Compile on Linux from Source Code

Environment preparation

  • Linux version (64 bit)

    • CentOS 7 (GPU 版本支持 CUDA 11.0 - 12.0)

    • Ubuntu 18.04 (GPU 版本支持 CUDA 11.0 - 12.0)

    • Ubuntu 20.04 (GPU 版本支持 CUDA 11.0 - 12.0)

  • Python 版本 3.8/3.9/3.10/3.11/3.12 (64 bit)

Choose CPU/GPU

  • If your computer doesn’t have NVIDIA® GPU, please install CPU version of PaddlePaddle

  • If your computer has NVIDIA® GPU, and the following conditions are met,GPU version of PaddlePaddle is recommended.

    • CUDA toolkit with cuDNN cuDNN 8(for multi card support, NCCL2.7 or higher)

    • Hardware devices with GPU computing power over 6.0

    You can refer to NVIDIA official documents for installation process and configuration method of CUDA and cudnn. Please refer toCUDAcuDNN

Installation steps

There are two compilation methods under Linux system. It’s recommended to use Docker to compile. The dependencies required for compiling Paddle are pre-installed in the Docker environment, which is simpler than the native compiling environment.

Compile with Docker

Docker is an open source application container engine. Using docker, you can not only isolate the installation and use of paddlepaddle from the system environment, but also share GPU, network and other resources with the host

Compiling PaddlePaddle with Docker,you need:

Please follow the steps below to install:

1. First select the path where you want to store PaddlePaddle, then use the following command to clone PaddlePaddle’s source code from github to a folder named Paddle in the local current directory:

git clone https://github.com/PaddlePaddle/Paddle.git

2. Go to the Paddle directory:

cd Paddle

3. Pull PaddlePaddle image:

For domestic users, when downloading docker is slow due to network problems, you can use the mirror provided by Baidu:

  • CPU version of PaddlePaddle:

    docker pull registry.baidubce.com/paddlepaddle/paddle:latest-dev
    
  • GPU version of PaddlePaddle:

    docker pull registry.baidubce.com/paddlepaddle/paddle:latest-dev-cuda12.0-cudnn8.9-trt8.6-gcc12.2
    

If your machine is not in mainland China, you can pull the image directly from DockerHub:

  • CPU version of PaddlePaddle:

    docker pull paddlepaddle/paddle:latest-dev
    
  • GPU version of PaddlePaddle:

    docker pull paddlepaddle/paddle:latest-dev-cuda12.0-cudnn8.9-trt8.6-gcc12.2
    

In the above example, latest-dev-cuda11.2-cudnn8.2-trt8.0-gcc82 is only for illustration, indicating that the GPU version of the image is installed. If you want to install another cuda/cudnn version of the image, you can replace it with latest-dev-cuda12.0-cudnn8.9-trt8.6-gcc12.2 etc.

You can see DockerHub to get the image that matches your machine.

4. Create and enter a Docker container that meets the compilation environment:

  • Compile CPU version of PaddlePaddle:

    Using the image pulled from Baidu.

    docker run --name paddle-test -v $PWD:/paddle --network=host -it registry.baidubce.com/paddlepaddle/paddle:latest-dev /bin/bash
    
    • --name paddle-test: names the Docker container you created as paddle-test;

    • -v $PWD:/paddle: mount the current directory to the /paddle directory in the docker container (PWD variable in Linux will be expanded to absolute path of the current path);

    • -it: keeps interaction with the host;

    • registry.baidubce.com/paddlepaddle/paddle:latest-dev: use the image named registry.baidubce.com/paddlepaddle/paddle:latest-dev to create Docker container, /bin/bash start the /bin/bash command after entering the container.

    If you are using the image pulled from DockerHub, just modify the image name.

    docker run --name paddle-test -v $PWD:/paddle --network=host -it paddlepaddle/paddle:latest-dev /bin/bash
    
  • Compile GPU version of PaddlePaddle:

    Using the image pulled from Baidu.

    docker run --gpus all --name paddle-test -v $PWD:/paddle --network=host -it registry.baidubce.com/paddlepaddle/paddle:latest-dev-cuda12.0-cudnn8.9-trt8.6-gcc12.2 /bin/bash
    
    • --gpus all: gpu resources can be used in Docker container;

    • --name paddle-test: names the Docker container you created as paddle-test;

    • -v $PWD:/paddle: mount the current directory to the /paddle directory in the docker container (PWD variable in Linux will be expanded to absolute path of the current path);

    • -it: keeps interaction with the host;

    • registry.baidubce.com/paddlepaddle/paddle:latest-dev-cuda12.0-cudnn8.9-trt8.6-gcc12.2: use the image named registry.baidubce.com/paddlepaddle/paddle:latest-dev-cuda12.0-cudnn8.9-trt8.6-gcc12.2 to create Docker container, /bin/bash start the /bin/bash command after entering the container.

