permalink | title | excerpt | author_profile | redirect_from | ||
---|---|---|---|---|---|---|
/ |
true |
|
{% if site.google_scholar_stats_use_cdn %} {% assign gsDataBaseUrl = "https://cdn.jsdelivr.net/gh/" | append: site.repository | append: "@" %} {% else %} {% assign gsDataBaseUrl = "https://raw.githubusercontent.com/" | append: site.repository | append: "/" %} {% endif %} {% assign url = gsDataBaseUrl | append: "google-scholar-stats/gs_data_shieldsio.json" %}
I am studying at University of Edinburgh for my PhD degree. I got my Master degree from Eindhoven University of Technology and my bachelor degree from Hefei University of Technology(HFUT), China.
I am interested in VLSI/ASIC/SoC Design, NoC, Neuromorphic Hardware and Efficient AI Hardware Accelerator.
Here is my CV.
- 2025.01: 🎉🎉 I started my PhD study in University of Edinburgh, supervised by Dr. Shady Agwa and Professor Themis Prodromakis.
- 2024.11: 🎉🎉 I finished my internship at NXP Semiconductors and successfully defenced my Master Thesis with 8.5/10.
- 2024.09: 🎉🎉 I got a conditional PhD offer at the University of Edinburgh.
- 2024.09: 🎉🎉 I also received 2 PhD interviews from National University of Singapore(NUS) and University of Groningen.
- 2024.09: 🎉🎉 I finished the half-way presentation at TU/e, committee member: Marc Geilen, Manil Dev Gomony and Chengmin Li.
- 2024.07: 🎉🎉 I received 2 PhD interview from University College Dublin(one from mix-signal group, and one from digital group).
- 2024.06: 🎉🎉 I received my first PhD interview from RWTH Aachen University.
- 2024.03: 🎉🎉 Congratulations Intrinsic ID is acquired by Synopsys.
- 2024.03: 🎉🎉 I change my master thesis to NXP Semiconductors in Nijmegen, supervised by Kimmo Salo(NXP), Gustavo Naspolini(NXP).
- 2023.11: 🎉🎉 I started working as a Oversea VC Intern(Remote) at Linear Capital.
- 2023.10: 🎉🎉 I finished my internship at Intrinsic ID and got 8.5 of internship. I also start my research on Neuromorphic at TU/e.
- 2023.07: 🎉🎉 I will be an intern at Intrinsic ID for 3.5 months, supervised by Rui Wang(Intrinsic ID), Roel Maes(Intrinsic ID) and Manil Dev Gomony(TU/e & Nokia Bell Labs).
- 2022.09: 🎉🎉 I started my master's study at TU/e.
- 2022.07: 🎉🎉 I am graduated from Hefei University of Technology with a bachelor degree, supervised by Zhenmin Li.
- Professional: Verilog/SystemVerilog/VHDL · FPGA · Lint · Linux · Cadence Tool(Xcelium,SimVision,Conformal Lint,Virtuoso,Genus,Innovus)
- Miscellaneous: Python · C/C++ · SystemC · MATLAB/Simulink · LaTex · CUDA · Pytorch · Perl/TCL/Shell · Git
- Language: Mandarian(Native), TOEFL(iBT) 92
- PhD student of Engneering
- Supervisor: Dr. Shady Agwa and Professor Themis Prodromakis
- Research interest:
- Msc.Eletrical Engneering(track:Electronic System)
- GPA: 7.8/10
- Relevant Course: Digital integrated circuit design, Embedded computer architecture, Electronic design automation, Applied combinatorial algorithms, Intelligent architectures(focus on DNN/Hardware co-design), Systems on silicon(focus on SoC backend), Neuro computation(focus on Neuromorphic computing).
- Thesis: LinkBo: A Robust Low-Latency 1-Wire Protocol for Chip-to-Chip Communications (8.5/10)
- Bachelor in Integrated Circuit Design and Integrated Systems
- GPA: 83.1/100 (TOP 22%)
- Relevant Class: Analysis and Design of Integrated Digital Circuit, Microprocessor architecture and design, Introduction to SoC design, Verilog HDL and FPGA implementation and so on.
- Thesis: The Research and Implementation of Router for Packet-Connect-Circuit Network-on-chip
-
Digital Design Intern
I am working in AA-APS, this thesis project supervised by Kimmo Salo(NXP), Gustavo Naspolini(NXP), and Manil Dev Gomony(Bell Lab & TU/e). My work as follow:
- Define a custom 1-wire digital communication protocol between two ICs.
