Feng Shi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Feng Shi has a total of 114 co-authors. Here are the top co-authors by number of shared publications:
- Karim Soliman
School of Computer Science, Beijing Institute of Technology, Beijing, China
5 shared publications - Yizhuo Wang
School of Computer Science and Technology, Beijing Institute of Technology, Beijing, China
4 shared publications Chunfeng Li
3 shared publications- Fei Yin
School of Computer Science and Technology, Beijing Institute of Technology, Beijing, China
3 shared publications - Weixing Ji
Beijing Normal University
3 shared publications - Xiaojun Wang
Beijing Institute of Technology
3 shared publications
Latest Publications
- 4 August 2025
- Published by Elsevier in Integration
- Vol. 105
Semi-adaptive distributed approach for triplet-based architecture inter-core communication Network-on-Chip
- 4 August 2025
- Published by Elsevier in Integration
- Vol. 105
Semi-adaptive distributed approach for triplet-based architecture inter-core communication Network-on-Chip
- 12 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 45 (1) , 190-203
Reactive Deadlock Avoidance Based on Focus Routing Graph Classification for Triplet-Based Architecture Network-on-Chip
The implementation of Network-on-Chip (NoC) architectures presents considerable advantages in performance relative to traditional bus-based systems. However, the sophisticated nature of NoC designs demands careful oversight of shared resources to mitigate potential performance issues. In this context, well-structured routing algorithms are essential, as they facilitate improved traffic management and minimize congestion. Furthermore, mechanisms for deadlock prevention and avoidance are integral to routing algorithms, ensuring continuous packet transmission and ultimately enhancing network performance. This paper introduces a novel reactive deadlock avoidance method for Triplet-Based Architecture Inter-Core NoC (TriBA-cNoC). that classifies routing based on Focus Routing Graph (FRG) size to address routing-level deadlocks caused by the combination of deterministic routing and TriBA-cNoC’s inherent network characteristics. Compared to proactive techniques, it improves downstream buffer utilization and reduces power consumption. Furthermore, two shortest-path routing algorithms are introduced: DM4T-M, which incorporates a round-robin selection mechanism to alleviate congestion on critical paths and minimize hot-node formation. TSR, a novel two-stage distributed routing algorithm, addresses the computational overhead associated with output port selection in previous algorithms. Simulation results obtained using gem5 show that the proposed approach, which integrates routing-level reactive deadlock avoidance with the proposed routing algorithms, yields improvements in latency, throughput, buffer utilization, and power consumption. TSR and DM4T-M achieve latency reductions of up to 34.89% and 28.2%, respectively. Throughput increases of up to 16.94% and 7.81% are observed for TSR and DM4T-M, respectively. Moreover, the proposed approaches enhance buffer utilization by up to 13.9% and 10.44%, while reducing power consumption by up to 9.64%.
- 12 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 45 (1) , 190-203
Reactive Deadlock Avoidance Based on Focus Routing Graph Classification for Triplet-Based Architecture Network-on-Chip
The implementation of Network-on-Chip (NoC) architectures presents considerable advantages in performance relative to traditional bus-based systems. However, the sophisticated nature of NoC designs demands careful oversight of shared resources to mitigate potential performance issues. In this context, well-structured routing algorithms are essential, as they facilitate improved traffic management and minimize congestion. Furthermore, mechanisms for deadlock prevention and avoidance are integral to routing algorithms, ensuring continuous packet transmission and ultimately enhancing network performance. This paper introduces a novel reactive deadlock avoidance method for Triplet-Based Architecture Inter-Core NoC (TriBA-cNoC). that classifies routing based on Focus Routing Graph (FRG) size to address routing-level deadlocks caused by the combination of deterministic routing and TriBA-cNoC’s inherent network characteristics. Compared to proactive techniques, it improves downstream buffer utilization and reduces power consumption. Furthermore, two shortest-path routing algorithms are introduced: DM4T-M, which incorporates a round-robin selection mechanism to alleviate congestion on critical paths and minimize hot-node formation. TSR, a novel two-stage distributed routing algorithm, addresses the computational overhead associated with output port selection in previous algorithms. Simulation results obtained using gem5 show that the proposed approach, which integrates routing-level reactive deadlock avoidance with the proposed routing algorithms, yields improvements in latency, throughput, buffer utilization, and power consumption. TSR and DM4T-M achieve latency reductions of up to 34.89% and 28.2%, respectively. Throughput increases of up to 16.94% and 7.81% are observed for TSR and DM4T-M, respectively. Moreover, the proposed approaches enhance buffer utilization by up to 13.9% and 10.44%, while reducing power consumption by up to 9.64%.
