Xiao Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xiao Zhang has a total of 64 co-authors. Here are the top co-authors by number of shared publications:
- Xiaoyong Du
Renmin University of China
5 shared publications - Feng Zhang
Renmin University of China
4 shared publications Shan Wang
Renmin University of China
4 shared publicationsNingnan Zhou
Renmin University of China
3 shared publications- Bingsheng He
National University of Singapore, Singapore, Singapore
2 shared publications - Jiawei Guan
Renmin University of China, Beijing, China
2 shared publications
Latest Publications
- Open Access
- 7 October 2026
- Published by Association for Computing Machinery (ACM)
- p. 4078-4090
Fuzz4DB: A Practice of LLM-Agent-Guided Fuzzing for Database Feature-Level Delta Testing
- 7 October 2026
- Published by Association for Computing Machinery (ACM)
- p. 4078-4090
Open AccessFuzz4DB: A Practice of LLM-Agent-Guided Fuzzing for Database Feature-Level Delta Testing
- Open Access
- 2 April 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 4  (1) , 1-26
AeroSketch: Near-Optimal Time Matrix Sketch Framework for Persistent, Sliding Window, and Distributed Streams
Many real-world matrix datasets arrive as high-throughput vector streams, making it impractical to store or process them in their entirety. To enable real-time analytics under limited computational, memory, and communication resources, matrix sketching techniques have been developed over recent decades to provide compact approximations of such streaming data. Some algorithms have achieved optimal space and communication complexity. However, these approaches often require frequent time-consuming matrix factorization operations. In particular, under tight approximation error bounds, each matrix factorization computation incurs cubic time complexity, thereby limiting their update efficiency. In this paper, we introduce AeroSketch, a novel matrix sketching framework that leverages recent advances in randomized numerical linear algebra (RandNLA). AeroSketch achieves optimal communication and space costs while delivering near-optimal update time complexity (within logarithmic factors) across persistent, sliding window, and distributed streaming scenarios. Extensive experiments on both synthetic and real-world datasets demonstrate that AeroSketch consistently outperforms state-of-the-art methods in update throughput. In particular, under tight approximation error constraints, AeroSketch reduces the cubic time complexity to the quadratic level. Meanwhile, it maintains comparable approximation quality while retaining optimal communication and space costs.
- 2 April 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 4  (1) , 1-26
Open AccessAeroSketch: Near-Optimal Time Matrix Sketch Framework for Persistent, Sliding Window, and Distributed Streams
Many real-world matrix datasets arrive as high-throughput vector streams, making it impractical to store or process them in their entirety. To enable real-time analytics under limited computational, memory, and communication resources, matrix sketching techniques have been developed over recent decades to provide compact approximations of such streaming data. Some algorithms have achieved optimal space and communication complexity. However, these approaches often require frequent time-consuming matrix factorization operations. In particular, under tight approximation error bounds, each matrix factorization computation incurs cubic time complexity, thereby limiting their update efficiency. In this paper, we introduce AeroSketch, a novel matrix sketching framework that leverages recent advances in randomized numerical linear algebra (RandNLA). AeroSketch achieves optimal communication and space costs while delivering near-optimal update time complexity (within logarithmic factors) across persistent, sliding window, and distributed streaming scenarios. Extensive experiments on both synthetic and real-world datasets demonstrate that AeroSketch consistently outperforms state-of-the-art methods in update throughput. In particular, under tight approximation error constraints, AeroSketch reduces the cubic time complexity to the quadratic level. Meanwhile, it maintains comparable approximation quality while retaining optimal communication and space costs.
