Sheng Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Sheng Liu has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Bo Yuan
College of Computer Science and Technology, National University of Defense Technology, Changsha, Hunan, China
6 shared publications - Zekun Jiang
College of Computer Science and Technology, National University of Defense Technology, Changsha, Hunan, China
5 shared publications - Jianfeng Cui
College of Computer Science and Technology, National University of Defense Technology, Changsha, Hunan, China
4 shared publications - Kai Lu
College of Computer Science and Technology, National University of Defense Technology, Changsha, China
4 shared publications - Jianfeng Cui
National University of Defense Technology, Changsha, Hunan, China
1 shared publication - Junsheng Chang
National University of Defense Technology
1 shared publication
Latest Publications
- Open Access
- 27 September 2026
- Published by Association for Computing Machinery (ACM)
- p. 975-985
RODIS: Accelerating Sparse Matrix Multiplication on GPU via Row-Orchestration and Dynamic Instruction Scheduling
- 27 September 2026
- Published by Association for Computing Machinery (ACM)
- p. 975-985
Open AccessRODIS: Accelerating Sparse Matrix Multiplication on GPU via Row-Orchestration and Dynamic Instruction Scheduling
- Open Access
- 24 September 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Design Automation of Electronic Systems
DCSR-GCN: A High-Performance GCN Accelerator Based on Dynamic Compression and Sparsity Reordering
Graph Convolutional Networks (GCNs) are widely used in tasks involving irregular graph data, such as recommendation. The hybrid execution pattern of sparse aggregation and dense combination during inference limits the efficiency of general processors like CPU and GPU. Therefore, designing dedicated accelerators for GCN inference has become an important research direction. However, existing GCN accelerators still suffer from high storage overhead caused by sparse indices and frequent read-after-write (RAW) conflicts during the aggregation phase, which limit system throughput and energy efficiency. To address these challenges, we adopt the hardware-software co-design approach and present DCSR-GCN, a high-performance GCN accelerator based on dynamic compression and sparsity reordering. First, we design an Enhanced Compressed Sparse Row (ECSR) format to reduce adjacency index storage space, and directly embed node degree fields to eliminate redundant computations in aggregation. Second, we propose a two-phase reordering algorithm that combines conflict-aware row scheduling with reuse-aware column grouping, thereby mitigating RAW conflicts and improving memory access efficiency. Finally, to fully exploit the benefits of sparse optimizations, we design a high-throughput vector systolic array that accelerates feature transformation in combination. Evaluations show that DCSR-GCN achieves up to 26.03 ×, 4.19 ×, and 3.52 × speedup compared to HyGCN, GCNAX, and MEGA, respectively.
- 24 September 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Design Automation of Electronic Systems
Open AccessDCSR-GCN: A High-Performance GCN Accelerator Based on Dynamic Compression and Sparsity Reordering
Graph Convolutional Networks (GCNs) are widely used in tasks involving irregular graph data, such as recommendation. The hybrid execution pattern of sparse aggregation and dense combination during inference limits the efficiency of general processors like CPU and GPU. Therefore, designing dedicated accelerators for GCN inference has become an important research direction. However, existing GCN accelerators still suffer from high storage overhead caused by sparse indices and frequent read-after-write (RAW) conflicts during the aggregation phase, which limit system throughput and energy efficiency. To address these challenges, we adopt the hardware-software co-design approach and present DCSR-GCN, a high-performance GCN accelerator based on dynamic compression and sparsity reordering. First, we design an Enhanced Compressed Sparse Row (ECSR) format to reduce adjacency index storage space, and directly embed node degree fields to eliminate redundant computations in aggregation. Second, we propose a two-phase reordering algorithm that combines conflict-aware row scheduling with reuse-aware column grouping, thereby mitigating RAW conflicts and improving memory access efficiency. Finally, to fully exploit the benefits of sparse optimizations, we design a high-throughput vector systolic array that accelerates feature transformation in combination. Evaluations show that DCSR-GCN achieves up to 26.03 ×, 4.19 ×, and 3.52 × speedup compared to HyGCN, GCNAX, and MEGA, respectively.
