Computer architecture &
machine-learning systems.

I am a postdoc at the School of Computer Science and Engineering, Beihang University, where I received my Ph.D. and M.S. in Computer Science under the supervision of Prof. Hailong Yang. I also received my B.S. in Computer Science from Beihang University.

My research lies at the intersection of computer architecture and machine-learning systems. I work on high-performance computing, performance analysis and tuning tools, deep-learning systems, and training & inference systems for large language models — in particular, how the next generation of AI workloads should reshape the hardware and software stacks beneath them. I currently serve as a core member on several national-level projects, including the National Key R&D Program of China and a Key Project of the National Natural Science Foundation of China (NSFC).

Some of my recent work includes profiling tools (GVARP, SimTrace) for diagnosing performance bottlenecks, variance, and trace analysis, and tools for GNN performance profiling and spatial-temporal analysis of parallel programs (GNNPerf, STAD).

Identifying Performance Inefficiencies of Parallel Program With Spatial and Temporal Trace Analysis

Zhibo Xuan, Xin Sun, Xin You, Hailong Yang, Zhongzhi Luan, Yi Liu, Depei Qian

IEEE Transactions on Parallel and Distributed Systems, 2025CCF-ADOI

· Combines spatial and temporal trace analysis to pinpoint performance inefficiencies in large-scale parallel programs.

SimTrace: Exploiting Spatial and Temporal Sampling for Large-Scale Performance Analysis

Zhibo Xuan, Xin You, Tianyu Feng, Hailong Yang, Zhongzhi Luan, Yi Liu, Depei Qian

ACM Transactions on Architecture and Code Optimization, 2025CCF-ADOI

· A sampling-based tracing framework that drastically reduces overhead while preserving diagnostic fidelity for HPC workloads.

GNNPerf: Towards Effective Performance Profiling and Analysis Across GNN Frameworks

Kejie Ma, Hailong Yang, Zizheng Zhang, Xin You, Zhibo Xuan, Qingxiao Sun, Zhongzhi Luan, Yi Liu, Depei Qian

IPDPS '25, 2025CCF-BDOI

· A cross-framework profiling tool that identifies and compares performance bottlenecks in GNN training pipelines.

Exploiting Synchronization-aware Transformer for Aligning Large-Scale MPI Traces

Zhibo Xuan, Xin You, Hailong Yang, Haoran Kong, Jingqi Chen, Tianyu Feng, Zhongzhi Luan, Yi Liu, Depei Qian

Frontiers of Computer Science, 2025CCF-B

· Leverages synchronization-aware transformers to efficiently align large-scale MPI traces for performance analysis.

GVARP: Detecting Performance Variance on Large-Scale Heterogeneous Systems

Xin You^*, Zhibo Xuan^*, Hailong Yang, Zhongzhi Luan, Yi Liu, Depei Qian* Equal contribution

SC '24, 2024CCF-ADOI

· Lightweight profiler that pinpoints the root causes of run-to-run performance variance on large-scale heterogeneous GPU systems.

Exploiting Transformer-Based Static Binary Analysis for Identifying Inefficient Locks

Zhibo Xuan, Xin You, Hailong Yang, Jingqi Chen, Zhongzhi Luan, Yi Liu, Depei Qian

NPC '25, 2025CCF-C

· A transformer-based static binary analysis tool that detects inefficient lock usage and contention in multithreaded programs.

Hotspy: Identifying Performance Hotspot with Graph Neural Network Based Static Analysis

Da Huo, Xin You, Zhibo Xuan, Hailong Yang, Zhongzhi Luan, Depei Qian

CCF Transactions on High Performance Computing, 2025

· Uses graph neural networks for static analysis to identify performance hotspots without runtime profiling overhead.

Retrospection on the Performance Analysis Tools for Large-Scale HPC Programs

Zhibo Xuan, Xin You, Hailong Yang, Mingzhen Li, Zhongzhi Luan, Yi Liu, Depei Qian

HiPC '24, 2024CCF-C

· A systematic survey and retrospective analysis of performance analysis tools for large-scale HPC programs.

gGMED: Towards GPU Accelerated Geometric Modeling Evaluation and Derivative Processes

Zhibo Xuan, Hailong Yang, Peng Wang, Xin Sun, Jie Hao, Siyuan Duan, Yiming Shi, Zhongzhi Luan, Depei Qian

Lecture Notes in Computer Science (APPT), 2024CCF-C

· Accelerates geometric modeling evaluation and derivative computation on GPUs for CAD/CAE applications.

Accelerating Big Data Application by Eliminating Redundancy on Hadoop Cluster

Ke Lei, Shitong Du, Xin You, Zhibo Xuan, Haoyu Kong, Hailong Yang, Jingya Shang, Zhongzhi Xiao, Zheng Wu, Zhongzhi Luan, Depei Qian

ICPADS '23, 2023CCF-C

· Identifies and eliminates redundant computation in Hadoop-based big data pipelines to improve cluster throughput.

PowerSpector: Towards Energy Efficiency with Calling-Context-Aware Profiling

Xin You, Hailong Yang, Zhibo Xuan, Zhongzhi Luan, Depei Qian

IPDPS '22, 2022CCF-B

· A calling-context-aware energy profiler that attributes power consumption to code regions for guided optimization.