Skip to content
This repository has been archived by the owner on Jan 7, 2020. It is now read-only.

Latest commit

 

History

History
229 lines (174 loc) · 6.52 KB

INSTALL_en.md

File metadata and controls

229 lines (174 loc) · 6.52 KB
Raw

Building Anakin from Source

We've built and tested Anakin on CentOS 7.3 and Ubuntu 16.04. The other operating systems installation are coming soon.

Installation overview

Installing on CentOS

1. System requirements

  • gcc 4.8.5
  • g++ 4.8.5
  • make 3.82+
  • cmake 2.8.12+
  • protobuf 3.4.0
  • wget

1.1 install wget

$ sudo yum -y install wget

1.2 install protobuf

$ wget --no-check-certificate https://mirror.sobukus.de/files/src/protobuf/protobuf-cpp-3.4.0.tar.gz
$ tar -xvf protobuf-cpp-3.4.0.tar.gz
$ cd protobuf-3.4.0
$ ./configure
$ make -j
$ make install

Check

$ protoc --version

Any problems for protobuf installation, Please see here

2. Building Anakin for CPU-only

Please make sure that all pre-requirements have been installed on your system.

2.1 Clone Anakin and build

$ git clone https://github.com/PaddlePaddle/Anakin.git  
$ cd Anakin 
$ git checkout developing  

# tools directory contains multi-platform build scripts.

# 1. using script to build.
$ ./tools/x86_build.sh

# 2. or you can build directly. This way you can build x86 and gpu  
# at the same time by set corresponding variables(USE_GPU_PLACE, 
# USE_X86_PLACE, USE_ARM_PLACE) in CMakeList.txt which is at Anakin 
# source root directory. In this situation, set USE_X86_PLACE with YES, 
# others with NO. And then type the following command:

# $ mkdir build  
# $ cd build  
# $ cmake ..  
# $ make

3. Building Anakin with NVIDIA GPU Support

3.1. Install dependences
3.1.1 CUDA Tookit

#set CUDA path(cuda 9.0 for example)
export PATH=/usr/local/cuda-9.0/bin:$PATH #the path where your CUDA installed.
export LD_LIBRARY_PATH=/usr/local/cuda-9.0/lib64:$LD_LIBRARY_PATH

# CUDNN
$ export CUDNN_ROOT=/usr/local/cudnn_v7 # the path where your cuDNN installed.
$ export LD_LIBRARY_PATH=${CUDNN_ROOT}/lib64:$LD_LIBRARY_PATH
$ export CPLUS_INCLUDE_PATH=${CUDNN_ROOT}/include:$CPLUS_INCLUDE_PATH

Don't forget to source your bashrc

3.2. Clone Anakin and build

$ git clone https://github.com/PaddlePaddle/Anakin.git  
$ cd Anakin 
$ git checkout developing 

# tools directory contains multi-platform build scripts.

# 1. using script to build.
$ ./tools/x86_build.sh

# 2. or you can build directly. This way you can build x86 and gpu  
# at the same time by set corresponding variables(USE_GPU_PLACE, 
# USE_X86_PLACE, USE_ARM_PLACE) in CMakeList.txt which is at Anakin 
# source root directory. In this situation, set USE_X86_PLACE with YES, 
# others with NO. And then type the following command:
# $ mkdir build  
# $ cd build  
# $ cmake ..  
# $ make

4. Building Anakin with AMD GPU Support

Coming soon..

Installing on Ubuntu

1. System requirements

  • gcc 4.8.5/5.4.0
  • g++ 4.8.5/5.4.0
  • make 3.82+
  • cmake 2.8.12+
  • protobuf 3.4.0
  • wget

Note: You need to use the gcc/g++ we recommend to build protobuf and Anakin, otherwise it might result some errors.

1.1 install wget

$ sudo apt-get install wget

1.2 install protobuf

$ wget --no-check-certificate https://mirror.sobukus.de/files/src/protobuf/protobuf-cpp-3.4.0.tar.gz
$ tar -xvf protobuf-cpp-3.4.0.tar.gz
$ cd protobuf-3.4.0
$ ./configure
$ make -j
$ make install

Check

$ protoc --version

Any problems for protobuf installation, Please see here

2. Building Anakin for CPU-only

Please make sure that all pre-requirements have been installed on your system.

2.1 Clone Anakin and build

$ git clone https://github.com/PaddlePaddle/Anakin.git  
$ cd Anakin 
$ git checkout developing  
# tools directory contains multi-platform build scripts.
# 1. using script to build.
$ ./tools/x86_build.sh
# 2. or you can build directly. This way you can build x86 and gpu  
# at the same time by set corresponding variables(USE_GPU_PLACE, 
# USE_X86_PLACE, USE_ARM_PLACE) in CMakeList.txt. In this situation, 
# set USE_X86_PLACE with YES, others with NO.
$ mkdir build  
$ cd build  
$ cmake ..  
$ make

3. Building Anakin with NVIDIA GPU Support

3.1. Install dependences
3.1.1 CUDA Tookit

#set CUDA path(cuda 9.0 for example)
export PATH=/usr/local/cuda-9.0/bin:$PATH #the path where your CUDA installed.
export LD_LIBRARY_PATH=/usr/local/cuda-9.0/lib64:$LD_LIBRARY_PATH

# CUDNN
$ export CUDNN_ROOT=/usr/local/cudnn_v7 # the path where your cuDNN installed.
$ export LD_LIBRARY_PATH=${CUDNN_ROOT}/lib64:$LD_LIBRARY_PATH
$ export CPLUS_INCLUDE_PATH=${CUDNN_ROOT}/include:$CPLUS_INCLUDE_PATH

Don't forget to source your bashrc

3.2. Clone Anakin and build

$ git clone https://github.com/PaddlePaddle/Anakin.git  
$ cd Anakin 
$ git checkout developing  
# tools directory contains multi-platform build scripts.
# 1. using script to build.
$ ./tools/gpu_build.sh
# 2. or you can build directly. This way you can build x86 and gpu  
# at the same time by set corresponding variables(USE_GPU_PLACE, 
# USE_X86_PLACE, USE_ARM_PLACE) in CMakeList.txt. In this situation, 
# set USE_GPU_PLACE with YES, others with NO.
$ mkdir build  
$ cd build  
$ cmake ..  
$ make

4. Building Anakin with AMD GPU Support

Coming soon..

Installing on ARM

Please refer to run on arm

Verifying installation

If build successfully, the libs will be in the directory output/, and you can run unit test in output/unit_test to verify your installation.