- Open Access
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
CODA: COoperative register Distribution and Arbitration among Partitioned GPU Sub-Cores
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessCODA: COoperative register Distribution and Arbitration among Partitioned GPU Sub-Cores
- Open Access
- 20 July 2026
- Published by Association for Computing Machinery (ACM) in ACM SIGMETRICS Performance Evaluation Review
- Vol. 54 (1) , 129-131
RELL-STC: An Efficient SpMM Extension of Sparse Tensor Core with Re-structured ELLPACK and Gustavson Dataflow
Sparsity is crucial for deep neural networks, reducing storage and bandwidth while improving computational efficiency. With the rapid development of deep neural networks, sparse computation accelerators have become a key research focus. NVIDIA introduced 2:4 fine-grained structured sparsity in the Ampere architecture, doubling Tensor Core peak throughput by retaining two out of four elements, but this may compromise model accuracy. In contrast, unstructured sparsity achieves maximum sparsity without accuracy loss. However, the irregular distribution of non-zero elements (NNZs) in unstructured sparse matrices causes complex memory access and load imbalance, preventing direct use of GPU Tensor Cores. Thus, computation relies on CUDA cores via sparse libraries like cuSPARSE, underutilizing Tensor Cores' capabilities. To address this, we extend the Tensor Core architecture based on Gustavson's dataflow and propose a re-structured row offline compression format for inference. This format integrates with the extended architecture to enable efficient unstructured sparse-dense matrix multiplication. Experiments show negligible area and power overheads, with an average speedup of 3.54x over cuSPARSE and 1.2x over state-of-the-art Tensor Core extensions.
- 20 July 2026
- Published by Association for Computing Machinery (ACM) in ACM SIGMETRICS Performance Evaluation Review
- Vol. 54 (1) , 129-131
Open AccessRELL-STC: An Efficient SpMM Extension of Sparse Tensor Core with Re-structured ELLPACK and Gustavson Dataflow
Sparsity is crucial for deep neural networks, reducing storage and bandwidth while improving computational efficiency. With the rapid development of deep neural networks, sparse computation accelerators have become a key research focus. NVIDIA introduced 2:4 fine-grained structured sparsity in the Ampere architecture, doubling Tensor Core peak throughput by retaining two out of four elements, but this may compromise model accuracy. In contrast, unstructured sparsity achieves maximum sparsity without accuracy loss. However, the irregular distribution of non-zero elements (NNZs) in unstructured sparse matrices causes complex memory access and load imbalance, preventing direct use of GPU Tensor Cores. Thus, computation relies on CUDA cores via sparse libraries like cuSPARSE, underutilizing Tensor Cores' capabilities. To address this, we extend the Tensor Core architecture based on Gustavson's dataflow and propose a re-structured row offline compression format for inference. This format integrates with the extended architecture to enable efficient unstructured sparse-dense matrix multiplication. Experiments show negligible area and power overheads, with an average speedup of 3.54x over cuSPARSE and 1.2x over state-of-the-art Tensor Core extensions.
- Open Access
- 8 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 129-131
RELL-STC: An Efficient SpMM Extension of Sparse Tensor Core with Re-structured ELLPACK and Gustavson Dataflow
- 8 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 129-131
Open AccessRELL-STC: An Efficient SpMM Extension of Sparse Tensor Core with Re-structured ELLPACK and Gustavson Dataflow
- Open Access
- 29 May 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Measurement and Analysis of Computing Systems
- Vol. 10 (2) , 1-22
RELL-STC: An Efficient SpMM Extension of Sparse Tensor Core with Re-structured ELLPACK and Gustavson Dataflow
Sparsity is a crucial optimization technique for deep neural networks, significantly reducing storage and bandwidth requirements while enhancing computational efficiency. In recent years, driven by the rapid advancement of deep neural networks, accelerators targeting sparse computation have emerged as a significant area of research. NVIDIA introduced 2:4 fine-grained structured sparsity starting with the Ampere architecture, which doubles the peak throughput of Tensor Cores by retaining only two out of every four elements. However, the constraints imposed by structured sparsity on parameters may compromise model accuracy. In contrast, unstructured sparsity enables maximum sparsity while maintaining model accuracy. Nevertheless, the irregular distribution of non-zero elements (NNZs) in unstructured sparse matrices introduces complex memory access patterns and load imbalances, preventing the direct utilization of GPU Tensor Cores. Consequently, computation typically relies on CUDA cores via sparse libraries such as cuSPARSE, leaving the powerful computational capabilities of Tensor Cores underutilized. To address this limitation, we extend the Tensor Core architecture inspired by Gustavson's dataflow and propose a re-structured row offline compression format tailored for the inference stage. This format integrates seamlessly with the extended architecture to achieve highly efficient unstructured sparse-dense matrix multiplication. Experimental results demonstrate that with negligible area and power overheads, our method achieves an average speedup of 3.54x compared to cuSPARSE and 1.2x compared to state-of-the-art Tensor Core extensions.