- Develope high-level model in Simulink and test the channel model with parasitic parameter.
- Develop and verify an IP module implementing the new communication protocol using SystemVerilog.
- Use Cadence Xcelium to simulate, Cadence SimVision to see the wave, Cadence Conformal Lint to lint and DesignSync to verision control.
- Build a prototype demonstrator with 2 FPGAs.
-
Oversea VC Intern
During my internship at Linear Capital, my responsibilities included:
- Mapping overseas talent in the academic and industrial sectors, providing vital insights for project recruitment and business collaborations.
- Conducting assessments for the commercialization and implementation of potential projects, analyzing market opportunities and risks to support investment decisions.
- Participating in the planning and execution of early-stage investments, collaborating with startups to drive project success.
-
IP Modeling and Digital IC Design Intern
I worked in Research and Security Department(now SoC Security of Synopsys). This internship supervised by Manil Dev Gomony(Bell Lab & TU/e) and Rui Wang & Roel Maes(Intrinsic ID). My work as follow:
- Study on a trellis-based Reed-Muller codec.(Algorithm)
- Use Python modeling it as a digital IP module.(software)
- Design the architecture of codec and implementation by VHDL.(harware)
- Verify on Arty-z7 FPGA(zynq) with vivado and vitis.(Embeded System)
- Under the 50MHz condition, after several improvements, the hardware resource consumption decreased from 938 LUTs to 843 LUTs. Meanwhile, the decoding latency reduced to 6 clock cycles, and continuous decoding became possible.
-
Collaborate with Dr.Changchun Zhou(PhD@PKU & postdoc@DUKE)
- Design hardware acceleration units for 3D point cloud applications, include Farthest Point Sampling(FPS), AXI interface(Master/Slave).
-
Eindhoven University of Technology, Netherlands
- I have completed the literature review on Neuromorphic Hardware Accelerator and Neuromorphic NoC.
- I have started architectural simulation using OpenSoC for the baseline NoC.
- Work in Eletronic System group with Prof. Manil Dev Gomony, Prof. Federico Corradi, and Prof. Henk Corporaal.
-
Institute of VLSI Design of HFUT, China
- Study the knowledge of Network-on-Chip(NoC) Router Based on Packet Connected Circuit(PCC).
- Implement the router and routing algorithm of PCC-NoC by using VerilogHDL.
- Verify it on FPGA and use UART(with FIFO) to communicate with PC.
- Use Python to verify result automatically.
- This project as my bachelor graduation project got A and advised by Zhenmin Li(HFUT).
2024.06 - 2024.09, Tiny SoC based on Rsic-V processor and Tiny LeViT accelerator(hobby project)
- Designed the SoC (System on Chip) architecture, incorporating an RISC-V core, memory, Levit accelerator, and AHB/APB buses with connected peripherals. Utilized previous project designs for the RISC-V core and Levit accelerator.
- Developed a Python program to convert assembly code to binary, enabling rapid generation of binary instruction files to meet SoC requirements.
- Constructed a warpper for the accelerator with input and output buffers, ensuring synchronized data input when all required data was present.
- Implemented AHB-compatible interfaces for the CPU, memory, and accelerator using SystemVerilog.
2024.04 - 2024.06, Tiny LeViT Hardware Accelerator(hobby project)
- Use System Verilog to design hardware accelerator for accelerate LeViT Network which contain Convolutional layer, Attention layer and Average pooling layer.
- Use row stationary (RS) and systolic array to get max parallel computing. The delay is only 3 cycle from input data to first output data. Also, it has specific core to accelerate the convolutional layer when stride=2 and padding=1.
- Use Tanh instead of softmax and use ReLU instead of Hardswish to simplify that difficulty of hardware calculation in attention layer.
- Used Verilog to design a SoC which include five-stage pipelined mMIPS processor core, AES encryption module, and AMBA bus and Used Cadence Incisive to simulation and functional verification.
- Used Cadence Genus for logic synthesis with low power strategies which is reduce 3% power consumption under 125MHz.
- Used Cadence Innovus for place and route with two power domain which is reduce 8% power consumption under 200MHz.
- Train a multilayer perceptron for handwritten digit classification(MNIST) using the PyTorch framework.
- Optimize a VGG5 for image classification using various quantization and pruning techniques. Explore the impact of these techniques on both accuracy and compute cost.