- Open Access
- 20 March 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 22 (1) , 1-26
A High Scalability Memory NoC with Shared-Inside Hierarchical-Groupings for Triplet-Based Many-Core Architecture
Innovative processor architecture designs are shifting towards Many-Core Architectures (MCAs) to meet the future demands of high-performance computing as the limits of Moore’s Law have almost been reached. Many-core processors utilize shared memory hierarchies to achieve high-speed memory systems, improving memory access efficiency. However, as the number of cores multiplies, the scalability of this system is significantly constrained by the increased proportion of long-distance and Non-Uniform Memory Access (NUMA). Improving the scalability of MCAs is crucial for achieving large/super-scale general-purpose many-core processors. This work proposes a high-scalability memory Network-on-Chip (NoC) for Triplet-Based Many-Core Architecture (TriBA), named TriBA-mNoC. TriBA-mNoC maintains a consistent core-to-core spacing as the network scale increases, effectively preventing increased long-distance memory access latency. Moreover, it leverages an inherent advantage of shared-inside hierarchical-groupings, alleviating common NUMA issues in the NoC design. Evaluations of static network characteristics show that TriBA-mNoC outperforms most classical NoCs in network diameter, average distance, and cost. TriBA-mNoC can be integrated with TriBA in the same silicon die with a tile-like floorplan, forming a novel NoC called TriBA-NoC, which can combine the strengths of both networks to maximize the architecture performance. We evaluated the memory access performance and scalability of TriBA-NoC using the mathematical evaluation models and actual simulations with real traffic (PARSEC 3.0 and SPLASH-2) at different network scales. The mathematical evaluation results indicate that TriBA-NoC achieves an aggregate speedup of approximately 3x compared with 2D-Mesh for a similar number of cores. Furthermore, TriBA-NoC’s single-core speedup efficiency remains stable as the number of cores increases under the same cache hit ratio, whereas 2D-Mesh experiences a rapid decline, highlighting TriBA-NoC’s exceptional scalability. Finally, the actual traffic simulation results show that TriBA-NoC achieves an average memory access latency and time reduction of 25.90% to 40.50% and 5.61% to 31.69%, respectively, compared with 2D-Mesh.
- 20 March 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 22 (1) , 1-26
Open AccessA High Scalability Memory NoC with Shared-Inside Hierarchical-Groupings for Triplet-Based Many-Core Architecture
Innovative processor architecture designs are shifting towards Many-Core Architectures (MCAs) to meet the future demands of high-performance computing as the limits of Moore’s Law have almost been reached. Many-core processors utilize shared memory hierarchies to achieve high-speed memory systems, improving memory access efficiency. However, as the number of cores multiplies, the scalability of this system is significantly constrained by the increased proportion of long-distance and Non-Uniform Memory Access (NUMA). Improving the scalability of MCAs is crucial for achieving large/super-scale general-purpose many-core processors. This work proposes a high-scalability memory Network-on-Chip (NoC) for Triplet-Based Many-Core Architecture (TriBA), named TriBA-mNoC. TriBA-mNoC maintains a consistent core-to-core spacing as the network scale increases, effectively preventing increased long-distance memory access latency. Moreover, it leverages an inherent advantage of shared-inside hierarchical-groupings, alleviating common NUMA issues in the NoC design. Evaluations of static network characteristics show that TriBA-mNoC outperforms most classical NoCs in network diameter, average distance, and cost. TriBA-mNoC can be integrated with TriBA in the same silicon die with a tile-like floorplan, forming a novel NoC called TriBA-NoC, which can combine the strengths of both networks to maximize the architecture performance. We evaluated the memory access performance and scalability of TriBA-NoC using the mathematical evaluation models and actual simulations with real traffic (PARSEC 3.0 and SPLASH-2) at different network scales. The mathematical evaluation results indicate that TriBA-NoC achieves an aggregate speedup of approximately 3x compared with 2D-Mesh for a similar number of cores. Furthermore, TriBA-NoC’s single-core speedup efficiency remains stable as the number of cores increases under the same cache hit ratio, whereas 2D-Mesh experiences a rapid decline, highlighting TriBA-NoC’s exceptional scalability. Finally, the actual traffic simulation results show that TriBA-NoC achieves an average memory access latency and time reduction of 25.90% to 40.50% and 5.61% to 31.69%, respectively, compared with 2D-Mesh.
