Wenxing Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Wenxing Li has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
- Rongliang Fu
The Chinese University of Hong Kong, Hong Kong, Hong Kong, China
2 shared publications - Tsung-Yi Ho
The Chinese University of Hong Kong
2 shared publications - Xiaochun Ye
Institute of Computing Technology Chinese Academy of Sciences
2 shared publications - Da Wang
SKLP, Institute of Computing Technology, CAS, Beijing, Beijing, China
1 shared publication - Dashan Shang
Institute of Microelectronics, Chinese Academy of Sciences
1 shared publication - Fei Wang
School of Integrated Circuits, Shandong University, Jinan 250101, China
1 shared publication
Latest Publications
- Open Access
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
AHASD: Asynchronous Heterogeneous Architecture for LLM Adaptive Drafting Speculative Decoding on Mobile Devices
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessAHASD: Asynchronous Heterogeneous Architecture for LLM Adaptive Drafting Speculative Decoding on Mobile Devices
- Open Access
- 22 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 31-37
THLR: A Top-down Hierarchical Logic Rewrite Framework for Xor-Majority-Inverter Graphs
With the increasing complexity of integrated circuits, multiple Boolean network types have been developed to support efficient logic rewriting methods. Although the Xor-Majority-Inverter Graphs (XMG) have a relatively compact expressive power, due to the characteristics of the rewriting itself and the inherent properties of XMG, the rewriting does not always perform optimally in terms of optimization performance on XMG. In this paper, we propose a novel top-down hierarchical logic rewriting framework for XMG that exploits the complementary expressive capabilities of multiple Boolean network types. To more effectively leverage the rewriting potential of hierarchical Boolean network types, we propose a type-aware partitioning strategy that decomposes the network into structurally meaningful sub-circuits. This enables targeted optimizations tailored to the structural characteristics of each sub-circuit, effectively balancing rewriting quality with computational efficiency. Experimental results demonstrate that our framework significantly improves circuit quality, achieving an approximate 4.31% reduction in node-depth product (NDP) compared to state-of-the-art rewriting methods, while also reducing the runtime by about 13.62%. Moreover, after ASIC mapping, THLR delivers a 3.10% improvement in area-delay product (ADP) over the state-of-the-art approaches.
- 22 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 31-37
Open AccessTHLR: A Top-down Hierarchical Logic Rewrite Framework for Xor-Majority-Inverter Graphs
With the increasing complexity of integrated circuits, multiple Boolean network types have been developed to support efficient logic rewriting methods. Although the Xor-Majority-Inverter Graphs (XMG) have a relatively compact expressive power, due to the characteristics of the rewriting itself and the inherent properties of XMG, the rewriting does not always perform optimally in terms of optimization performance on XMG. In this paper, we propose a novel top-down hierarchical logic rewriting framework for XMG that exploits the complementary expressive capabilities of multiple Boolean network types. To more effectively leverage the rewriting potential of hierarchical Boolean network types, we propose a type-aware partitioning strategy that decomposes the network into structurally meaningful sub-circuits. This enables targeted optimizations tailored to the structural characteristics of each sub-circuit, effectively balancing rewriting quality with computational efficiency. Experimental results demonstrate that our framework significantly improves circuit quality, achieving an approximate 4.31% reduction in node-depth product (NDP) compared to state-of-the-art rewriting methods, while also reducing the runtime by about 13.62%. Moreover, after ASIC mapping, THLR delivers a 3.10% improvement in area-delay product (ADP) over the state-of-the-art approaches.
