前言

Caffe的安裝大體分為5個步驟：參考地址

• 安裝 CUDA 和 cuDNN
• 安裝軟體 Python 和 Matlab
• 安裝依賴
• 編譯 Caffe，pycaffe 和 matcaffe
• 路徑設定

安裝 CUDA 和 cuDNN

安裝 CUDA 參考地址 下載地址

• Verify You Have a CUDA-Capable GPU: lspci | grep -i nvidia
• Verify You Have a Supported Version of Linux: uname -m && cat /etc/*release
• Verify the System Has gcc Installed: gcc --version
• Verify the System has the Correct Kernel Headers and Development Packages Installed: uname -r,sudo apt-get install linux-headers-$(uname -r)
• Create a file at /etc/modprobe.d/blacklist-nouveau.conf with the following contents:
  blacklist nouveau
options nouveau modeset=0
• Regenerate the kernel initramfs: sudo update-initramfs -u
• Reboot into text mode (runlevel 3): systemctl set-default runlevel3.target
• 重啟reboot
• 確認nouveau被遮蔽：lsmod | grep nouveau，無輸出則被遮蔽
• 安裝：sh cuda_8.0.61_375.26_linux.run，examples路徑設定為：/usr/local/cuda-8.0
• 進入影象介面 systemctl set-default runlevel5.target
• 重啟：reboot
• 檢視顯示卡：cat /proc/driver/nvidia/version
• 編譯示例：cd /usr/local/cuda-8.0/NVIDIA_CUDA-8.0_Samples/
  
  make
• 新增PATH路徑：vim /etc/profile
#CUDA
export PATH=/usr/local/cuda-8.0/bin${PATH:+:${PATH}}
. /etc/profile
• 新增動態連結庫:vim /etc/ld.so.conf
#CUDA
/usr/local/cuda-8.0/lib64
• 編譯例子 cd /usr/local/cuda-8.0/NVIDIA_CUDA-8.0_Samples
make
• 檢驗安裝結果：cd 1_Utilities/deviceQuery
./deviceQuery
  若出現下列結果，則安裝成功

./deviceQuery Starting...
CUDA Device Query (Runtime API) version (CUDART static linking)
Detected 1 CUDA Capable device(s)
Device 0: "GeForce GT 740"
  CUDA Driver Version / Runtime Version          8.0 / 8.0
  CUDA Capability Major/Minor version number:    3.0
  Total amount of global memory:                 974 MBytes (1021706240 bytes)
  ( 2) Multiprocessors, (192) CUDA Cores/MP:     384 CUDA Cores
  GPU Max Clock rate:                            993 MHz (0.99 GHz)
  Memory Clock rate:                             2500 Mhz
  Memory Bus Width:                              128-bit
  L2 Cache Size:                                 262144 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(65536), 2D=(65536, 65536), 3D=(4096, 4096, 4096)
  Maximum Layered 1D Texture Size, (num) layers  1D=(16384), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(16384, 16384), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  2048
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 1 copy engine(s)
  Run time limit on kernels:                     Yes
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 1 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >
deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 8.0, CUDA Runtime Version = 8.0, NumDevs = 1, Device0 = GeForce GT 740
Result = PASS

安裝 cuDNN

下載地址

• 解壓下載檔案
  tar -zxvf cudnn-8.0-linux-x64-v5.1.tgz
• 拷貝檔案到相應路徑
  sudo cp cuda/include/cudnn.h /usr/local/cuda-8.0/include
sudo cp cuda/lib64/libcudnn* /usr/local/cuda-8.0/lib64

安裝軟體 Python 和 Matlab

安裝 Python

Ubuntu 16.04 自帶Python2和Python3想用哪個版本都可以，這裡就不介紹Python的安裝了

安裝Matlab

• 拷貝安裝檔案到硬碟
• 修改安裝檔案許可權chmod -R +777 R2015b_glnxa64
• 執行安裝程式cd R2015_glnxa64 sudo ./install
• 選擇不聯網安裝，序列號為09806-07443-53955-64350-21751-41297，若安裝Matlab Production Server，使用40236-45817-26714-51426-39281
• 拷貝破解檔案sudo cp R2015b/bin/glnxa64/* /usr/local/MATLAB/R2015b/bin/glnxa64/
• 啟用cd /usr/local/MATLAB/R2015b/bin，sudo ./matlab，選擇standalone進行啟用
• 建立快捷方式, sudo vim /usr/share/applications/matlab.desktop輸入下列內容：

[Desktop Entry]
Encoding=UTF-8
Name=Matlab R2015b
Exec=/usr/local/MATLAB/R2015b/bin/matlab -desktop
Icon=/usr/local/MATLAB/R2015b/toolbox/shared/dastudio/resources/MatlabIcon.png
Terminal=false
Type=Application
Categories=Development;IDE;
Name[zh_CN]=Matlab R2015b

• 注意剛安裝完執行時可能遇到兩個問題：
  1.啟動時終端顯示matlab is selecting software opengl rendering，執行imshow() 會提示錯誤，原因是沒安裝顯示卡驅動，所以要最好先安裝CUDA，再安裝Matlab
  2.用快捷方式啟動Matlab時，畫面卡住不動，而且在終端不用root許可權執行matlab時會出錯，解決辦法：
  sudo chmod a+w -R /home/sibo/.matlab

安裝依賴

General dependencie

sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler
sudo apt-get install --no-install-recommends libboost-all-dev