- 31 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Software Engineering
- Vol. 51  (12) , 3654-3668
A Comprehensive Study of Bugs in Relational DBMS
Relational Database Management Systems (RDBMSs) are crucial infrastructures supporting a wide range of applications, making bug mitigation within these systems essential. This study presents the first comprehensive analysis of bugs in three popular open-source RDBMSs—MySQL, SQLite, and openGauss. We manually examined 777 bugs across four dimensions, i.e., bug root causes, bug symptoms, bug distribution across modules, and the correlations between the studied aspects. We also analyzed the bug-triggering SQL statements to uncover test cases that cannot be generated by existing tools. We have made 12 findings, which throw lights on the development, maintenance and testing of RDBMS systems. Particularly, our findings reveal that bugs related to SQL data types and complex features, such as database triggers, procedures and database parameter settings, present significant opportunities for enhancing RDBMS bug detection and mitigation. Leveraging these insights, we developed a tool, SQLT, which effectively identified eight RDBMS bugs (five type-related), all verified by developers, with four subsequently fixed.
- 31 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Software Engineering
- Vol. 51  (12) , 3654-3668
A Comprehensive Study of Bugs in Relational DBMS
Relational Database Management Systems (RDBMSs) are crucial infrastructures supporting a wide range of applications, making bug mitigation within these systems essential. This study presents the first comprehensive analysis of bugs in three popular open-source RDBMSs—MySQL, SQLite, and openGauss. We manually examined 777 bugs across four dimensions, i.e., bug root causes, bug symptoms, bug distribution across modules, and the correlations between the studied aspects. We also analyzed the bug-triggering SQL statements to uncover test cases that cannot be generated by existing tools. We have made 12 findings, which throw lights on the development, maintenance and testing of RDBMS systems. Particularly, our findings reveal that bugs related to SQL data types and complex features, such as database triggers, procedures and database parameter settings, present significant opportunities for enhancing RDBMS bug detection and mitigation. Leveraging these insights, we developed a tool, SQLT, which effectively identified eight RDBMS bugs (five type-related), all verified by developers, with four subsequently fixed.
- 28 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 75  (2) , 503-515
MI-LLM: Multiplier-Free LLM Inference on Commodity Processing-in-Memory Hardware
Large language models (LLMs) are prominent for their superior ability in language understanding and generation. However, a notorious problem for LLM inference is low computational utilization caused by the memory bottleneck, since it typically requires large memory capacity and high bandwidth to process neural weights. By integrating processing cores into memory, Processing-In-Memory (PIM) architecture excels at alleviating memory bottleneck; with the recent release of the first commodity near-bank PIM hardware (NBP), PIM becomes off-the-shelf and shows great potential for accelerating LLM inference practically. However, simply shoehorning LLM inference on NBP can not achieve satisfactory performance due to its inherent limitations: weak compute performance, frequent cache misses caused by the limited working memory capacity, and poor inter-PIM-core communication bandwidth. To address these limitations, we propose MI-LLM, an efficient system deploying LLM inference on NBP hardware. Its key idea is to build NBP-aware Lookup Tables (LUTs) and completely replace multiplications with lookups on LUTs, thereby mitigating the limitation of weak compute performance. 1) To reduce the model accuracy drop caused by the use of LUT, MI-LLM tailors a learning-based LUT construction method to maintain the model accuracy. 2) To cope with frequent cache misses caused by LUT sizes far exceeding PIM working memory capacity, MI-LLM introduces the design of PIM-aware linear kernel, with the optimization of intra-row and inter-row reordering enabled, to enhance LUT lookup locality. 3) MI-LLM further proposes a model partitioning scheme to minimize inter-PIM-core communication. Kernel-level benchmarks reveal that MI-LLM achieves a 9% throughput improvement and an 11% increase in energy efficiency over GPU implementations. Compared to FP8 quantization, MI-LLM incurs only a 0.24 times increase in perplexity, demonstrating minimal accuracy degradation. Moreover, in our end-to-end evaluation, MI-LLM requires 80% fewer ALU operation ticks per output token than the GPU baseline.
