Leibin Ni
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Leibin Ni has a total of 22 co-authors. Here are the top co-authors by number of shared publications:
Hao Yu
School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798
4 shared publicationsHantao Huang
School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798
3 shared publicationsMenghui Xu
Huawei Technologies Co., Ltd
2 shared publications- Ruicheng Dai
Shanghai Jiao Tong University
2 shared publications - Weikang Qian
Shanghai Jiao Tong University
2 shared publications Xiaolong Shen
Huawei Technologies Co., Ltd
2 shared publications
Latest Publications
- 27 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 44 (5) , 1620-1633
AccALS 2.0: Accelerating Approximate Logic Synthesis by Simultaneous Selection of Multiple Local Approximate Changes
Approximate computing emerges as an energy-efficient computing paradigm designed for applications that can tolerate errors. Many iterative methods for approximate logic synthesis (ALS) have been developed to automatically synthesize approximate circuits. Nonetheless, most of them overlook the potential of applying multiple local approximate changes (LACs) simultaneously in one iteration, which can significantly reduce the overall computation time. In this paper, we propose AccALS 2.0, a novel framework for further Accelerating iterative ALS flows, which is based on simultaneous selection of multiple LACs in a single round. However, there are two challenges for selecting multiple LACs. The first is that the mutual influence of multiple LACs can affect the estimation of the circuit error. The second is that there may exist conflicts among multiple LACs. To address these issues, first, we propose an efficient measure for the mutual influence between two LACs. With its help, we transform the problems of solving the LAC conflicts and selecting multiple LACs into a unified maximum independent set problem for solving. The experimental results showed that AccALS 2.0 outperforms state-of-the-art ALS methods in runtime, while achieving similar or better circuit quality.
- 27 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 44 (5) , 1620-1633
AccALS 2.0: Accelerating Approximate Logic Synthesis by Simultaneous Selection of Multiple Local Approximate Changes
Approximate computing emerges as an energy-efficient computing paradigm designed for applications that can tolerate errors. Many iterative methods for approximate logic synthesis (ALS) have been developed to automatically synthesize approximate circuits. Nonetheless, most of them overlook the potential of applying multiple local approximate changes (LACs) simultaneously in one iteration, which can significantly reduce the overall computation time. In this paper, we propose AccALS 2.0, a novel framework for further Accelerating iterative ALS flows, which is based on simultaneous selection of multiple LACs in a single round. However, there are two challenges for selecting multiple LACs. The first is that the mutual influence of multiple LACs can affect the estimation of the circuit error. The second is that there may exist conflicts among multiple LACs. To address these issues, first, we propose an efficient measure for the mutual influence between two LACs. With its help, we transform the problems of solving the LAC conflicts and selecting multiple LACs into a unified maximum independent set problem for solving. The experimental results showed that AccALS 2.0 outperforms state-of-the-art ALS methods in runtime, while achieving similar or better circuit quality.
- Open Access
- 16 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Circuits and Systems
- Vol. 5, 57-68
A Low-Power DNN Accelerator With Mean-Error-Minimized Approximate Signed Multiplier
Approximate computing is an emerging and effective method for reducing energy consumption in digital circuits, which is critical for energy-efficient performance improvement of edge-computing devices. In this paper, we propose a low-power DNN accelerator with novel signed approximate multiplier based on probability-optimized compressor and error compensation. The probability-optimized compressor is customized for partial product matrix (PPM) of signed operands, which gets the optimal logic circuit after probabilistic analysis and optimization. At the same time, we explored the PPM truncation method, found out the impact of different partial product (PP) truncation numbers on circuit benefit and error, and achieved a more ideal performance-error tradeoff through a reasonable error compensation method. In the optimal case of 8 bits, the proposed approximate multiplier saves 49.84% power, 46.41% area and 24.65% delay compared to the exact multiplier. We employed the proposed approximate multiplier in the vector systolic array as the processing element (PE). Under the VGG-16 evaluation, the proposed accelerator achieves performance improvement of energy efficiency $1.96\times $ , while the error loss was only 0.95%.
- 16 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Circuits and Systems
- Vol. 5, 57-68
Open AccessA Low-Power DNN Accelerator With Mean-Error-Minimized Approximate Signed Multiplier
Approximate computing is an emerging and effective method for reducing energy consumption in digital circuits, which is critical for energy-efficient performance improvement of edge-computing devices. In this paper, we propose a low-power DNN accelerator with novel signed approximate multiplier based on probability-optimized compressor and error compensation. The probability-optimized compressor is customized for partial product matrix (PPM) of signed operands, which gets the optimal logic circuit after probabilistic analysis and optimization. At the same time, we explored the PPM truncation method, found out the impact of different partial product (PP) truncation numbers on circuit benefit and error, and achieved a more ideal performance-error tradeoff through a reasonable error compensation method. In the optimal case of 8 bits, the proposed approximate multiplier saves 49.84% power, 46.41% area and 24.65% delay compared to the exact multiplier. We employed the proposed approximate multiplier in the vector systolic array as the processing element (PE). Under the VGG-16 evaluation, the proposed accelerator achieves performance improvement of energy efficiency $1.96\times $ , while the error loss was only 0.95%.
