INSTALL.md

Installation Guide

Table of Contents

• Build from Source
  • Install Dependencies
  • Install Gurobi (Optional)
• Build with Docker
  • Docker Image Options
  • Running the Docker Container
  • Enter Container
• Build and Install OpenPARF
  • Get the Source
  • Install
  • Build Options (Optional)

OpenPARF provides two ways to build and install:

• Build from source
• Build with Docker

Build from Source

Install Dependencies

We recommend using Anaconda/Mamba for dependency management:

# Create and activate conda environment
mamba create --name openparf python=3.7
mamba activate openparf

# Install common packages
mamba install cmake boost bison

# Install PyTorch 1.7.1
mamba install pytorch==1.7.1 torchvision==0.8.2 cudatoolkit=11.0 -c pytorch

# Install Python packages
pip install hummingbird-ml pyyaml networkx tqdm

Install Gurobi (Optional)

1. Download gurobi9.5.1_linux64.tar.gz
2. Extract to <your Gurobi home>
3. Obtain a license from Gurobi website
4. Set environment variables in ~/.bashrc:

export GUROBI_HOME="<your Gurobi home>/linux64"
export GRB_LICENSE_FILE="<your Gurobi license path>"
export PATH="${PATH}:${GUROBI_HOME}/bin"
export LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:${GUROBI_HOME}/lib"

Build with Docker

Docker Image Options

Using Pre-built Images

docker pull magic3007/openparf:2.0

Building the Image Yourself

cd <source directory>/docker
docker build . -t openparf:2.0 -f openparf.dockerfile

Running the Docker Container

Without CUDA Support:

docker run -itd --restart=always --network host -e TERM=$TERM \
  --name openparf \
  -v /etc/localtime:/etc/localtime:ro \
  -v <project directory on host>:/root/OpenPARF \
  -v <benchmark directory on host>:/root/benchmarks \
  openparf:2.0 \
  /bin/bash

With CUDA Support:

docker run -itd --restart=always --network host -e TERM=$TERM \
  --name openparf \
  --gpus all \
  -v /etc/localtime:/etc/localtime:ro \
  -v <project directory on host>:/root/OpenPARF \
  -v <benchmark directory on host>:/root/benchmarks \
  openparf:2.0 \
  /bin/bash

Enter Container

docker exec -it openparf /bin/bash

Build and Install OpenPARF

Get the Source

git clone --recursive https://github.com/PKU-IDEA/OpenPARF.git

Install

mkdir build
cd build
cmake ../OpenPARF -DCMAKE_PREFIX_PATH=$CONDA_PREFIX -DPYTHON_EXECUTABLE=$(which python) -DPython3_EXECUTABLE=$(which python) -DCMAKE_INSTALL_PREFIX=<installation directory>
make -j8
make install

Build Options (Optional)

Adjustable CMake variables:

• CMAKE_INSTALL_PREFIX: Installation directory
• CMAKE_BUILD_TYPE: Release/Debug (Default: Release)
• USE_CCACHE: Enable ccache (Default: OFF)
• ENABLE_ROUTER: Enable router compilation (Default: OFF)