- 20 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 75 (8) , 2870-2884
AgeBalance: Low-Cost Lifetime Extension for SRAM-Based PIM Accelerators
Although processing-in-memory (PIM) techniques have widely been used for deep neural networks (DNNs) acceleration, the inference performance of aged PIM-based accelerators remains to be investigated. This paper makes the first attempt to study Hot Carrier Injection (HCI) and Negative Bias Temperature Instability (NBTI) aging impacts on SRAM-based DNN accelerators, which provides a novel and unified framework, termed AgeBalance for aging detection, analysis and mitigation. First, we discuss a convenient aging detection scheme. Then, we benchmark the inference accuracy drops of DNNs running on aged SRAM-based PIM accelerators. Finally, we propose a low-cost anti-aging training method without incurring additional hardware overhead on SRAM-based DNN accelerators. Extensive experimental results on MNIST, CIFAR10 and AG News datasets show that aging can cause the inference accuracy of shallow or deep DNNs to drop to about 10%, close to random guessing. The aging mitigation scheme proposed in this paper can largely restore the accuracy to the original. Moreover, the SRAM write overhead of our method is much reduced thanks to a score-based training approach, leading to a reduction of 5× to 10× writing energy compared to the traditional training method.
- 20 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 75 (8) , 2870-2884
AgeBalance: Low-Cost Lifetime Extension for SRAM-Based PIM Accelerators
Although processing-in-memory (PIM) techniques have widely been used for deep neural networks (DNNs) acceleration, the inference performance of aged PIM-based accelerators remains to be investigated. This paper makes the first attempt to study Hot Carrier Injection (HCI) and Negative Bias Temperature Instability (NBTI) aging impacts on SRAM-based DNN accelerators, which provides a novel and unified framework, termed AgeBalance for aging detection, analysis and mitigation. First, we discuss a convenient aging detection scheme. Then, we benchmark the inference accuracy drops of DNNs running on aged SRAM-based PIM accelerators. Finally, we propose a low-cost anti-aging training method without incurring additional hardware overhead on SRAM-based DNN accelerators. Extensive experimental results on MNIST, CIFAR10 and AG News datasets show that aging can cause the inference accuracy of shallow or deep DNNs to drop to about 10%, close to random guessing. The aging mitigation scheme proposed in this paper can largely restore the accuracy to the original. Moreover, the SRAM write overhead of our method is much reduced thanks to a score-based training approach, leading to a reduction of 5× to 10× writing energy compared to the traditional training method.
- Open Access
- 23 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-6
SmartATPG: Learning-based Automatic Test Pattern Generation with Graph Convolutional Network and Reinforcement Learning
Automatic test pattern generation (ATPG) is a critical technology in integrated circuit testing. It searches for effective test vectors to detect all possible faults in the circuit as entirely as possible, thereby ensuring chip yield and improving chip quality. However, the process of searching for test vectors is NP-complete. At the same time, the large amount of backtracking generated during the search for test vectors can directly affect the performance of ATPG. In this paper, a learning-based ATPG framework, SmartATPG, is proposed to search for high-quality test vectors, reduce the number of backtracking during the search process, and thereby improve the performance of ATPG. SmartATPG utilizes graph convolutional network (GCN) to fully extract circuit feature information and efficiently explore the ATPG search space through reinforcement learning (RL). Experimental results indicate that the proposed SmartATPG outperforms traditional and artificial neural network (ANN)-based heuristic strategies on most benchmark circuits.
- 23 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Open AccessSmartATPG: Learning-based Automatic Test Pattern Generation with Graph Convolutional Network and Reinforcement Learning
Automatic test pattern generation (ATPG) is a critical technology in integrated circuit testing. It searches for effective test vectors to detect all possible faults in the circuit as entirely as possible, thereby ensuring chip yield and improving chip quality. However, the process of searching for test vectors is NP-complete. At the same time, the large amount of backtracking generated during the search for test vectors can directly affect the performance of ATPG. In this paper, a learning-based ATPG framework, SmartATPG, is proposed to search for high-quality test vectors, reduce the number of backtracking during the search process, and thereby improve the performance of ATPG. SmartATPG utilizes graph convolutional network (GCN) to fully extract circuit feature information and efficiently explore the ATPG search space through reinforcement learning (RL). Experimental results indicate that the proposed SmartATPG outperforms traditional and artificial neural network (ANN)-based heuristic strategies on most benchmark circuits.