    If you are using the image pulled from DockerHub, just modify the image name.

    docker run --gpus all --name paddle-test -v $PWD:/paddle --network=host -it paddlepaddle/paddle:latest-dev-cuda12.0-cudnn8.9-trt8.6-gcc12.2 /bin/bash
    

Note: Please make sure to allocate at least 4g of memory for docker, otherwise the compilation process may fail due to insufficient memory.

When using GPU version of image, please make sure the NVIDIA Container Toolkit is successfully installed, or GPU can not be used in docker container. And the latest version of image is recommended, or some compiling error may occur.

5. After entering Docker, go to the paddle directory:

cd /paddle

6. Switch to develop version to compile:

git checkout develop

Paddle supports Python version 3.8 and above

7. Create and enter the /paddle/build path:

mkdir -p /paddle/build && cd /paddle/build

8. Use the following command to install the dependencies:

  • Install protobuf 3.20.2

pip3.10 install protobuf

Note: We used Python3.10 command as an example above, if the version of your Python is 3.8/3.9/3.11/3.12, please change pip3.10 in the commands to pip3.8/pip3.9/pip3.11/pip3.12

  • Installing patchelf, PatchELF is a small and useful program for modifying the dynamic linker and RPATH of ELF executables.

apt install patchelf

9. Execute cmake:

  • For users who need to compile the CPU version PaddlePaddle:

    cmake .. -DPY_VERSION=3.10 -DWITH_GPU=OFF
    
  • For users who need to compile the GPU version PaddlePaddle:

    cmake .. -DPY_VERSION=3.10 -DWITH_GPU=ON
    
  • For details on the compilation options, see the compilation options table.

  • Please attention to modify parameters -DPY_VERSION for the version of Python you want to compile with, for example -DPY_VERSION=3.10 means the version of python is 3.10

  • We currently do not support the compilation of the GPU version PaddlePaddle under CentOS 6.

10. Execute compilation:

Use multicore compilation

make -j$(nproc)

Note: During the compilation process, you need to download dependencies from github. Please make sure that your compilation environment can download the code from github normally.

11. After compiling successfully, go to the /paddle/build/python/dist directory and find the generated .whl package:

cd /paddle/build/python/dist

12. Install the compiled .whl package on the current machine or target machine:

For Python3:

pip3.10 install -U [whl package name]

Note: We used Python3.10 command as an example above, if the version of your Python is 3.8/3.9/3.11/3.12, please change pip3.10 in the commands to pip3.8/pip3.9/pip3.11/3.12.

Congratulations, now that you have successfully installed PaddlePaddle using Docker, you only need to run PaddlePaddle after entering the Docker container. For more Docker usage, please refer to the official Docker documentation.

Local compilation

1. Check that your computer and operating system meet the compilation standards we support:

uname -m && cat /etc/*release

2. Update the system source

  • CentOS system

    Update the source of yum: yum update, and add the necessary yum source:

    yum install -y epel-release
    
  • Ubuntu system

    Update the source of apt:

    apt update
    

3. Install NCCL (optional)

  • If you need to use multi card environment, please make sure that you have installed nccl2 correctly, or install nccl2 according to the following instructions (here is the installation instructions of nccl2 under CUDA11.2 and cuDNN8. For more version of installation information, please refer to NVIDIAofficial website):

    • CentOS system can refer to the following commands

      rm -f /usr/local/lib/libnccl.so
      wget --no-check-certificate -q https://nccl2-deb.cdn.bcebos.com/libnccl-2.10.3-1+cuda11.4.x86_64.rpm
      wget --no-check-certificate -q https://nccl2-deb.cdn.bcebos.com/libnccl-devel-2.10.3-1+cuda11.4.x86_64.rpm
      wget --no-check-certificate -q https://nccl2-deb.cdn.bcebos.com/libnccl-static-2.10.3-1+cuda11.4.x86_64.rpm
      rpm -ivh libnccl-2.10.3-1+cuda11.4.x86_64.rpm
      rpm -ivh libnccl-devel-2.10.3-1+cuda11.4.x86_64.rpm
      rpm -ivh libnccl-static-2.10.3-1+cuda11.4.x86_64.rpm
      