- 29 May 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Measurement and Analysis of Computing Systems
- Vol. 10 (2) , 1-22
Open AccessRELL-STC: An Efficient SpMM Extension of Sparse Tensor Core with Re-structured ELLPACK and Gustavson Dataflow
Sparsity is a crucial optimization technique for deep neural networks, significantly reducing storage and bandwidth requirements while enhancing computational efficiency. In recent years, driven by the rapid advancement of deep neural networks, accelerators targeting sparse computation have emerged as a significant area of research. NVIDIA introduced 2:4 fine-grained structured sparsity starting with the Ampere architecture, which doubles the peak throughput of Tensor Cores by retaining only two out of every four elements. However, the constraints imposed by structured sparsity on parameters may compromise model accuracy. In contrast, unstructured sparsity enables maximum sparsity while maintaining model accuracy. Nevertheless, the irregular distribution of non-zero elements (NNZs) in unstructured sparse matrices introduces complex memory access patterns and load imbalances, preventing the direct utilization of GPU Tensor Cores. Consequently, computation typically relies on CUDA cores via sparse libraries such as cuSPARSE, leaving the powerful computational capabilities of Tensor Cores underutilized. To address this limitation, we extend the Tensor Core architecture inspired by Gustavson's dataflow and propose a re-structured row offline compression format tailored for the inference stage. This format integrates seamlessly with the extended architecture to achieve highly efficient unstructured sparse-dense matrix multiplication. Experimental results demonstrate that with negligible area and power overheads, our method achieves an average speedup of 3.54x compared to cuSPARSE and 1.2x compared to state-of-the-art Tensor Core extensions.
- Open Access
- 2 February 2026
- Published by Elsevier in SSRN Electronic Journal
MMC: GNSS-Based High-Precision Clock Generation and Holdover System with MRAC Control and MLP Aging Prediction
Modern electronic systems impose strict requirements on clock accuracy and stability. To address this, this paper proposes a high-precision clock generation and retention method based on GNSS timing and all-digital control logic. Taking GNSS second pulse (1PPS) as a high-precision reference, a fully digital closed-loop control system based on heterogeneous computing platforms was constructed to achieve highly integrated hardware and flexible deployment of algorithms. In the control strategy, a Model Reference Adaptive Control (MRAC) algorithm was designed to estimate and compensate the nonlinear and time-varying characteristics of OCXO tuning sensitivity in real time, thereby enhancing the frequency control accuracy. In view of the loss of external reference time, a semi-parametric aging prediction model combining physical prior and lightweight Multilayer Perceptron (MLP) was used to actively predict the frequency drift and compensate the voltage, so as to prolong the high-precision retention time of the clock. Experimental results show that the proposed system can improve the long-term stability of the OCXO output clock to the order of 10−13, and maintain the clock accuracy in the sub-ppb (part per billion) accuracy range for several hours after GNSS loss, which effectively enhances the reliability and adaptability of the clock system under the abnormal reference signal. This study provides an integrated and intelligent solution for the design of high-precision clock sources, which is suitable for application scenarios with stringent time synchronization requirements in communication, navigation, finance and other fields.
- 2 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessMMC: GNSS-Based High-Precision Clock Generation and Holdover System with MRAC Control and MLP Aging Prediction
Modern electronic systems impose strict requirements on clock accuracy and stability. To address this, this paper proposes a high-precision clock generation and retention method based on GNSS timing and all-digital control logic. Taking GNSS second pulse (1PPS) as a high-precision reference, a fully digital closed-loop control system based on heterogeneous computing platforms was constructed to achieve highly integrated hardware and flexible deployment of algorithms. In the control strategy, a Model Reference Adaptive Control (MRAC) algorithm was designed to estimate and compensate the nonlinear and time-varying characteristics of OCXO tuning sensitivity in real time, thereby enhancing the frequency control accuracy. In view of the loss of external reference time, a semi-parametric aging prediction model combining physical prior and lightweight Multilayer Perceptron (MLP) was used to actively predict the frequency drift and compensate the voltage, so as to prolong the high-precision retention time of the clock. Experimental results show that the proposed system can improve the long-term stability of the OCXO output clock to the order of 10−13, and maintain the clock accuracy in the sub-ppb (part per billion) accuracy range for several hours after GNSS loss, which effectively enhances the reliability and adaptability of the clock system under the abnormal reference signal. This study provides an integrated and intelligent solution for the design of high-precision clock sources, which is suitable for application scenarios with stringent time synchronization requirements in communication, navigation, finance and other fields.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64, 5648514
Geo-Sparse Reconstruction Transformer for Remote Sensing Image Super-Resolution
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64, 5648514
Geo-Sparse Reconstruction Transformer for Remote Sensing Image Super-Resolution