-cycle and 59.9% for device-to-device. It enables high-accuracy VMM calculation and, therefore, achieves 94.7% accuracy with a typical BNN, showing only 0.7% degradation from the ideal variation-free case.
- 27 December 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 67 (2) , 469-473
An Improved RRAM-Based Binarized Neural Network With High Variation-Tolerated Forward/Backward Propagation Module
Binarized neural network (BNN) enables resistive switching random access memory (RRAM) with high nonlinearity and nonsymmetry to realize online training, using an RRAM & comparator structure. In this work, a new hardware implementation approach is proposed to improve the efficiency of BNN. In the approach, an 1T1R array-based propagation module is introduced and designed to realize the computing acceleration of fully parallel vector-matrix multiplication (VMM) in both forward and backward propagations. Using the 1T1R-based propagation module, high computing efficiency is achieved in both training and inference tasks, improving by 50x and 177x, respectively. To solve the computation error caused by device variation, a novel operation scheme with low gate voltage is proposed. With the operation scheme, the RRAM variation is dramatically suppressed by 74.8% for cycle-to-cycle and 59.9% for device-to-device. It enables high-accuracy VMM calculation and, therefore, achieves 94.7% accuracy with a typical BNN, showing only 0.7% degradation from the ideal variation-free case.
- Open Access
- 30 July 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 40 (9) , 1538-1541
Stateful Logic Operations in One-Transistor-One- Resistor Resistive Random Access Memory Array
Nonvolatile and cascadable stateful logic operations are experimentally demonstrated within a 1 k-bit one-transistor-one-resistor (1T1R) resistive random access memory (RRAM) array, where NAND gates serve as the building blocks. A robust dual-gate-voltage operation scheme is proposed. The effects of the transistor ON logic operation and the robustness to device parameter variations are discussed. The parallel 4-bit bitwise XOR operation is experimentally implemented in the 1T1R array by cascading NAND gates. This letter presents a feasible approach to in-memory computing for large-scale circuits.
- 30 July 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 40 (9) , 1538-1541
Open AccessStateful Logic Operations in One-Transistor-One- Resistor Resistive Random Access Memory Array
Nonvolatile and cascadable stateful logic operations are experimentally demonstrated within a 1 k-bit one-transistor-one-resistor (1T1R) resistive random access memory (RRAM) array, where NAND gates serve as the building blocks. A robust dual-gate-voltage operation scheme is proposed. The effects of the transistor ON logic operation and the robustness to device parameter variations are discussed. The parallel 4-bit bitwise XOR operation is experimentally implemented in the 1T1R array by cascading NAND gates. This letter presents a feasible approach to in-memory computing for large-scale circuits.
- Open Access
- 8 January 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 66 (5) , 1692-1703
A Novel Convolution Computing Paradigm Based on NOR Flash Array With High Computing Speed and Energy Efficiency
Convolution is one of the key operations in signal processing and machine learning applications. In this paper, we propose a novel convolution computing paradigm based on the NOR Flash array (NFA) that is capable of executing the 2-D convolution computing in one clock cycle. In order to demonstrate the feasibility and efficiency of the proposed convolution computing paradigm, the feature extraction task on an image with the size of 20x 20 is executed using the NFA structure. We also prove the NOR Flash-driven convolution computing is capable of processing the image with a larger size. This paper presents a new approach to realize convolution computing with high speed and energy efficiency for the signal processing and convolution neural network.
- 8 January 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 66 (5) , 1692-1703
Open AccessA Novel Convolution Computing Paradigm Based on NOR Flash Array With High Computing Speed and Energy Efficiency
Convolution is one of the key operations in signal processing and machine learning applications. In this paper, we propose a novel convolution computing paradigm based on the NOR Flash array (NFA) that is capable of executing the 2-D convolution computing in one clock cycle. In order to demonstrate the feasibility and efficiency of the proposed convolution computing paradigm, the feature extraction task on an image with the size of 20x 20 is executed using the NFA structure. We also prove the NOR Flash-driven convolution computing is capable of processing the image with a larger size. This paper presents a new approach to realize convolution computing with high speed and energy efficiency for the signal processing and convolution neural network.
- 17 April 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 65 (9) , 2726-2738
Design and Hardware Implementation of Neuromorphic Systems With RRAM Synapses and Threshold-Controlled Neurons for Pattern Recognition
In this paper, a hardware-realized neuromorphic system for pattern recognition is presented. The system directly captures images from the environment, and then conducts classification using a single layer neural network. Metal-oxide resistive random access memory (RRAM) is used as electronic synapses, and threshold-controlled neurons are proposed as postsynaptic neurons to save the system area and simplify the operation. In the proposed threshold-controlled neuron, no capacitor is utilized, which contributes to higher integration density. The total energy consumption of RRAM synapses for classifying an example is 0.31μJ on average. The proposed system has been implemented on hardware, and has been experimentally demonstrated to show the capability of pattern recognition.
