zoukankan      html  css  js  c++  java
  • Setting up caffe on Ubuntu

    Setting up caffe on ubuntu1404 LTS +CUDA7.5 and ubuntu1604 LTS +CUDA8.0

    @(sinbad)[360sinbad@gmail.com]

    Prepare nVidia driver and CUDA Toolkit

    • Download cuda toolkit 7.5( support ubuntu 1404) or 8.0(support ubuntu 1604) on nvidia web. Now the download page is shown the latest version of cuda toolkit, you should find CUDA Toolkit Archive folder or else you only see current latest version, and choose the compatable version with your system link like as below:
      https://developer.nvidia.com/cuda-75-downloads-archive

      Now the latest version of CUDA is CUDA 9.0, there are some errors after installing CUDA 9.0 with ubuntu1604 LTS. So i suggest CUDA 8.0 for your choose.Download resource please search Baidu cloud.

    • Download GPU Graphics driver on web: http://www.geforce.cn/drivers/results/105888#, choose version depend on your graphics card.

    Install CUDA driver and config system env

    Step 1. nVidia Graphics Driver and CUDA Toolkit
    Forbid nouveau.
    sudo gedit /etc/modprobe.d/disable-nouveau.conf
    Appending two lines as below.
    blacklist nouveau
    options nouveau modeset=0
    then save and exit: (:wq)
    Note: excute lspci | grep nouveau check if nouveau was forbid. If there is nothing to show, you can ignore step above.
    Enter TTY Ctrl + Alt +F1 and close GUI
    sudo stop lightdm
    sudo chmod -x NVIDIA_XXX.run
    sudo ./NVIDIA_XXX.run

    CUDA Toolkit install
    Installation Instructions:
    1. sudo dpkg -i cuda-repo-ubuntu1404-7-5-local_7.5-18_amd64.deb
    2. sudo apt-get update
    3. sudo apt-get install cuda

    **Step 2.Set up system env for CUDA **
    sudo gedit /etc/profile
    Appending two lines as below.
    export PATH=/usr/local/cuda/bin:$PATH
    export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PAT

    After saving, create a linkfile through
    sudo gedit /etc/ld.so.conf.d/cuda.conf
    Add cuda path in the end.
    /usr/local/cuda/lib64
    Save and reboot the computer.

    Download Caffe

    After such preparation, now let us start installing caffe.
    Download Caffe from ithub.
    git clone https://github.com/BVLC/caffe.git
    Install several dependencies for caffe
    sudo apt-get install build-essential
    sudo apt-get install cmake
    sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libboost-all-dev libhdf5-serial-dev libgflags-dev libgoogle-glog-dev liblmdb-dev protobuf-compiler

    Install Python

    The main requirements are bumpy and boost.python (provided by boost).
    You can install the dependencies with
    cd caffe/python/
    for req in $(cat requirements.txt); do pip install $req; done
    it may meet permission problem.so, try the other command:
    cd ~/caffe
    sudo pip install -r ./python/requirements.txt
    Maybe you install pip first.
    sudo apt-get install python-pip
    Note: You can use #sudo apt-get install python-numpy python-skimage python-matplotlib if above step failed.

    To import the caffe Python module after completing the installation, add the module directory to your $PYTHONPATH by
    export PYTHONPATH=/path/to/caffe/python:$PYTHONPATH

    Install Matlab 2015b

    • Download Matlab 2015b for Linux
      Link: http://pan.baidu.com/s/1i5BO0eh password: hraf
    • Mount ISO file
      sudo mkdir /MATLAB
      sudo mount -o loop R2015b_glnxa64.so /MATLAB
    • Excute instal without connect Internet. Note: Choose install standalone
      cd /MATLAB
      sudo ./inatall
    • Copy /crack/bin to usr/local/MATLAB/R2015b/bin after installing matlab
      sudo cp /[Your crack directory]/R2015b/bin/glnxa64/* /usr/local/MATLAB/R2015b/bin/glnxa64
    • Run matlab
      cd /usr/local/MATLAB/R2015b/bin
      sudo ./matlab
    • Umount ISO
      sudo umount /matlab

    Build Caffe follow the caffe installation instructions here:

  • 相关阅读:
    c/c++ linux 进程间通信系列7,使用pthread mutex
    c/c++ linux 进程间通信系列6,使用消息队列(message queue)
    c/c++ linux 进程间通信系列5,使用信号量
    eos 创建两对的公钥和私钥, 钱包,交易所转账到主网,主网到交易所
    c/c++ linux 进程间通信系列4,使用共享内存
    python基础-网络编程part02
    idea新建项目相关名词意义
    idea中当前模块资源目录上显示属于其他模块
    centos下安装rabbitmq
    JAVA中值传递,引用传递
  • 原文地址:https://www.cnblogs.com/sinbad360/p/6445837.html
Copyright © 2011-2022 走看看