本文共 2799 字，大约阅读时间需要 9 分钟。

fork from : https://astrail.com/guide/

Install Ubuntu 14.04

1. partition:
   1. swap: ~16 GiB
   2. /: ~100 GiB
   3. /home: the rest
2. grub: use default setting
3. Attention, if you cannot enter tty1-6:
   1. sudo gedit /etc/default/grub
   2. set GRUB_CMDLINE_LINUX_DEFAULT = nomodeset
   3. sudo update-grub
   4. sudo reboot

Install Nvidia driver 3xx.xx

1. Ctrl+Alt+F1 (enter into tty1)
2. sudo service lightdm stop (stop x-window service)
3. sudo bash ./NVIDIA-Linux-x86_64-3xx.xx.run (latest version)
4. sudo service lightdm start

Install CUDA 8.0

1. sudo bash ./cuda_8.0.xx_linux-run.run (use .run version instead of .deb )
   1. DO NOT Install NVIDIA Accelerated Graphics Driver (n)
   2. choose y for other options
2. sudo apt-get update
3. sudo gedit /etc/ld.so.conf.d/cuda.conf (add)
   1. /usr/local/cuda/lib64
   2. /lib
4. sudo ldconfig -v

Install cuDNN v5.x

1. tar xzvf cudnn-8.0-linux-x64-v5.x.tgz
2. sudo cp cuda/include/cudnn.h /usr/local/cuda/include
3. sudo cp cuda/lib64/libcudnn.* /usr/local/cuda/lib64
4. sudo ln -sf /usr/local/cuda/lib64/libcudnn.so.5.x.x /usr/local/cuda/lib64/libcudnn.so.5
5. sudo ln -sf /usr/local/cuda/lib64/libcudnn.so.5 /usr/local/cuda/lib64/libcudnn.so
6. sudo ldconfig -v

Install dependence

1. sudo apt-get install -y build-essential git
2. sudo apt-get install -y libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libboost-all-dev libhdf5-serial-dev libgflags-dev libgoogle-glog-dev liblmdb-dev protobuf-compiler protobuf-c-compiler libyaml-dev
3. sudo apt-get install -y python-pip python-dev python-opencv python-numpy python-scipy python-matplotlib python-sklearn python-skimage python-h5py python-protobuf python-leveldb python-networkx python-nose python-pandas python-gflags cython ipython
4. sudo pip install --upgrade pip
5. sudo pip install jupyter (optional)

Install BLAS

1. choose ATLAS，MKL or OpenBLAS
2. ATLAS:
   1. sudo apt-get install -y libatlas-base-dev
3. MKL (optional):
   1. sudo bash install_GUI.sh
   2. sudo gedit /etc/ld.so.conf.d/intel_mkl.conf (add)
      1. /opt/intel/lib/intel64
      2. /opt/intel/mkl/lib/intel64
   3. sudo ldconfig -v

Install glog 0.3.3

1. wget
2. tar xzvf v0.3.3.tar.gz
3. cd glog-0.3.3
4. ./configure && make && make install

Install caffe

1. git clone
2. cd caffe
3. cp Makefile.config.example Makefile.config
4. gedit Makefile.config
   1. USE_CUDNN := 1 (del “#”)
   2. BLAS := atlas (or mkl, open)
   3. WITH_PYTHON_LAYER := 1 (del “#” to build with python support)
   4. MATLAB_DIR := /usr/local/MATLAB/R20xxx/ (del “#” to build with MATLAB support)
   5. INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/lib/x86_64-linux-gnu/hdf5/serial/include
   6. LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib /usr/lib/x86_64-linux-gnu/hdf5/serial
5. make all -j8
6. make test -j8 (optional)
7. make runtest -j8 (optional)
8. make pycaffe (optional)
9. make matcaffe (optional)

Issues:

1. Error ‘fatal error: pyconfig.h: No such file or directory’
   1. export CPLUS_INCLUDE_PATH=/usr/include/python2.7
2. Login loop, cannot enter desktop

转载地址：https://blog.csdn.net/xiaojiajia007/article/details/71514624 如侵犯您的版权，请留言回复原文章的地址，我们会给您删除此文章，给您带来不便请您谅解！