4. Install the necessary tools

bzip2 and make:

yum install -y bzip2 make

or

apt install -y bzip2 make

cmake requires version 3.18, we recommend that you use 3.18.0 version:

wget -q https://cmake.org/files/v3.18/cmake-3.18.0-Linux-x86_64.tar.gz
tar -zxvf cmake-3.18.0-Linux-x86_64.tar.gz
rm cmake-3.18.0-Linux-x86_64.tar.gz
PATH=/home/cmake-3.18.0-Linux-x86_64/bin:$PATH

gcc requires version 8.2:

wget -q --no-proxy https://paddle-ci.gz.bcebos.com/gcc-8.2.0.tar.xz && \
tar -xvf gcc-8.2.0.tar.xz && \
cd gcc-8.2.0 && \
sed -i 's#ftp://gcc.gnu.org/pub/gcc/infrastructure/#https://paddle-ci.gz.bcebos.com/#g' ./contrib/download_prerequisites && \
unset LIBRARY_PATH CPATH C_INCLUDE_PATH PKG_CONFIG_PATH CPLUS_INCLUDE_PATH INCLUDE && \
./contrib/download_prerequisites && \
cd .. && mkdir temp_gcc82 && cd temp_gcc82 && \
../gcc-8.2.0/configure --prefix=/usr/local/gcc-8.2 --enable-threads=posix --disable-checking --disable-multilib && \
make -j8 && make install

5. We support compiling and installing with virtualenv. First, create a virtual environment called paddle-venv with the following command:

  • a. Install Python-dev:

    (Please refer to the official Python installation process)

  • b. Install pip:

    (Please refer to the official Python installation process, and ensure that the pip3 version 20.2.2 and above, please note that in python3.8 and above, pip3 does not necessarily correspond to the python version, such as python3.10 default only Pip3.10)

  • c. (Only For Python3) set Python3 related environment variables, here is python3.10 version example, please replace with the version you use (3.8, 3.9, 3.11, 3.12):

    1. First find the path to the Python lib using

      find `dirname $(dirname $(which python3))` -name "libpython3.so"
      

      If it is 3.8/3.9/3.10/3.11/3.12, change python3 to python3.8, python3.9, python3.10, python3.11, python3.12, then replace [python-lib-path] in the following steps with the file path found.

    2. Set PYTHON_LIBRARIES:

      export PYTHON_LIBRARY=[python-lib-path]
      
    3. Secondly, use

      find `dirname $(dirname $(which python3))`/include -name "python3.10"
      

      to find the path to Python Include, please pay attention to the python version, then replace the following [python-include-path] to the file path found.

    4. Set PYTHON_INCLUDE_DIR:

      export PYTHON_INCLUDE_DIRS=[python-include-path]
      
    5. Set the system environment variable path:

      export PATH=[python-lib-path]:$PATH
      

      (here replace the last two levels content of [python-lib-path] with /bin/)

  • d. Install the virtual environment virtualenv and virtualenvwrapper and create a virtual environment called paddle-venv: (please note the pip3 commands corresponding to the python version, such as pip3.8, pip3.9, pip3.10, pip3.11, pip3.12)

    1. Install virtualenv:

      pip install virtualenv
      

      or

      pip3 install virtualenv
      
    2. Install virtualenvwrapper

      Pip install virtualenvwrapper
      

      or

      pip3 install virtualenvwrapper
      
    3. Find virtualenvwrapper.sh:

      find / -name virtualenvwrapper.sh
      

      (please find the corresponding Python version of virtualenvwrapper.sh)

    4. See the installation method in virtualenvwrapper.sh:

      cat vitualenvwrapper.sh
      

      this shell file describes the steps and commands

    5. Install virtualwrapper as described in virtualenvwrapper.sh

    6. Set VIRTUALENVWRAPPER_PYTHON:

      export VIRTUALENVWRAPPER_PYTHON=[python-lib-path]:$PATH
      

      (here replace the last two levels content of [python-lib-path] with /bin/)

    7. Create virtual environment named paddle-venv:

      mkvirtualenv paddle-venv
      

6. Enter the virtual environment:

workon paddle-venv

8. Put the PaddlePaddle source cloned in the Paddle folder in the current directory and go to the Paddle directory:

git clone https://github.com/PaddlePaddle/Paddle.git
cd Paddle

9. Switch to develop branch for compilation (Paddle supports Python version 3.8 and above):

git checkout develop

10. And please create and enter a directory called build:

mkdir build && cd build

11. Execute cmake:

For details on the compilation options, see the compilation options table.