- 28 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 75  (2) , 503-515
MI-LLM: Multiplier-Free LLM Inference on Commodity Processing-in-Memory Hardware
Large language models (LLMs) are prominent for their superior ability in language understanding and generation. However, a notorious problem for LLM inference is low computational utilization caused by the memory bottleneck, since it typically requires large memory capacity and high bandwidth to process neural weights. By integrating processing cores into memory, Processing-In-Memory (PIM) architecture excels at alleviating memory bottleneck; with the recent release of the first commodity near-bank PIM hardware (NBP), PIM becomes off-the-shelf and shows great potential for accelerating LLM inference practically. However, simply shoehorning LLM inference on NBP can not achieve satisfactory performance due to its inherent limitations: weak compute performance, frequent cache misses caused by the limited working memory capacity, and poor inter-PIM-core communication bandwidth. To address these limitations, we propose MI-LLM, an efficient system deploying LLM inference on NBP hardware. Its key idea is to build NBP-aware Lookup Tables (LUTs) and completely replace multiplications with lookups on LUTs, thereby mitigating the limitation of weak compute performance. 1) To reduce the model accuracy drop caused by the use of LUT, MI-LLM tailors a learning-based LUT construction method to maintain the model accuracy. 2) To cope with frequent cache misses caused by LUT sizes far exceeding PIM working memory capacity, MI-LLM introduces the design of PIM-aware linear kernel, with the optimization of intra-row and inter-row reordering enabled, to enhance LUT lookup locality. 3) MI-LLM further proposes a model partitioning scheme to minimize inter-PIM-core communication. Kernel-level benchmarks reveal that MI-LLM achieves a 9% throughput improvement and an 11% increase in energy efficiency over GPU implementations. Compared to FP8 quantization, MI-LLM incurs only a 0.24 times increase in perplexity, demonstrating minimal accuracy degradation. Moreover, in our end-to-end evaluation, MI-LLM requires 80% fewer ALU operation ticks per output token than the GPU baseline.
- Top Cited Papers
- 9 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cloud Computing
- Vol. 13  (2) , 694-710
Breaking the Edge: Enabling Efficient Neural Network Inference on Integrated Edge Devices
Edge computing has gained widespread attention in cloud computing due to the increasing demands of AIoT applications and the evolution of edge architectures. One prevalent application in this domain is neural network inference on edge for computing and processing. This article presents an in-depth exploration of inference on integrated edge devices and introduces EdgeNN, a groundbreaking solution for inference specifically designed for CPU-GPU integrated edge devices. EdgeNN offers three key innovations. First, EdgeNN adaptively employs zero-copy optimization by harnessing unified physical memory. Second, EdgeNN introduces an innovative approach to CPU-GPU hybrid execution tailored for inference tasks. This technique enables concurrent CPU and GPU operation, effectively leveraging edge platforms’ computational capabilities. Third, EdgeNN adopts a finely tuned adaptive inference tuning technique that analyzes complex inference structures. It divides computations into sub-tasks, intelligently assigning them to the two processors for better performance. Experimental results demonstrate EdgeNN's superiority across six popular neural network inference processing. EdgeNN delivers average speed improvements of 3.97×, 4.10×, 3.12×, and 8.80× when compared to inference on four distinct edge CPUs. Furthermore, EdgeNN achieves significant time advantages compared to the direct execution of original programs. This improvement is attributed to better unified memory utilization (44.37%) and the innovative CPU-GPU hybrid execution approach (17.91%). Additionally, EdgeNN exhibits superior energy efficiency, providing 29.14× higher energy efficiency than edge CPUs and 5.70× higher energy efficiency than discrete GPUs. EdgeNN is now open source at https://github.com/ChenyangZhang-cs/EdgeNN.
