Ubuntu14.04+Cuda8.0+Opencv2.14.3_caffe.txt

### Cuda 安装 Note
系统：Ubuntu 14.04.5 x64
Cuda: 8.0.44
显卡：nvidia-GeForce 1070

## 目的:因为需要重编译FlowNet(caffe)，DispNet,需要Cuda环境
主要参考</br>
[http://weibo.com/p/2304189db078090102vdvx]</br>
[https://yangcha.github.io/GTX-1080/    ]</br>
[www.cnblogs.com/platero/p/3993877.html]</br>

##Steps:
1. 重装系统 ubuntu 14.04.5
1.1  ubuntu下 vi输入方向键会变成ABCD，这是ubuntu预装的是vim tiny版本，安装vim full版本即可解决。
     先卸载vim-tiny：
     $ sudo apt-get remove vim-common
     再安装vim full：
     $ sudo apt-get install vim
     #modified by Lingfeng

2. sudo apt-get install build-essential
   sudo add-apt-repository ppa:graphics-drivers/ppa
   sudo apt-get update
  sudo apt-get install nvidia-367 [手动下载驱动，此处可能下载较慢] (icecubic 机器装的是nvidia-375)
   sudo apt-get install mesa-common-dev
   sudo apt-get install freeglut3-dev
  sudo sh cuda_8.0.44_linux.run [其中，Install NVIDIA Accelerated Graphics Driver... 选择n,前面已经安装了驱动且可行]

3. 配置Cuda环境变量
3.1  1. $vi ~/.bashrc 按G 跳到最后一行，添加入下面这行：
        export PATH=/usr/local/cuda-8.0/bin:$PATH
        source 生效：
        $source /etc/profile
     2. $vi /etc/ld.so.conf
        往文件里增加一行： /usr/local/cuda-8.0/lib64
        $source ldconfig
     #mmodified by Lingfeng

4. 安装库：sudo apt-get install freeglut3-dev libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev
4.1  上述安装会报错：详见: [http://github.com/linrio/git_R/issues/1]</br>
     要先卸载2个库:
     $sudo apt-get autoremove libcheese-gtk23 libcheese7
     再:
     $sudo apt-get install freeglut3-dev libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev
   #modified by Lingfeng


5.  make Cuda的example，此后nvcc, nvidia-smi, ./deviceQuery均已有效
5.1  跑example的方法:
     cd /usr/local/cuda-6.5/samples
     sudo make
     #modified by Lingfeng
5.2  跑example 报错：
     cuda usr/bin/ld:cannot find -lnvcuvid
     collect2: error: ld returned 1 exit status
     Makefile:381: recipe for target 'cudaDecodeGL' failed
     解决方法：https://github.com/linrio/mycaffe/issues/6


6. 安装python: sudo aot-get install python-dev python-pip
7. 安装opencv 2.4.13 在opencv的github master tag 选不同的版本。
   cd ~/opencv
   mkdir build
   cd build
   cmake -D CMAKE_BUILD_TYPE=RELEASE \
   -D CMAKE_INSTALL_PREFIX=/usr/local \
   -D INSTALL_C_EXAMPLES=ON \
   -D INSTALL_PYTHON_EXAMPLES=ON \
   -D BUILD_EXAMPLES=ON \
   -D WITH_CUDA=ON ..
   
   make -j 8
   sudo make install


8. 安装boost,配置环境

9. 安装blas,atlas,配置环境
9.1 安装 libatlas-base-dev
    sudo apt-get install libatlas-base-dev

10. copy cudnn include, lib内容

11.1 操作：
     下载cudnn.
     tar -zxvf cudnn-8.0-linux-x64-v5.1.tgz
     sudo cp cuda/include/cudnn.h /usr/local/cuda/include
     sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
     sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn*

12 安装anaconda2  [默认环境配置和默认安装位置]

13 编译caffe
13.1 修改Makefile.config
     CUDNN = 1 去掉注释#
     去掉anaconda的三行注释#
     把${home}/anaconda 改成 ${home}/anaconda2