BLAS

sudo apt-get install libatlas-base-dev

Python headers

apt-get install python3-dev

編譯 Caffe，pycaffe 和 matcaffe

編譯Caffe

• 解壓檔案
  cd /usr/local
unzip /media/sibo/UUI/軟體/caffe-master.zip
• 修改Python依賴包
  cd caffe-master/python
vim requirements.txt
  內容修改為
Cython
numpy==1.11.0
scipy
scikit-image
matplotlib
ipython
h5py
leveldb
networkx
nose
pandas
python-dateutil
protobuf
python-gflags
pyyaml
Pillow
six

• 安裝Python包
  sudo apt-get install python3-pip
pip3 install --upgrade pip
for i in $(cat requirements.txt); do sudo pip3 install $i; done
• 複製配置檔案cp Makefile.config.example Makefile.config
• 修改配置檔案

USE_CUDNN := 1
CUDA_DIR := /usr/local/cuda-8.0
BLAS := atlas
MATLAB_DIR := /usr/local/MATLAB/R2015b
PYTHON_LIBRARIES := boost_python3 python3.5m
PYTHON_INCLUDE := /usr/include/python3.5m \
                 /usr/local/lib/python3.5/dist-packages/numpy/core/include
PYTHON_LIB := /usr/lib/python3.5/config-3.5m-x86_64-linux-gnu
WITH_PYTHON_LAYER := 1
INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include
LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib
BUILD_DIR := build
DISTRIBUTE_DIR := distribute
TEST_GPUID := 0

• 編譯make all
  出現錯誤：fatal error: gflags/gflags.h: 沒有那個檔案或目錄
  解決辦法：apt-get install libgflags-dev
出現錯誤：fatal error: glog/logging.h: 沒有那個檔案或目錄
  解決辦法：apt-get install libgoogle-glog-dev
  出現錯誤：fatal error: hdf5.h: 沒有那個檔案或目錄
  解決辦法：vim Makefile.config 新增
  INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include \
/usr/include/hdf5/serial
  出現錯誤：fatal error: lmdb.h: 沒有那個檔案或目錄
  解決辦法：apt-get install liblmdb-dev
  出現錯誤：/usr/bin/ld: 找不到 -lhdf5_hl
/usr/bin/ld: 找不到 -lhdf5
/usr/bin/ld: 找不到 -lboost_python3
  解決辦法：vim Makefile.config 新增
  LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib \
/usr/lib/x86_64-linux-gnu/hdf5/serial \
/usr/lib/x86_64-linux-gnu
cd /usr/lib/x86_64-linux-gnu/
ln -s libboost_python-py35.so libboost_python3.so
• make test, make runtest
• make pycaffe

• 修改python測試命令
  vim /usr/local/caffe-master/Makefile

  pytest: py
  cd python; python -m unittest discover -s caffe/test
  修改為
  pytest: py
  cd python; python3 -m unittest discover -s caffe/test
  
  make pytest

• make matcaffe
make mattest
  出現錯誤：Invalid MEX-file '/usr/local/caffe-master/matlab/+caffe/private/caffe_.mexa64':
/usr/local/MATLAB/R2015b/bin/glnxa64/../../sys/os/glnxa64/libstdc++.so.6: versionGLIBCXX_3.4.20’
  not found `
  解決辦法：

cd /usr/local/MATLAB/R2015b/sys/os/glnxa64/
mv libstdc++.so.6 libstdc++.so.6.bak
ln -s /usr/lib/x86_64-linux-gnu/libstdc++.so.6 libstdc++.so.6

出現錯誤：caffe_.mexa64: undefined symbol:
_ZN2cv8imencodeERKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEERKNS_11_InputArrayERSt6vectorIhSaIhEERKSB_IiSaIiEE
原因是caffe在編譯的時候用的是系統的opencv庫，而Matlab軟體用的是自己的opencv庫，所以有衝突
解決辦法:參考地址

root@sibo-All-Series:/usr/local/caffe-master# cd /usr/local/MATLAB/R2015b/bin/glnxa64/
root@sibo-All-Series:/usr/local/MATLAB/R2015b/bin/glnxa64# mv libopencv_imgproc.so.2.4 libopencv_imgproc.so.2.4.bak
root@sibo-All-Series:/usr/local/MATLAB/R2015b/bin/glnxa64# mv libopencv_highgui.so.2.4 libopencv_highgui.so.2.4.bak
root@sibo-All-Series:/usr/local/MATLAB/R2015b/bin/glnxa64# mv libopencv_core.so.2.4 libopencv_core.so.2.4.bak
root@sibo-All-Series:/usr/local/MATLAB/R2015b/bin/glnxa64# ln -s /usr/lib/x86_64-linux-gnu/libopencv_core.so.2.4.9 libopencv_core.so.2.4
root@sibo-All-Series:/usr/local/MATLAB/R2015b/bin/glnxa64# ln -s /usr/lib/x86_64-linux-gnu/libopencv_highgui.so.2.4.9 libopencv_highgui.so.2.4
root@sibo-All-Series:/usr/local/MATLAB/R2015b/bin/glnxa64# ln -s /usr/lib/x86_64-linux-gnu/libopencv_imgproc.so.2.4.9 libopencv_imgproc.so.2.4

設定路徑

• 對於Python
  
vim /home/sibo/.bashrc
# caffe
export PYTHONPATH=/usr/local/caffe-master/python${PYTHONPATH:+:${PYTHONPATH}}
. /home/sibo/.bashrc

• 對於Matlab
  編寫程式前將新增路徑
  addpath /usr/local/caffe-master/matlab