Home

I am an associate professor in State Key Laboratory of Processors (SKLP), Institute of Computing Technology (ICT), Chinese Academy of Sciences (CAS). I got my B.Eng degree and M.Eng degree from Harbin Institute of Technology, and Ph.D degree from The University of Hong Kong advised by Prof. Hayden So in 2016. I worked as a research fellow in National University of Singapore from 2016 to 2018, and thereafter, I joined ICT as an associate professor. My current research interest focuses on domain specific architecture and system, LLM for Chip Design.

Vancancies

I am looking for self-motivated master/intern students for LLM-based intelligent chip designs. It includes various design tasks such as RISC-V SoC design and verification, RISC-V ASIP design automation, domain-specific accelerator generation, Low-power design and optimization. Students with RISC-V processor design and LLM experience are highly preferred. It is possible to work fully remotely. Check the topics in the following list and contact me if you are interested.

  • LLM-based SoC/DSA/ASIP design
  • LLM-based high-level synthesis
  • LLM-based ASIC design, verification, and debugging
  • LLM-based EDA parallelization

News

  • [Jun 2020] Shengwen Liang and Rick Lee won the Third Prize (FPGA Track) of the 2020 IEEE Low-Power Computer Vision Challenge (LPCVC).
  • [Jul 2022] Haitong Huang, Erjing Luo, and Cangyuan Li in my group got the Third Place in DAC’22 SDC.
  • [Jan 2023] Our work “EnGN: A High-Throughput and Energy-Efficient Accelerator for Large Graph Neural Networks” won the Best Paper Award of TC’21.
  • [Jun 2023] Congratulations! Haitong Huang, Erjing Luo, and Guoyu Li got the Second Place in DAC’23 SDC.
  • [Nov 2023] A HW/SW co-design framework for mixed-precision neural network acceleration on FPGAs (DeepBurning-MixQ) is accepted by ICCAD’23. We are adding LLM support to enable more intelligent code generation. The code is open sourced on github and welcome to try it.
  • [Mar 2024] A multi-resolution fault injection framework for deep learning is accepted by TVLSI’24. It is well documented and validated sufficiently, and provides many convenient evaluation features that are typically required in fault-tolerant deep learning study. Check it on github.
  • [May 2024] Our work on computational storage got Huawei OlympusMons Awards in 2024.
  • [Jun 2024] HLSPilot: LLM-based High-level Synthesis is accepted by ICCAD’24. This work presents an LLM-based high-level synthesis design framework for typical CPU-FPGA architecture. Particularly, it leverages LLM and RAG techniques to generate HLS-based accelerator design for arbitrary sequential C/C++ code, which is also a key component for our LLM-driven SoC generation framework.
  • [Oct 2024] Our work on CIM-based LLM acceleration collaborated with Dr. Luo is accepted by IEEE TPAMI.
  • [Oct 2024] Start to serve as an associate editor of IEEE Transactions on Emerging Topics in Computing.
  • [Nov 2024] Two papers about DSA are accepted by HPCA’25.

Fundings

  • HW/SW platform for intelligent cross-layer design and optimization of processors, 11 000 000¥, xdb(2024-2028)
  • Natural weather disturbed image generation for remote sensing, 500 000¥, XXX(2023-2024)
  • AI assisted DSA design automation, 300 000¥, SKLP(2023-2024)
  • Automatic Cross-Layer DSA Design and Optimization, 1 600 000¥, National Key Research and Development Program of China(2023-2025)
  • Intelligent In-Storage Big Data Processing System, 1 300 000¥, 1XX(2023-2024)
  • Fault-tolerant Deep Learning Toolchain for COTS Devices, 300 000¥, XXX(2023)
  • Elastic Fault-tolerant Deep Learning Processor Design, 570 000¥, NSFC(2022-2025)
  • Customized Energy-efficient Graph Processing Acceleration on FPGAs, 300 000¥, NSFC(2020-2022)
  • Fault-tolerant Deep Learning Processor Design Automation, 300 000¥, SKLCA(2021-2022)

Talks

  • 基于大语言模型的全自动CPU-FPGA异构硬件加速, 高性能异构计算与人工智能优化论坛,CCF-HPC, 2024
  • 异构硬件加速器设计、优化以及编程, 高性能异构计算与人工智能优化论坛,CCF-HPC,2023
  • 基于计算存储器的图处理系统设计, 面向复杂图计算应用的新型高能效体系结构论坛,CNCC,2022
  • 容错深度学习处理器微结构设计,集成电路设计与自动化学术会议(CCF-DAC), 2021
  • DeepBurning2.0: An Automatic End-to-end Neural Network Acceleration System on FPGAs, 华南理工大学,软件学院, 2019

Services

  • AE of IEEE Transactions on Emerging Topics in Computing
  • TPC for DFTS’22, FPT’22, ITC’22
  • TPC for ATS’23, FPT’23, DFTS’23, ITC’23, NeurIPS’23
  • TPC for DFTS’24, FPT’24, ICLR’24, FCCM’24, ICML’24
  • Review for TC, TPDS, TCAD. TVLSI, TETC, JETC, JSA, TNNLS

Graduated Students

  • Xinghua Xue (Ph.D, Hangzhou Institute of Advanced Study, UCAS)
  • Zheng Feng (Master student, Huawei, AI Compilation)
  • Guoyu Li (Master student, Baidu, Kunlun Chip)
  • Haitong Huang (Master student, Tencent, AI Lab)
  • Xuejian Sun (Intern from Harbin Institute of Technology, Master student in Fudan University)
  • Erjing Luo (Intern from Beijing Institute of Technology, Mphil in University of Alberta)
  • Miaoxin Wang (Intern from Harbin Institute of Technology, Master student in Nanjing University)
  • Jinming Zhao (Intern from Wuhan University, PhD candidate in the University of Hong Kong)
  • Zhiyu Zhu (Intern from Harbin Institute of Technology, CETC 47)
  • Cheng Chu (Intern from Hefei University of Technology, PhD candidate in Indiana University Bloomington)
  • Meng He (Intern from Hefei University of Technology, 字节跳动)
  • Ziyang Zhu (Intern from Hefei University of Technology, 紫光展锐)
  • Qiang Zhang (Master Student, 快手)
  • Kouzi Xing (Intern from Hefei University of Technology, 商汤科技)
  • Li Li (Intern from Hefei University of Technology, 京东)
  • Kexin Chu (Intern from Hefei University of Technology, 百度)
  • Kaijie Tu (Intern from Hefei University of Technology, 计算所)
  • Chang Shi (Master student, Alibaba)
  • Peibin Wu (Master student, MSRA)