- 9 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cloud Computing
- Vol. 13  (2) , 694-710
Top Cited PapersBreaking the Edge: Enabling Efficient Neural Network Inference on Integrated Edge Devices
Edge computing has gained widespread attention in cloud computing due to the increasing demands of AIoT applications and the evolution of edge architectures. One prevalent application in this domain is neural network inference on edge for computing and processing. This article presents an in-depth exploration of inference on integrated edge devices and introduces EdgeNN, a groundbreaking solution for inference specifically designed for CPU-GPU integrated edge devices. EdgeNN offers three key innovations. First, EdgeNN adaptively employs zero-copy optimization by harnessing unified physical memory. Second, EdgeNN introduces an innovative approach to CPU-GPU hybrid execution tailored for inference tasks. This technique enables concurrent CPU and GPU operation, effectively leveraging edge platforms’ computational capabilities. Third, EdgeNN adopts a finely tuned adaptive inference tuning technique that analyzes complex inference structures. It divides computations into sub-tasks, intelligently assigning them to the two processors for better performance. Experimental results demonstrate EdgeNN's superiority across six popular neural network inference processing. EdgeNN delivers average speed improvements of 3.97×, 4.10×, 3.12×, and 8.80× when compared to inference on four distinct edge CPUs. Furthermore, EdgeNN achieves significant time advantages compared to the direct execution of original programs. This improvement is attributed to better unified memory utilization (44.37%) and the innovative CPU-GPU hybrid execution approach (17.91%). Additionally, EdgeNN exhibits superior energy efficiency, providing 29.14× higher energy efficiency than edge CPUs and 5.70× higher energy efficiency than discrete GPUs. EdgeNN is now open source at https://github.com/ChenyangZhang-cs/EdgeNN.
- 11 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Parallel and Distributed Systems
- Vol. 34  (10) , 2699-2717
Enabling Efficient Random Access to Hierarchically Compressed Text Data on Diverse GPU Platforms
The tremendous computing capacity of GPU offers significant potential in processing hierarchically compressed text data without decompression. However, current GPU techniques offer only traversal-based text data analytics; random access is exceedingly inefficient, limiting their utility significantly. To address this issue, we develop a novel and widely applicable solution that prompts random access to hierarchically compressed text data without decompression in GPU memory. We address three main challenges for enabling efficient random access to compressed text data on GPUs. The first challenge is designing GPU data structures that facilitate random access. The second challenge is efficiently generating data structures on GPU. The CPU is inefficient when generating data structures for random access, and this inefficiency increases considerably when PCIe transmission is incorporated. The third challenge is query processing on compressed text data in GPU memory. Random accesses, such as data updates, cause massive conflicts among countless threads. In order to address the first challenge, we develop several compressed GPU data structures, including indexing within the intricate GPU memory hierarchy. To handle the second challenge, we propose a two-phase process for producing these data structures on GPU. For the third challenge, a double-parsing design is proposed as a solution to avoid conflicts. We evaluate our solution on three platforms, two server-grade GPU platforms and one edge-grade GPU platform, using five real-world datasets. Experimental results show that random access operations on GPU achieve an average speedup of $52.98\times$ compared to the state-of-the-art solution.
- 11 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Parallel and Distributed Systems
- Vol. 34  (10) , 2699-2717
Enabling Efficient Random Access to Hierarchically Compressed Text Data on Diverse GPU Platforms
The tremendous computing capacity of GPU offers significant potential in processing hierarchically compressed text data without decompression. However, current GPU techniques offer only traversal-based text data analytics; random access is exceedingly inefficient, limiting their utility significantly. To address this issue, we develop a novel and widely applicable solution that prompts random access to hierarchically compressed text data without decompression in GPU memory. We address three main challenges for enabling efficient random access to compressed text data on GPUs. The first challenge is designing GPU data structures that facilitate random access. The second challenge is efficiently generating data structures on GPU. The CPU is inefficient when generating data structures for random access, and this inefficiency increases considerably when PCIe transmission is incorporated. The third challenge is query processing on compressed text data in GPU memory. Random accesses, such as data updates, cause massive conflicts among countless threads. In order to address the first challenge, we develop several compressed GPU data structures, including indexing within the intricate GPU memory hierarchy. To handle the second challenge, we propose a two-phase process for producing these data structures on GPU. For the third challenge, a double-parsing design is proposed as a solution to avoid conflicts. We evaluate our solution on three platforms, two server-grade GPU platforms and one edge-grade GPU platform, using five real-world datasets. Experimental results show that random access operations on GPU achieve an average speedup of $52.98\times$ compared to the state-of-the-art solution.