- 1 September 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 280-285
LTNN: An energy-efficient machine learning accelerator on 3D CMOS-RRAM for layer-wise tensorized neural network
An energy efficient machine learning requires an effective construction of neural network during training. This paper introduces a tensorized formulation of neural network during training such that weight matrix can be significantly compressed. The tensorized neural network can be further naturally mapped to a 3D CMOS-RRAM based accelerator with significant bandwidth boosting from vertical I/O connections. As such, high throughput and low power can be achieved simultaneously. Simulation results using the benchmark MNIST show that the proposed accelerator has 1.294x speed-up, 2.393× energy-efficiency and 7.59× area saving compared to 3D CMOS-ASIC implementation. Moreover, our proposed accelerator can achieve 370.64 GOPS throughput and 1055.95 GOPS/W energy efficiency, which is equivalent to 7.661 TOPS/W for uncompressed neural network. In addition, 142× model compression can be achieved by tensorization with acceptable accuracy loss.
- 1 September 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 280-285
LTNN: An energy-efficient machine learning accelerator on 3D CMOS-RRAM for layer-wise tensorized neural network
An energy efficient machine learning requires an effective construction of neural network during training. This paper introduces a tensorized formulation of neural network during training such that weight matrix can be significantly compressed. The tensorized neural network can be further naturally mapped to a 3D CMOS-RRAM based accelerator with significant bandwidth boosting from vertical I/O connections. As such, high throughput and low power can be achieved simultaneously. Simulation results using the benchmark MNIST show that the proposed accelerator has 1.294x speed-up, 2.393× energy-efficiency and 7.59× area saving compared to 3D CMOS-ASIC implementation. Moreover, our proposed accelerator can achieve 370.64 GOPS throughput and 1055.95 GOPS/W energy efficiency, which is equivalent to 7.661 TOPS/W for uncompressed neural network. In addition, 142× model compression can be achieved by tensorization with acceptable accuracy loss.
- Open Access
- 20 May 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
- Vol. 3, 37-46
An Energy-Efficient Digital ReRAM-Crossbar-Based CNN With Bitwise Parallelism
There is great attention to develop hardware accelerator with better energy efficiency, as well as throughput, than GPUs for convolutional neural network (CNN). The existing solutions have relatively limited parallelism as well as large power consumption (including leakage power). In this paper, we present a resistive random access memory (ReRAM)-accelerated CNN that can achieve significantly higher throughput and energy efficiency when the CNN is trained with binary constraints on both weights and activations, and is further mapped on a digital ReRAM-crossbar. We propose an optimized accelerator architecture tailored for bitwise convolution that features massive parallelism with high energy efficiency. Numerical experiment results show that the binary CNN accelerator on a digital ReRAM-crossbar achieves a peak throughput of 792 GOPS at the power consumption of 4.5 mW, which is 1.61 times faster and 296 times more energy-efficient than a high-end GPU.
- 20 May 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
- Vol. 3, 37-46
Open AccessAn Energy-Efficient Digital ReRAM-Crossbar-Based CNN With Bitwise Parallelism
There is great attention to develop hardware accelerator with better energy efficiency, as well as throughput, than GPUs for convolutional neural network (CNN). The existing solutions have relatively limited parallelism as well as large power consumption (including leakage power). In this paper, we present a resistive random access memory (ReRAM)-accelerated CNN that can achieve significantly higher throughput and energy efficiency when the CNN is trained with binary constraints on both weights and activations, and is further mapped on a digital ReRAM-crossbar. We propose an optimized accelerator architecture tailored for bitwise convolution that features massive parallelism with high energy efficiency. Numerical experiment results show that the binary CNN accelerator on a digital ReRAM-crossbar achieves a peak throughput of 792 GOPS at the power consumption of 4.5 mW, which is 1.61 times faster and 296 times more energy-efficient than a high-end GPU.