  • For users who need to compile the CPU version PaddlePaddle:

    cmake .. -DPY_VERSION=3.10 -DPYTHON_INCLUDE_DIR=${PYTHON_INCLUDE_DIRS} \
    -DPYTHON_LIBRARY=${PYTHON_LIBRARY} -DWITH_GPU=OFF
    

    If you encounter Could NOT find PROTOBUF (missing: PROTOBUF_LIBRARY PROTOBUF_INCLUDE_DIR), you can re-execute the cmake command. Please note that the PY_VERSION parameter is replaced with the python version you need.

  • For users who need to compile the GPU version PaddlePaddle:

    1. Please make sure that you have installed nccl2 correctly, or install nccl2 according to the following instructions (here is ubuntu 20.04, CUDA11.2, cuDNN8 nccl2 installation instructions, for more information on the installation information please refer to the NVIDIA official website):

      rm -f /usr/local/lib/libnccl.so
      wget --no-check-certificate -q https://nccl2-deb.cdn.bcebos.com/libnccl-2.10.3-1+cuda11.4.x86_64.rpm
      wget --no-check-certificate -q https://nccl2-deb.cdn.bcebos.com/libnccl-devel-2.10.3-1+cuda11.4.x86_64.rpm
      wget --no-check-certificate -q https://nccl2-deb.cdn.bcebos.com/libnccl-static-2.10.3-1+cuda11.4.x86_64.rpm
      rpm -ivh libnccl-2.10.3-1+cuda11.4.x86_64.rpm
      rpm -ivh libnccl-devel-2.10.3-1+cuda11.4.x86_64.rpm
      rpm -ivh libnccl-static-2.10.3-1+cuda11.4.x86_64.rpm
      
    2. If you have already installed nccl2 correctly, you can start cmake: (For Python3: Please configure the correct python version for the PY_VERSION parameter)

      cmake .. -DPYTHON_EXECUTABLE:FILEPATH=[您可执行的 Python3 的路径] -DPYTHON_INCLUDE_DIR:PATH=[之前的 PYTHON_INCLUDE_DIRS] -DPYTHON_LIBRARY:FILEPATH=[之前的 PYTHON_LIBRARY] -DWITH_GPU=ON
      

Note: For the command involving Python 3, we use Python 3.10 as an example above, if the version of your Python is 3.8/3.9/3.11/3.12, please change Python3.10 in the commands to Python3.8/Python3.9/Python3.11/Python3.12

12. Compile with the following command:

make -j$(nproc)

Use multicore compilation

If “Too many open files” error is displayed during compilation, please use the instruction ulimit -n 8192 to increase the number of files allowed to be opened by the current process. Generally speaking, 8192 can ensure the completion of compilation.

13. After compiling successfully, go to the /paddle/build/python/dist directory and find the generated .whl package:

cd /paddle/build/python/dist

14. Install the compiled .whl package on the current machine or target machine:

Pip install -U (whl package name)

or

pip3 install -U (whl package name)

Congratulations, now you have completed the process of compiling PaddlePaddle natively.



Verify installation

After the installation is complete, you can use python or python3 to enter the Python interpreter and then use

import paddle

and then

paddle.utils.run_check()

to verify that the installation was successful.

If PaddlePaddle is installed successfully! appears, it means the installation was successful.



How to uninstall

Please use the following command to uninstall PaddlePaddle (users who use Docker to install PaddlePaddle should use the following command in the container containing PaddlePaddle. Please use the corresponding version of pip):

  • CPU version of PaddlePaddle:

    pip uninstall paddlepaddle
    

    or

    pip3 uninstall paddlepaddle
    
  • GPU version of PaddlePaddle:

    pip uninstall paddlepaddle-gpu
    

    or

    pip3 uninstall paddlepaddle-gpu
    

Users installing PaddlePaddle with Docker, please use above commands in the container involved PaddlePaddle and attention to use the corresponding version of Pip