- 21 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 35  (10) , 10181-10196
Expanding the Edge: Enabling Efficient Winograd CNN Inference With Deep Reuse on Edge Device
Deep learning on edge devices is becoming increasingly important, especially with the explosion of IoT devices. For example, the total number of devices connected to IoT reaches 29 billion in 2022. Convolutional neural networks (CNNs), as common deep learning representatives, are among the most popular neural networks in knowledge and data engineering. However, CNN employs a high degree of computing. In comparison to the training phase, the inference process is more frequently done on low-power computing equipments, such as edge devices. The limited computing resource and high computation pressure limit the effective use of CNN algorithms at the edge. Fortunately, a minimal filtering algorithm called Winograd can reduce convolution calculations by minimizing multiplication operations. We find that Winograd convolution can be accelerated further by deep reuse technique, which reuses the similar data and computation processes. In this paper, we propose a new inference method, called DREW, which combines deep reuse with Winograd for further accelerating CNNs. DREW handles three difficulties. First, it can detect the similarities from the complex minimal filtering patterns by clustering. Second, it reduces the online clustering cost in a reasonable range. Third, it provides an adjustable method in clustering granularity balancing the performance and accuracy. We perform evaluation on Raspberry PI and NVIDIA Jetson AGX Xavier edge devices, and experiments show that on five popular networks, 1) DREW further accelerates the Winograd convolution by an average of 8.27× speedup. Even for the highly parallel Winograd implementation, DREW still can provide 2.21× speedup. 2) When DREW is applied to end-to-end Winograd CNN inferences, DREW achieves 5.94× the average performance speedup with no ( $< $ 0.4%) accuracy loss. 3) Energy consumption is an important factor for inference in practice. DREW reduces the number of convolution operations to 10% of the original operations, thus achieving up to 60% energy-efficiency benefits than the original Winograd inference.
- 21 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 35  (10) , 10181-10196
Expanding the Edge: Enabling Efficient Winograd CNN Inference With Deep Reuse on Edge Device
Deep learning on edge devices is becoming increasingly important, especially with the explosion of IoT devices. For example, the total number of devices connected to IoT reaches 29 billion in 2022. Convolutional neural networks (CNNs), as common deep learning representatives, are among the most popular neural networks in knowledge and data engineering. However, CNN employs a high degree of computing. In comparison to the training phase, the inference process is more frequently done on low-power computing equipments, such as edge devices. The limited computing resource and high computation pressure limit the effective use of CNN algorithms at the edge. Fortunately, a minimal filtering algorithm called Winograd can reduce convolution calculations by minimizing multiplication operations. We find that Winograd convolution can be accelerated further by deep reuse technique, which reuses the similar data and computation processes. In this paper, we propose a new inference method, called DREW, which combines deep reuse with Winograd for further accelerating CNNs. DREW handles three difficulties. First, it can detect the similarities from the complex minimal filtering patterns by clustering. Second, it reduces the online clustering cost in a reasonable range. Third, it provides an adjustable method in clustering granularity balancing the performance and accuracy. We perform evaluation on Raspberry PI and NVIDIA Jetson AGX Xavier edge devices, and experiments show that on five popular networks, 1) DREW further accelerates the Winograd convolution by an average of 8.27× speedup. Even for the highly parallel Winograd implementation, DREW still can provide 2.21× speedup. 2) When DREW is applied to end-to-end Winograd CNN inferences, DREW achieves 5.94× the average performance speedup with no ( $< $ 0.4%) accuracy loss. 3) Energy consumption is an important factor for inference in practice. DREW reduces the number of convolution operations to 10% of the original operations, thus achieving up to 60% energy-efficiency benefits than the original Winograd inference.