- 1 December 2016
- Published by Springer Nature in Synthesis Lectures on Emerging Engineering Technologies
- Vol. 2 (7) , 1-161
Non-Volatile In-Memory Computing by Spintronics
Exa-scale computing needs to re-examine the existing hardware platform that can support intensive data-oriented computing. Since the main bottleneck is from memory, we aim to develop an energy-efficient in-memory computing platform in this book. First, the models of spin-transfer torque magnetic tunnel junction and racetrack memory are presented. Next, we show that the spintronics could be a candidate for future data-oriented computing for storage, logic, and interconnect. As a result, by utilizing spintronics, in-memory-based computing has been applied for data encryption and machine learning. The implementations of in-memory AES, Simon cipher, as well as interconnect are explained in details. In addition, in-memory-based machine learning and face recognition are also illustrated in this book. Table of Contents: Preface / Acknowledgments / Introduction / Non-volatile Spintronic Device and Circuit / In-memory Data Encryption / In-memory Data Analytics / Authors' Biographies
- 1 December 2016
- Published by Springer Nature in Synthesis Lectures on Emerging Engineering Technologies
- Vol. 2 (7) , 1-161
Non-Volatile In-Memory Computing by Spintronics
Exa-scale computing needs to re-examine the existing hardware platform that can support intensive data-oriented computing. Since the main bottleneck is from memory, we aim to develop an energy-efficient in-memory computing platform in this book. First, the models of spin-transfer torque magnetic tunnel junction and racetrack memory are presented. Next, we show that the spintronics could be a candidate for future data-oriented computing for storage, logic, and interconnect. As a result, by utilizing spintronics, in-memory-based computing has been applied for data encryption and machine learning. The implementations of in-memory AES, Simon cipher, as well as interconnect are explained in details. In addition, in-memory-based machine learning and face recognition are also illustrated in this book. Table of Contents: Preface / Acknowledgments / Introduction / Non-volatile Spintronic Device and Circuit / In-memory Data Encryption / In-memory Data Analytics / Authors' Biographies
- 1 October 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 186-188
A 3D multi-layer CMOS-RRAM accelerator for multi-layer machine learning
Fast machine learning is required for future real-time data analytics. This paper introduces a 3D multi-layer CMOS-RRAM accelerator for learning on neural network. Given input of buffered data hold on the layer of a RRAM memory, intensive matrix-vector multiplication can be firstly accelerated on the layer of a digitized RRAM-crossbar. The remaining algorithmic operations such as feature extraction and classifier training can be accelerated on the layer of CMOS ASIC with consideration of parallelism and pipeline. Experiment results have shown that such a 3D accelerator can significantly reduce training time with acceptable accuracy. Compared to 3D-CMOS-ASIC implementation, it can achieve 1.28× smaller area, 2.05× faster runtime and 12.4× energy reduction.
- 1 October 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 186-188
A 3D multi-layer CMOS-RRAM accelerator for multi-layer machine learning
Fast machine learning is required for future real-time data analytics. This paper introduces a 3D multi-layer CMOS-RRAM accelerator for learning on neural network. Given input of buffered data hold on the layer of a RRAM memory, intensive matrix-vector multiplication can be firstly accelerated on the layer of a digitized RRAM-crossbar. The remaining algorithmic operations such as feature extraction and classifier training can be accelerated on the layer of CMOS ASIC with consideration of parallelism and pipeline. Experiment results have shown that such a 3D accelerator can significantly reduce training time with acceptable accuracy. Compared to 3D-CMOS-ASIC implementation, it can achieve 1.28× smaller area, 2.05× faster runtime and 12.4× energy reduction.
- 1 March 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-3
Future brain-like computing with non-volatile memory device
On-line machine learning has become the need for future data analytics. This work will show an RRAM-crossbar based l2-norm hardware solver for on-line machine learning that can significantly reduce training time when compared to the traditional gradient-based solution using backward propagation. Intensive matrix-vector multiplication in l2-norm solution can be mapped onto recent resistive switching random access memory (RRAM) devices. One incremental Cholesky factorization based l2-norm is proposed based on RRAM-crossbar implementation. Experimental results have shown that significant speedup can be achieved for matrix-vector multiplication in the l2-norm solver such that the overall training and testing time can be reduced respectively. In addition, large energy saving can be also achieved when compared to the traditional CMOS-based out-of-memory computing architecture.
- 1 March 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-3
Future brain-like computing with non-volatile memory device
On-line machine learning has become the need for future data analytics. This work will show an RRAM-crossbar based l2-norm hardware solver for on-line machine learning that can significantly reduce training time when compared to the traditional gradient-based solution using backward propagation. Intensive matrix-vector multiplication in l2-norm solution can be mapped onto recent resistive switching random access memory (RRAM) devices. One incremental Cholesky factorization based l2-norm is proposed based on RRAM-crossbar implementation. Experimental results have shown that significant speedup can be achieved for matrix-vector multiplication in the l2-norm solver such that the overall training and testing time can be reduced respectively. In addition, large energy saving can be also achieved when compared to the traditional CMOS-based out-of-memory computing architecture.