- 17 April 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 65 (9) , 2726-2738
Design and Hardware Implementation of Neuromorphic Systems With RRAM Synapses and Threshold-Controlled Neurons for Pattern Recognition
In this paper, a hardware-realized neuromorphic system for pattern recognition is presented. The system directly captures images from the environment, and then conducts classification using a single layer neural network. Metal-oxide resistive random access memory (RRAM) is used as electronic synapses, and threshold-controlled neurons are proposed as postsynaptic neurons to save the system area and simplify the operation. In the proposed threshold-controlled neuron, no capacitor is utilized, which contributes to higher integration density. The total energy consumption of RRAM synapses for classifying an example is 0.31μJ on average. The proposed system has been implemented on hardware, and has been experimentally demonstrated to show the capability of pattern recognition.
- 1 April 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-2
A novel RRAM-based adaptive-threshold LIF neuron circuit for high recognition accuracy
A novel leaky integrate-and-fire (LIF) neuron circuit based on the gradual switching in resistive random access memory (RRAM) device is put forward, in which threshold modulation can be achieved. Its threshold modulation and spike generating functions are verified through HSPICE simulation. In unsupervised pattern recognition for handwritten digits in MNIST dataset, its advantage in improving the accuracy (from about 70% to more than 95%) is demonstrated. Benchmarking results indicate that this novel neuron is much faster and can save about 66% area compared to one previously proposed.
- 1 April 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-2
A novel RRAM-based adaptive-threshold LIF neuron circuit for high recognition accuracy
A novel leaky integrate-and-fire (LIF) neuron circuit based on the gradual switching in resistive random access memory (RRAM) device is put forward, in which threshold modulation can be achieved. Its threshold modulation and spike generating functions are verified through HSPICE simulation. In unsupervised pattern recognition for handwritten digits in MNIST dataset, its advantage in improving the accuracy (from about 70% to more than 95%) is demonstrated. Benchmarking results indicate that this novel neuron is much faster and can save about 66% area compared to one previously proposed.
- 6 March 2018
- Published by IOP Publishing in Japanese Journal of Applied Physics
- Vol. 57 (4S) , 04FE05
An energy efficient and high speed architecture for convolution computing based on binary resistive random access memory
In this work we present a novel convolution computing architecture based on metal oxide resistive random access memory (RRAM) to process the image data stored in the RRAM arrays. The proposed image storage architecture shows performances of better speed-device consumption efficiency compared with the previous kernel storage architecture. Further we improve the architecture for a high accuracy and low power computing by utilizing the binary storage and the series resistor. For a 28 × 28 image and 10 kernels with a size of 3 × 3, compared with the previous kernel storage approach, the newly proposed architecture shows excellent performances including: 1) almost 100% accuracy within 20% LRS variation and 90% HRS variation; 2) more than 67 times speed boost; 3) 71.4% energy saving.
- 6 March 2018
- Published by IOP Publishing in Japanese Journal of Applied Physics
- Vol. 57 (4S) , 04FE05
An energy efficient and high speed architecture for convolution computing based on binary resistive random access memory
In this work we present a novel convolution computing architecture based on metal oxide resistive random access memory (RRAM) to process the image data stored in the RRAM arrays. The proposed image storage architecture shows performances of better speed-device consumption efficiency compared with the previous kernel storage architecture. Further we improve the architecture for a high accuracy and low power computing by utilizing the binary storage and the series resistor. For a 28 × 28 image and 10 kernels with a size of 3 × 3, compared with the previous kernel storage approach, the newly proposed architecture shows excellent performances including: 1) almost 100% accuracy within 20% LRS variation and 90% HRS variation; 2) more than 67 times speed boost; 3) 71.4% energy saving.
- 21 February 2018
- Published by IOP Publishing in Japanese Journal of Applied Physics
- Vol. 57 (4S) , 04FE02
A novel ternary content addressable memory design based on resistive random access memory with high intensity and low search energy
A novel ternary content addressable memory (TCAM) design based on resistive random access memory (RRAM) is presented. Each TCAM cell consists of two parallel RRAM to both store and search for ternary data. The cell size of the proposed design is 8F 2 , enable a ∼60× cell area reduction compared with the conventional static random access memory (SRAM) based implementation. Simulation results also show that the search delay and energy consumption of the proposed design at the 64-bit word search are 2 ps and 0.18 fJ/bit/search respectively at 22 nm technology node, where significant improvements are achieved compared to previous works. The desired characteristics of RRAM for implementation of the high performance TCAM search chip are also discussed.
- 21 February 2018
- Published by IOP Publishing in Japanese Journal of Applied Physics
- Vol. 57 (4S) , 04FE02
A novel ternary content addressable memory design based on resistive random access memory with high intensity and low search energy
A novel ternary content addressable memory (TCAM) design based on resistive random access memory (RRAM) is presented. Each TCAM cell consists of two parallel RRAM to both store and search for ternary data. The cell size of the proposed design is 8F 2 , enable a ∼60× cell area reduction compared with the conventional static random access memory (SRAM) based implementation. Simulation results also show that the search delay and energy consumption of the proposed design at the 64-bit word search are 2 ps and 0.18 fJ/bit/search respectively at 22 nm technology node, where significant improvements are achieved compared to previous works. The desired characteristics of RRAM for implementation of the high performance TCAM search chip are also discussed.