当我重装ubuntu时我该怎么做

Posted on | In

引言

这篇post不做过多介绍，主要方便我个人记录参考使用。

具体实施步骤

# Make Sure the NVIDIA-Linux-x86_64-390.25.run, cuda-8.0.run, and cudnn-8.0-linux-x64-v5.1.tgz has been in the /home/yann
    1  sudo gedit /etc/modprobe.d/blacklist.conf 
#  blacklist nouveau  
#  options nouveau modeset=0
# above two lines, add to files' ending
# then restart the system
    2  lsmod | grep nouveau
    3  sudo service lightdm stop
    4  sudo chmod 777 NVIDIA-Linux-x86_64-390.25.run 
    5  sudo ./NVIDIA-Linux-x86_64-390.25.run 
    6  sudo service lightdm start
## then set your system setting and make sure you have been on Internet
    7  sudo apt-get update
    8  sudo apt-get install git 
    9  sudo apt-get install python-pip
   10  sudo apt-get install python-pip python-dev build-essential
   11  sudo apt-get install python-numpy python-scipy python-matplotlib
   12  sudo apt-get install vim
   13  sudo apt-get install ipython
   14  sudo apt-get install firefox
   15  sudo pip install -U pip
   17  pip -V
   19  ibus-daemon -drx # fix the input pinyin
   20  cat /proc/driver/nvidia/version # check the nvidia driver
   21  ls
   22  sudo sh cuda-8.0.run
   23  cd /usr/local/cuda-8.0/samples/1_Utilities/deviceQuery/
   24  ls
   25  sudo make
   26  ./deviceQuery 
# if you have see the sentence ' result = PASS ' at last line, then OK!
   27  sudo gedit /etc/profile
# enter this file and add——
# 'export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64'
# 'export PATH=$PATH:/usr/local/cuda-8.0/bin'
   28  cd
   29  source /etc/profile
   30  echo $PATH
   31  nvcc -V
   32  ls
   33  tar -zxvf cudnn-7.5-linux-x64-v5.0-ga.tgz 
   34  sudo cp cuda/include/cudnn.h /usr/local/cuda/include/
   35  sudo cp cuda/lib64/* //usr/local/cuda/lib64/
   36  cd /usr/local/cuda/lib64/
   37  ls
   38  sudo chmod a+r libcudnn*
   40  sudo rm -r libcudnn.so libcudnn.so.5
   41  sudo ln -s libcudnn.so.5.1.10 libcudnn.so.5
   42  sudo ln -s libcudnn.so.5 libcudnn.so
   43  sudo ldconfig # check compile
   44  cd 
   45  nvidia-smi
   46  sudo pip install tensorflow-gpu==1.2.1
   47  python
   48  history>when-i-just-install-the-system.sh # import your history command for next configuation

驱动及库链接

以下链接可以直接通过wget的方式得到。另外，即便不是cuda8.0，cuda9.0也不错的，但是要注意cudnn和cuda要对应！

附录

送一点福利：

wallet：{"secret":false,"public":"04fdd2eeb9cc15d255e6f4bbe2270d38b6a810eb10aba4ab6149d3dc004e4b8a9107a1267f5025d0ccd43ac66ae470bdc26371d75ef2436eb1248d5967f0fc55c4","private":"d41c9485296b952b0ac48b6e5aeae65a5c6370984100218c9845ad33580a4bfc","address":"Qr5KXpYJUJUCevnxLZG4GMVuVrjTnCwrW2pkUGAvNc7y"}