[xdesign] linear algebra SVD

Table of Contents

• About the Project
• Usage
• References
• Contributing

About The Project

Using high level synthesis to accelerate singular value decomposition (SVD), the most common matrix method.

Directory structure

README.md
code/
    svd.h
    svd.cpp
code_opt/
    svd.h
    svd.cpp
    hls_library/
        hls_svd.cpp
testdata/
    svd_tb.cpp
    pynq_python/
        svd.py
script/
    directives.tcl
    run_hls.tcl
impl/
    svd_top_csynth.rpt

Usage

1. fpga board setup

We use Xilinx ZedBoard Evaluation and Development Kit to evaulate this project

2. using HLS vivado to simulation and synthesis

3. export RTL

4. generate .bit from vivado

5. python test

 python svd.py

References

• xilinx vivado design examples

Contributing

• fix negtive slack
• experiment on different scale floating number