In this section, we demonstrate how to set up the environment for fpgaConvNet.
As a prerequisite, install the following software on your machine:
-
miniconda: https://docs.conda.io/projects/miniconda/en/latest/
-
grahviz: https://graphviz.org/download/
-
protocol buffer compiler: https://grpc.io/docs/protoc-installation/
-
Vivado 2019.1: https://www.xilinx.com/support/download.html
-
Vivado Y2K22 patch: https://support.xilinx.com/s/article/76960?language=en_US
Note for using Vivado, you may set environment variable as instructed here.
Afterwards, create a new conda environment named as fpgaconvnet-tutorial
and run the provided script setup.sh
to finish the setup
conda create -n fpgaconvnet-tutorial python=3.10
conda activate fpgaconvnet-tutorial
./setup.sh
To check the installation is successful, run the following commands in your bash shell, and make sure there is no error showing.
python -c "import torch"
python -c "import fpgaconvnet"
The rest of the tutorial is built on Jupyter notebook, and simply run notebook_shell.sh
to continue.
./notebook_shell.sh