- 31 August 2024
- Published by Elsevier in Microprocessors and Microsystems
- Vol. 110
Proactive deadlock prevention based on traffic classification sub-graphs for triplet-based NoC TriBA-cNoC
- 31 August 2024
- Published by Elsevier in Microprocessors and Microsystems
- Vol. 110
Proactive deadlock prevention based on traffic classification sub-graphs for triplet-based NoC TriBA-cNoC
- 24 July 2024
- Published by Elsevier in Parallel Computing
- Vol. 121
NxtSPR: A deadlock-free shortest path routing dedicated to relaying for Triplet-Based many-core Architecture
- 24 July 2024
- Published by Elsevier in Parallel Computing
- Vol. 121
NxtSPR: A deadlock-free shortest path routing dedicated to relaying for Triplet-Based many-core Architecture
- Open Access
- 31 October 2023
- Published by Springer Nature in Research Square
A lightweight parallel execution engine for IoT data analytics in fog computing
With the rise of fog computing, the architecture of IOT is evolving into a cloud-fog-end architecture, where fog computing and cloud computing complement each other. The Internet of Things produces a large amount of data with great value. More and more real-time data analytics scenarios need to be carried out in the fog environment. Data processing and analytics has formed a relatively mature technology stack in the cloud, but in the fog environment with limited resources, the supporting framework of data analytics is still in its infancy. This paper reviews relevant research work in this field. Furthermore, this paper proposes a new parallel programming model and its API, and constructs a lightweight parallel execution engine (LwPEE) suitable for fog computing environment. Finally, we evaluate the performance of LwPEE and conduct relevant analysis.
- 31 October 2023
- Published by Springer Nature in Research Square
Open AccessA lightweight parallel execution engine for IoT data analytics in fog computing
With the rise of fog computing, the architecture of IOT is evolving into a cloud-fog-end architecture, where fog computing and cloud computing complement each other. The Internet of Things produces a large amount of data with great value. More and more real-time data analytics scenarios need to be carried out in the fog environment. Data processing and analytics has formed a relatively mature technology stack in the cloud, but in the fog environment with limited resources, the supporting framework of data analytics is still in its infancy. This paper reviews relevant research work in this field. Furthermore, this paper proposes a new parallel programming model and its API, and constructs a lightweight parallel execution engine (LwPEE) suitable for fog computing environment. Finally, we evaluate the performance of LwPEE and conduct relevant analysis.
- Open Access
- 4 May 2023
- Published by Wiley in Angewandte Chemie
- Vol. 135 (21)
Graphisches Inhaltsverzeichnis: Angew. Chem. 21/2023
- 4 May 2023
- Published by Wiley in Angewandte Chemie
- Vol. 135 (21)
Open AccessGraphisches Inhaltsverzeichnis: Angew. Chem. 21/2023
- Open Access
- 9 August 2022
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Sensor Networks
Retracted March 9, 2026: Heterogeneous Big Data Parallel Computing Optimization Model using MPI/OpenMP Hybrid and Sensor Networks
For the heterogeneous big data parallel computing model, two levels of parallelism between nodes are not considered, resulting in low efficiency of heterogeneous big data parallel computing and bandwidth to send and receive information, high communication overhead, long model running time and small computational volume. In the paper, we propose an optimization model of heterogeneous big data parallel computing based on a hybrid Multi Point Interface (MPI)/Open Multi-Processing (OpenMP) and Sensor Networks. First, the processor characteristics of heterogeneous big data architecture is analyzed, the parallel tasks among processors are divided, collect the heterogeneous big data to be computed and cluster them, and use the processing results as the input items of the model. Then, a parallel load balancing mechanism is established to optimally divide the parallel computing load of heterogeneous big data, and a parallel computing optimization program is written by combining the hybrid programming mode of MPI and OpenMP and using the hybrid MPI/OpenMP, and finally, the parallel computing optimization of heterogeneous big data is realized by optimizing the parallel communication and determining the model parameters. The results show that the proposed model has a communication bandwidth of 510Mbps, a computational volume of 1.16GB, a model runtime of 24s, and an improved network bandwidth utilization of 93%, which can effectively reduce the communication overhead, and improve the efficiency of parallel computing and bandwidth sending and receiving information in sensor networks, and shorten the model running time.
