Shixun Wu

Shixun Wu

Phd Candidate at UC, Riverside

---
permalink: /
title: "About Me"
excerpt: "About me"
author_profile: true
redirect_from: 
  - /about/
  - /about.html
---

About Me

I am a Ph.D. student in Computer Science at the University of California, Riverside, advised by Prof. Zizhong Chen.

Previously, I received my M.S. in Electrical Engineering from Columbia University and B.S. in Computer Science (with a double major in Economics) from Peking University.

My research interests lie broadly in High-Performance Computing (HPC), Reinforcement Learning, Fault Tolerance & Resilience in Deep Learning Systems, Parallel, Distributed & Heterogeneous Systems, and Lossy Compression & Data Management.

News

  • June 2023: Gave a talk at the 37th ACM International Conference on Supercomputing (ICS 2023).
  • April 2023: Our work on high-performance and fault-tolerant GEMM on GPUs was accepted by ICS 2023.
  • …

(若需要，你可在此处按时间顺序补充更多动态)

Education

  • Ph.D. in Computer Science (Sep. 2022 – Present)
    University of California, Riverside (UCR)
    Advisor: Prof. Zizhong Chen

  • M.S. in Electrical Engineering (Sep. 2020 – May 2022)
    Columbia University

  • B.S. in Computer Science (Sep. 2016 – Jul. 2020)
    B.S. in Economics (Double Major)
    Peking University

Research Experience

  1. USC ISI / Argonne National Laboratory (Jan. 2024 – Present)
    Los Angeles, CA / Lemont, IL
    Scientific Workflow Applications on Resilient Metasystem
    Mentors: Dr. Franck Cappello, Dr. Sheng Di, Dr. Krishnan Raghavan (ANL); Dr. Ewa Deelman (USC ISI)
    • Designed a Q-learning + GNN-based topology protocol (DGRO) that reduces network diameter by optimizing virtual rings over heterogeneous, failure-prone systems.
    • Implemented a single-hop gossip-based failure detector, resilient to network jitter and churn, enabling decentralized membership monitoring across 20+ globally distributed sites.
    • Deployed DGRO on the FABRIC testbed spanning Japan, Europe, Hawaii, and 15+ U.S. locations, demonstrating fast convergence and robustness at international scale.
  2. UCR / Lawrence Berkeley National Laboratory (Sep. 2022 – Present)
    Riverside, CA
    Data-driven Exascale Control of Optically Driven Excitations in Chemical and Material Systems
    Mentor: Dr. Zizhong Chen
    • Designed and implemented in-kernel ABFT GEMM using tensor cores, achieving higher performance than cuBLAS while ensuring fault detection and correction under soft errors.
    • Developed a fully GPU-resident ABFT FFT pipeline, outperforming cuFFT, and enabling error-resilient spectral analysis in scientific simulations.
    • Proposed the first ABFT-enabled K-means clustering framework on GPUs, exceeding cuML performance with integrated resilience support.
    • Innovated lightweight, low-overhead in-kernel fault tolerance mechanisms across linear algebra and ML workloads, demonstrating resilience-performance co-design in exascale systems.
  3. Nvidia (Jun. 2024 – Sep. 2024)
    Santa Clara, CA
    Compiler Optimization for OpenMP Target Offload on Heterogeneous GPU Architectures
    Mentor: Dr. David Appelhans
    • Investigated performance bottlenecks of OpenMP target offload in SPEChpc 2021 on GH200/H200 GPUs.
    • Developed compiler/runtime optimizations achieving up to 10× speedup without source code changes.
    • Analyzed OpenMP vs. OpenACC performance and contributed optimized versions to SPEChpc 1.1.9.
    • Work adopted by RWTH Aachen University, demonstrating both research impact and practical relevance.
  4. Columbia University / AI4Finance Foundation (Aug. 2021 – Jul. 2022)
    New York, NY
    ElegantRL: Massively Parallel Deep Reinforcement Learning Library
    Mentors: Dr. Xiaoyang Liu, Dr. Xiaodong Wang
    • Developed multi-agent RL algorithms in ElegantRL, a popular RL library with ~4k GitHub stars.
    • Co-led ElegantRL_Solver, a high-performance solver that outperforms Gurobi for dense MaxCut problems.

