个人简介
我的研究聚焦于高效人工智能架构与系统,尤其关注算法、系统与硬件之间的协同设计。
- 2026.01 – 至今华为香港研究所 2012 实验室,高级研究员
- 2023.02 – 2025.12香港中文大学,博士后
- 2022.07 – 2023.01读博的两年半太累了,回家钓鱼种菜睡觉
- 2020.01 – 2022.06香港中文大学(深圳),计算机与信息工程博士(获校长杰出博士研究奖)
- 2018.11 – 2020.01无业,打游戏,思考人生
- 2015.08 – 2018.10荷兰埃因霍温理工大学 & 德国柏林工业大学,多核系统双硕士(欧盟全额奖学金)
- 2011.08 – 2015.06华中科技大学光学与电子信息学院,电子科学与技术学士(国家留学基金委资助赴德国亚琛工业大学交换)
研究方向
- 高效大模型架构
- 低比特量化、线性模型、稀疏注意力。
- 软硬件协同的人工智能系统
- GPU/NPU 算子优化、跨模型缓存复用。
- 稀疏计算和图智能
- 稀疏矩阵计算、并行图算法、图神经网络系统。
学术论文
Sparsity in Linear Attention Models: Bregman Testing Time Learning.
Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2026, to appear.
ThunderGNN: Unlocking Tensor Cores for Graph Neural Networks.
Proceedings of the VLDB Endowment (PVLDB), vol. 19, no. 10, pp. 2699–2712, 2026.
High-Throughput Non-uniformly Quantized 3-bit LLM Inference.
Proceedings of the 31st ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming (PPoPP), pp. 288–300, 2026.
Rebo: Locality-Aware Graph Processing via Reordering and Blocking.
Proceedings of the 40th IEEE International Parallel and Distributed Processing Symposium (IPDPS), pp. 1037–1049, 2026. Best Paper Nomination (3 of 550 papers)
ToT: Triangle Counting on Tensor Cores.
IEEE Transactions on Parallel and Distributed Systems (TPDS), vol. 36, no. 12, pp. 2679–2692, 2025.
Groot: Graph-Centric Row Reordering with Tree for Sparse Matrix Multiplications on Tensor Cores.
Proceedings of the 20th European Conference on Computer Systems (EuroSys), pp. 803–817, 2025.
POSTER: Triangle Counting on Tensor Cores.
Proceedings of the 30th ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming (PPoPP), pp. 560–562, 2025.
Accelerating SpMV for Scale-Free Graphs with Optimized Bins.
Proceedings of the 40th IEEE International Conference on Data Engineering (ICDE), pp. 2407–2420, 2024.
Efficient SpMV for Graph Matrices Through Vectoring and Caching.
Proceedings of the 30th European Conference on Parallel and Distributed Processing (Euro-Par), LNCS vol. 14803, pp. 356–370, 2024.
Bitmap-Based Sparse Matrix-Vector Multiplication with Tensor Cores.
Proceedings of the 53rd International Conference on Parallel Processing (ICPP), pp. 1135–1144, 2024.
An Unequal Caching Strategy for Shared-Memory Graph Analytics.
IEEE Transactions on Parallel and Distributed Systems (TPDS), vol. 34, no. 3, pp. 955–967, 2023.
Connectivity-Aware Link Analysis for Skewed Graphs.
Proceedings of the 52nd International Conference on Parallel Processing (ICPP), pp. 482–491, 2023.
Workload Balancing via Graph Reordering on Multicore Systems.
IEEE Transactions on Parallel and Distributed Systems (TPDS), vol. 33, no. 5, pp. 1231–1245, 2022.
HiPa: Hierarchical Partitioning for Fast PageRank on NUMA Multicore Systems.
Proceedings of the 50th International Conference on Parallel Processing (ICPP), Article 24, pp. 1–10, 2021.
POSTER: Corder: Cache-Aware Reordering for Optimizing Graph Analytics.
Proceedings of the 26th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming (PPoPP), pp. 472–473, 2021.
教学经历
- 计算机体系结构Computer Architecture
- 操作系统Operating System
- 编译器设计Compiler Design
- 程序设计方法导论Introduction to Programming Methodology
技术笔记
关于现代 C++、图算法与性能优化的学习记录。