- 9 August 2022
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Sensor Networks
Open AccessRetracted March 9, 2026: Heterogeneous Big Data Parallel Computing Optimization Model using MPI/OpenMP Hybrid and Sensor Networks
For the heterogeneous big data parallel computing model, two levels of parallelism between nodes are not considered, resulting in low efficiency of heterogeneous big data parallel computing and bandwidth to send and receive information, high communication overhead, long model running time and small computational volume. In the paper, we propose an optimization model of heterogeneous big data parallel computing based on a hybrid Multi Point Interface (MPI)/Open Multi-Processing (OpenMP) and Sensor Networks. First, the processor characteristics of heterogeneous big data architecture is analyzed, the parallel tasks among processors are divided, collect the heterogeneous big data to be computed and cluster them, and use the processing results as the input items of the model. Then, a parallel load balancing mechanism is established to optimally divide the parallel computing load of heterogeneous big data, and a parallel computing optimization program is written by combining the hybrid programming mode of MPI and OpenMP and using the hybrid MPI/OpenMP, and finally, the parallel computing optimization of heterogeneous big data is realized by optimizing the parallel communication and determining the model parameters. The results show that the proposed model has a communication bandwidth of 510Mbps, a computational volume of 1.16GB, a model runtime of 24s, and an improved network bandwidth utilization of 93%, which can effectively reduce the communication overhead, and improve the efficiency of parallel computing and bandwidth sending and receiving information in sensor networks, and shorten the model running time.
- 26 April 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Parallel and Distributed Systems
- Vol. 33 (12) , 3732-3745
TaiChi: A Hybrid Compression Format for Binary Sparse Matrix-Vector Multiplication on GPU
Binary Sparse Matrix-Vector Multiplication (SpMV) is a heavy computational kernel in weblink analysis, integer factorization, compressed sensing, spectral graph theory, and other domains. Testing several popular GPU-based SpMV implementations on 400 sparse matrices, we observed that data transfer to GPU memory accounts for a large part of the total computation time. The transfer of constant value 1s can be easily eliminated for binary sparse matrices. However, compressing index arrays has always been a great challenge. This paper proposes a new compression format TaiChi to further reduce index data copies and improve the performance of SpMV, especially for diagonally dominant binary sparse matrices. Input matrices are firstly partitioned into relatively dense and ultra-sparse areas. Then the dense areas are encoded inversely by marking ``0"s, while the ultra-sparse area is encoded by marking ``1"s. We also designed a new SpMV algorithm only using addition and subtraction for binary matrices based on our partition and encoding format. Evaluation results on real-world binary sparse matrices show that our hybrid encoding for binary matrix significantly reduces the data transfer and speeds up the kernel execution. It achieves the highest transfer and kernel execution speedups of 5.63x and 3.84x on GTX 1080 Ti, 3.39x and 3.91x on Tesla V100.
- 26 April 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Parallel and Distributed Systems
- Vol. 33 (12) , 3732-3745
TaiChi: A Hybrid Compression Format for Binary Sparse Matrix-Vector Multiplication on GPU
Binary Sparse Matrix-Vector Multiplication (SpMV) is a heavy computational kernel in weblink analysis, integer factorization, compressed sensing, spectral graph theory, and other domains. Testing several popular GPU-based SpMV implementations on 400 sparse matrices, we observed that data transfer to GPU memory accounts for a large part of the total computation time. The transfer of constant value 1s can be easily eliminated for binary sparse matrices. However, compressing index arrays has always been a great challenge. This paper proposes a new compression format TaiChi to further reduce index data copies and improve the performance of SpMV, especially for diagonally dominant binary sparse matrices. Input matrices are firstly partitioned into relatively dense and ultra-sparse areas. Then the dense areas are encoded inversely by marking ``0"s, while the ultra-sparse area is encoded by marking ``1"s. We also designed a new SpMV algorithm only using addition and subtraction for binary matrices based on our partition and encoding format. Evaluation results on real-world binary sparse matrices show that our hybrid encoding for binary matrix significantly reduces the data transfer and speeds up the kernel execution. It achieves the highest transfer and kernel execution speedups of 5.63x and 3.84x on GTX 1080 Ti, 3.39x and 3.91x on Tesla V100.
- 27 August 2021
- Published by Springer Nature in Cluster Computing
- Vol. 25 (4) , 2601-2611
Cluster optimization algorithm based on CPU and GPU hybrid architecture
- 27 August 2021
- Published by Springer Nature in Cluster Computing
- Vol. 25 (4) , 2601-2611
Cluster optimization algorithm based on CPU and GPU hybrid architecture