- 23 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Parallel and Distributed Systems
- Vol. 32  (7) , 1740-1752
iMLBench: A Machine Learning Benchmark Suite for CPU-GPU Integrated Architectures
Utilizing heterogeneous accelerators, especially GPUs, to accelerate machine learning tasks has shown to be a great promise in recent years. GPUs bring huge performance improvements to machine learning and greatly promote the widespread adoptions of machine learning. However, the discrete CPU-GPU architecture design with high PCIe transmission overhead decreases the GPU computing benefits in machine learning training tasks. To overcome such limitations, hardware vendors release CPU-GPU integrated architectures with shared unified memory. In this paper, we design a benchmark suite for machine learning training on CPU-GPU integrated architectures, called iMLBench, covering a wide range of machine learning applications and kernels. We mainly explore two features on integrated architectures: 1) zero-copy, which means that the PCIe overhead has been eliminated for machine learning tasks, and 2) co-running, which means that CPU and GPU co-run together to process the same machine learning tasks. Our experimental results on iMLBench show that integrated architecture brings an average 14.18 performance improvement over the original implementations. Specifically, zero-copy design brings 4.65 performance improvement, and co-running brings 1.78 improvement. Moreover, integrated architectures exhibit promising results from both performance-per-dollar and energy perspectives, achieving 6.50 performance-price ratio while 4.06 energy efficiency over discrete GPUs.
- 23 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Parallel and Distributed Systems
- Vol. 32  (7) , 1740-1752
iMLBench: A Machine Learning Benchmark Suite for CPU-GPU Integrated Architectures
Utilizing heterogeneous accelerators, especially GPUs, to accelerate machine learning tasks has shown to be a great promise in recent years. GPUs bring huge performance improvements to machine learning and greatly promote the widespread adoptions of machine learning. However, the discrete CPU-GPU architecture design with high PCIe transmission overhead decreases the GPU computing benefits in machine learning training tasks. To overcome such limitations, hardware vendors release CPU-GPU integrated architectures with shared unified memory. In this paper, we design a benchmark suite for machine learning training on CPU-GPU integrated architectures, called iMLBench, covering a wide range of machine learning applications and kernels. We mainly explore two features on integrated architectures: 1) zero-copy, which means that the PCIe overhead has been eliminated for machine learning tasks, and 2) co-running, which means that CPU and GPU co-run together to process the same machine learning tasks. Our experimental results on iMLBench show that integrated architecture brings an average 14.18 performance improvement over the original implementations. Specifically, zero-copy design brings 4.65 performance improvement, and co-running brings 1.78 improvement. Moreover, integrated architectures exhibit promising results from both performance-per-dollar and energy perspectives, achieving 6.50 performance-price ratio while 4.06 energy efficiency over discrete GPUs.
- 13 November 2018
- Published by Springer Nature in Frontiers of Computer Science
- Vol. 13  (6) , 1282-1295
Timestamp reassignment: taming transaction abort for serializable snapshot isolation
- 13 November 2018
- Published by Springer Nature in Frontiers of Computer Science
- Vol. 13  (6) , 1282-1295
Timestamp reassignment: taming transaction abort for serializable snapshot isolation
- Open Access
- 15 November 2017
- Published by Springer Nature in Data Science and Engineering
- Vol. 2  (4) , 301-315
Reordering Transaction Execution to Boost High-Frequency Trading Applications
- 15 November 2017
- Published by Springer Nature in Data Science and Engineering
- Vol. 2  (4) , 301-315
Open AccessReordering Transaction Execution to Boost High-Frequency Trading Applications