Publications

For the complete list, please visit my Google Scholar.

Conference Papers

  • [PPoPP ‘25]
    Shixun Wu, Yujia Zhai, Jinyang Liu, Jiajun Huang, Zizhe Jian, Sheng Di, Franck Cappello, Zizhong Chen.
    TurboFFT: Co-Designed High-Performance and Fault-Tolerant Fast Fourier Transform on GPUs.
    ACM Symposium on Principles and Practice of Parallel Programming (PPoPP), 2025.
    [Paper]

  • [Cluster ‘24]
    Shixun Wu*, Yitong Ding*, Yujia Zhai, Jinyang Liu, Jiajun Huang, Zizhe Jian, Huangliang Dai, Sheng Di, Bryan Wong, Zizhong Chen, Franck Cappello.
    FT K-means: A High-Performance K-means on GPU with Fault Tolerance.
    2024 IEEE International Conference on Cluster Computing (CLUSTER).
    [Paper]

  • [SC ‘24]
    Jinyang Liu*, Jiannan Tian*, Shixun Wu*, Sheng Di, Boyuan Zhang, Robert Underwood, Yafan Huang, Jiajun Huang, Kai Zhao, Guanpeng Li, Dingwen Tao, Zizhong Chen, Franck Cappello.
    cuSZ-I: High-Fidelity Error-Bounded Lossy Compression for Scientific Data on GPUs.
    SC24: International Conference for High Performance Computing, Networking, Storage and Analysis, 2024.
    [Paper]

  • [ICS ‘23]
    Shixun Wu*, Yujia Zhai*, Jinyang Liu, Jiajun Huang, Zizhe Jian, Bryan Wong, Zizhong Chen.
    Anatomy of High-Performance GEMM with Online Fault Tolerance on GPUs.
    37th ACM International Conference on Supercomputing (ICS), 2023.
    [Paper]

  • [IPDPS ‘24]
    Zizhe Jian, Sheng Di, Jinyang Liu, Kai Zhao, Xin Liang, Haiying Xu, Robert Underwood, Shixun Wu, Jiajun Huang, Zizhong Chen, Franck Cappello.
    CliZ: Optimizing Lossy Compression for Climate Datasets with Adaptive Fine-tuned Data Prediction.

  • [SIGMOD ‘24]
    Jinyang Liu, Sheng Di, Kai Zhao, Xin Liang, Sian Jin, Zizhe Jian, Jiajun Huang, Shixun Wu, Zizhong Chen, Franck Cappello.
    High-performance Effective Scientific Error-bounded Lossy Compression with Auto-tuned Multi-component Interpolation.

  • [BigData ‘23]
    Jiajun Huang, Jinyang Liu, Sheng Di, Yujia Zhai, Zizhe Jian, Shixun Wu, Kai Zhao, Zizhong Chen, Yanfei Guo, Franck Cappello.
    Exploring Wavelet Transform Usages for Error-bounded Scientific Data Compression.

  • [Allerton ‘23]
    Jeremy Johnston, Xiaoyang Liu, Shixun Wu, Xiaodong Wang.
    Downlink beamforming optimization via deep learning.
    59th Annual Allerton Conference on Communication, Control, and Computing (Allerton), 2023.

Journal Papers

  • [IJHPCA ‘25]
    Balaprakash Prasanna, Krishnan Raghavan, Franck Cappello, Ewa Deelman, Anirban Mandal, Hongwei Jin, Imtiaz Mahmud, Komal Thareja, Shixun Wu, Pawel Zuk, Mariam Kiran, Zizhong Chen, Sheng Di, Kesheng Wu.
    SWARM: Reimagining Scientific Workflow Management Systems in a Distributed World.
    International Journal of High Performance Computing Applications, 2025.