- Use open source Tensil AI for generating tensor computing units(TCU), compiling and accelerating ResNet20 by systolic array on PYNQ .
- I have implemented a RISC-V five-stage pipeline processor with full hazard handling. The RTL level design using Verilog and simulated in Modelsim for simulation.
- The processor can run the basic RV32i instructions, solves data conflicts, structure conflicts, and control conflicts, and supports stalling, flushing and forwarding.
- Completed CMOS circuit design and layout design for a 45nm full custom 16-bit Brent-Kung adder. Circuit design and layout design via Cadence Virtuoso, DRC and LVS verification of the layout using Calibre. The circuit design can be functionally verified by simulation at 500MHz, 90◦C with an output rise and fall time of less than 100ps, and the layout results can be functionally verified at a post-simulation of 500MHz.
- Mapping Grayscale processing and convolution 2D kernels from C to CUDA and optimaize the loop, then running on the Nvidia GPU.
- The result is that the processing time of 13 images is accelerated from 4872ms to 27ms.
- The CPU is designed in modules, and each module uses Verilog HDL to design, which is implemented on the Basys3 after being verified by Modelsim pre-imitation. The complex C language load can run, and UART serial communication function controlled by softcore can be realized.
- Studying the structural design of soc, and use SystemC to write Bus and UART serial interface, function processing module, arbiter module, data receiving and sending module to form a lightweight system and verify it. Drive data enters two processing modules, and one of the results is selected by the arbiter in the bus and sent to the receiving module through UART interface and displayed.
- Use VCS to verify the function of Tinycore based on RISC-V on the test platform.
- Use DC compiler to synthesize four different hardware description ALUs, and analyze the comprehensive report to compare and analyze different parameters.
-
System on Chip (SOC) Architecture: A Practical Approach,Veena S. Chakravarthi & Shivananda R. Koteshwar.
- Provides the most up-to-date information on current SOCs and architectural insights for the design of future semiconductor systems
- Explores concepts such as parallelism, pipelining, data-driven or instruction-driven, and event-driven systems and their respective tradeoffs in SOC architectures
- Provides a practical approach to defining SOC architecture with real case studies
-
A Practical Approach to VLSI System on Chip (SoC) Design: A Comprehensive Guide,Veena S. Chakravarthi.
- A comprehensive practical guide for VLSI designers
- Covers end-to-end VLSI SoC design flow
- Includes source code, case studies, and application examples
-
The Missing Semester of Your CS Education, MIT.
- Content: Shell Tools and Scriptin, Editors (Vim), Data Wrangling, Command-line Environment, Version Control (Git). Debugging and Profiling, Metaprogramming, Security and Cryptography, Potpourri.
-
How to Start a Startup, Sam Altman in Stanford.
- Current progress: Lecture 3.
-
TinyML and Efficient Deep Learning Computing, MIT HAN LAB.
- Content: Basics of Deep Learning, Pruning and Sparsity, Quantization, Neural Architecture Search, Knowledge distillation.
-
一生一芯ysyx, Institute of Computing technology, CAS.
- Current progress: Prestudy->Stop.
-
SoC 101, Adam Teman(Bar-Ilan University).
- Focus on SoC architecture.
-
Digital Design and Computer Architecture, ETH.
- beignner.
- 2022, My bachelor thesis of NoC get A grade.
- 2021/2022, Unergraduate Scholarship.
Director, External Relations Department
- Participated in writing the introduction of the center. Led the Computer Science College's party branch in visiting and introducing the Big Data Center and led high school students from Hefei to visit the center.
- Organized and planned the Innovation and Entrepreneurship Forum at the Big Data Center. Invited teachers from various colleges to give lectures. Attracted active participation from 500 students across the university.
- Participated in editing the WeChat official account of INOW Creators.
Director, Organizing Department
- Responsible for the Organization Department of the new District's daily work, organized and planned the new district association recruitment activities, thus the association became the largest association at our university.
- Coordinated the students' Union and other departments, carried out targeted basic teaching work, held "no innovation, not young" electronic science and technology exchange lectures and other activities, the association was rated as the annual model association.
Director, Innovation and Entrepreneurship Department
- Responsible for organizing and promoting the Microelectronics College Innovation and Entrepreneurship Competition and the National College Student Electronics Design Competition training.
- Actively collaborated with the Big Data Center to facilitate the entry of Microelectronics College's innovation teams into the center.