  • [TSP ‘23]
    Jeremy Johnston, Xiaoyang Liu, Shixun Wu, Xiaodong Wang.
    A Curriculum Learning Approach to Optimization with Application to Downlink Beamforming.
    IEEE Transactions on Signal Processing, 2023.

Poster Papers

  • [HPDC ‘23]
    Shixun Wu*, Yujia Zhai*, Jiajun Huang, Zizhe Jian, Zizhong Chen.
    FT-GEMM: A Fault Tolerant High Performance GEMM Implementation on x86 CPUs.
    The 32nd ACM International Symposium on High-Performance Parallel and Distributed Computing (HPDC), 2023.
    [Poster]

Workshop Papers

  • [SC ‘24 Workshop]
    Ewa Deelman, Prasanna Balaprakash, Mariam Kiran, Anirban Mandal, Krishnan Raghavan, Sheng Di, Franck Cappello, John Wu, Zizhong Chen, Shixun Wu, Hongwei Jin, Cong Wang, Imtiaz Mahmud, Komal Thareja, Erik Scott, Pawel Zuk, Aiden Hamade.
    SWARM: Scientific Workflow Applications on Resilient Metasystem.

  • [ICLR ‘23 Workshop]
    Xiaoyang Liu, Zechu Li, Shixun Wu, Xiaodong Wang.
    Stationary Deep Reinforcement Learning With Quantum K-Spin Hamiltonian Regularization.
    ICLR 2023 Workshop on Physics for Machine Learning, 2023.

arXiv (In Submission)

  1. Shixun Wu, Sheng Di, Krishnan Ragahvan, Kesheng Wu, Ewa Deelman, Franck Cappello.
    DGRO: Diameter-Guided Ring Optimization for Integrated Research Infrastructure Membership.

  2. Huangliang Dai, Shixun Wu, Zizhong Chen.
    FT-Transformer.

  3. Shixun Wu, Zizhong Chen.
    TurboFNO: High-Performance Fourier Neural Operator with Fused FFT-GEMM-iFFT on GPU.

  4. Shixun Wu*, Jinyang Liu*, Jiannan Tian*, Jinwen Pan*, Sheng Di, Zizhong Chen, Franck Cappello.
    Optimizing GPU-Based Error-Bounded Lossy Compression with Advanced Interpolation and Synergistic Lossy-Lossless Scheme.

Professional Services

  • Reviewer, Parallel Computing, 2025
  • SubReviewer, ICS (International Conference on Supercomputing), 2025
  • SubReviewer, CCGrid (International Symposium on Cluster, Cloud and Grid Computing), 2025
  • Reviewer, GPGPU (Workshop on General Purpose Processing Using GPU), 2025
  • Reviewer, ICDCS (International Conference on Distributed Computing Systems), 2025
  • SubReviewer, IPDPS (International Parallel and Distributed Processing Symposium), 2024, 2023
  • Reviewer, Computing Surveys, 2023
  • Reviewer, Journal of Intelligent & Fuzzy Systems, 2023

Projects Involved

  • DOE SWARM: Scientific Workflow Applications on Resilient Metasystem
  • DOE DECODE: Data-driven Exascale Control of Optically Driven Excitations in Chemical and Material Systems

Teaching

  • Teaching Assistant, CS161: Design & Architecture of Computer Systems, University of California, Riverside
    • Fall 2024, Spring 2024
  • Teaching Assistant, CS160: Concurrent Programming and Parallel Systems, University of California, Riverside
    • Fall 2023

Honors and Awards

  • NSF PPoPP Travel Grant (Feb. 2025)
  • Outstanding Teaching Award, University of California, Riverside (May 2024)
  • Third Prize, UCRPC, University of California, Riverside (Nov. 2023)
  • Distinguished Dean’s Fellowship, University of California, Riverside (Sep. 2022)
  • Second Prize, PKU ACM (2017, 2018)
  • PKU May 4th Scholarship (2017)