14  ubuntu 图形界面进不去
    重新安装nvidia驱动：
    sudo service lightdm stop
    sudo apt-get remove --purge nvidia-*
    sudo apt-get install ubuntu-desktop
    sudo rm /etc/X11/xorg.conf
    echo 'nouveau' | sudo tee -a /etc/modules
    cd进驱动所在文件夹
    sudo sh NVIDIA-Linux-x86_64-346.59.run(之后要自己选择一下选项)
    sudo reboot




--------
[错误版]

1. 重装系统

2. 安装Anaconda2,主要为了安装python2.x，似乎会附带很多其他有用的包，就直接安装anaconda2了[!!!不能提前安装Anaconda2，可能导致后面显卡驱动问题，登陆无法进入界面]

3. 安装SSH-SERVER,安装Cuda需要退出GUI界面，因此用另一台机器用SSH-SERVER控制本机,方便操作

4. 安装步骤参考上述链接，不过实际过程中没出现博客中的toolkit等安装失败问题

5. 安装build-essential

6. 设置环境变量

---其中会出现的问题---
a. 执行nvcc -V查看版本的时候会提示nvcc没有安装，其实就是之前装的NVIDIA-CUDA-Toolkit的编译器没有安装完整，根据提示安装就好：
sudo apt-get install nvidia-cuda-toolkit

b. 执行examples中的sudo ./deviceQuery的时候，会显示cuda driver version isinsufficient for cuda runtime，因为当前驱动版本比刚安装的cuda版本低，需要重装驱动： sudo apt-get purge nvidia*
       sudo apt-get install nvidia-current
       sudoreboot

c. 执行example的sudo make 时，报错说找不到 -lnvcuvid

e. 重装显卡驱动：

第一步、Ctrl+Alt+F1进终端，log in之后，输入watch nvidia-smi，查看英伟达的显卡驱动还在不在。如果有显卡信息显示，转到第三步，如果没有转第二步。

第二步、安装Nvidia显卡驱动，按部就班：

sudo service lightdm stop
sudo apt-get remove --purge nvidia-*
sudo apt-get install ubuntu-desktop
sudo rm /etc/X11/xorg.conf
echo 'nouveau' | sudo tee -a /etc/modules
cd进驱动所在文件夹
sudo sh NVIDIA-Linux-x86_64-346.59.run(之后要自己选择一下选项)
sudo reboot

第三步：如果，桌面还是不能显示，则重装桌面。

sudo apt-get install --reinstall ubuntu-desktop
sudo apt-get install unity
sudo apt-get purge nvidia* bumblebee*
sudo apt-get install nvidia-prime
sudo shutdown -r now

DISPLAY=:0 unity
dconf reset -f /org/compiz/
unity
setsid unity

到这里基本都能解决问题。

**************************************************************************************
                        一些注意事项/报错及解决方法：
**************************************************************************************

4.
icecubic@icecubic:~/icecubic$ sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev
Reading package lists... Done
Building dependency tree
Reading state information... Done
build-essential is already the newest version.
freeglut3-dev is already the newest version.
libglu1-mesa is already the newest version.
libglu1-mesa set to manually installed.
libglu1-mesa-dev is already the newest version.
libglu1-mesa-dev set to manually installed.
libx11-dev is already the newest version.
libx11-dev set to manually installed.
Some packages could not be installed. This may mean that you have
requested an impossible situation or if you are using the unstable
distribution that some required packages have not yet been created
or been moved out of Incoming.
The following information may help to resolve the situation:

The following packages have unmet dependencies:
 unity-control-center : Depends: libcheese-gtk23 (>= 3.4.0) but it is not going to be installed
                        Depends: libcheese7 (>= 3.0.1) but it is not going to be installed
E: Error, pkgProblemResolver::Resolve generated breaks, this may be caused by held packages.

Fixed method:

$ sudo apt-get autoremove libcheese-gtk23 libcheese7
$ sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev

**************************************************************************************************************************************

5.
```
cuda usr/bin/ld:cannot find -lnvcuvid
collect2: error: ld returned 1 exit status
Makefile:381: recipe for target 'cudaDecodeGL' failed

```
解决了

1.

1.  `vi  ~/.bashrc `
     adding following line into end:
     **LD_LIBRARY_PATH=/usr/lib/nvidia-340:$LD_LIBRYARY_PATH**
     check the LD_LIBRARY_PATH:
     `echo $LD_LIBRARY_PATH: `

2.  `$grep "nvidia-367" -r ./`
      **将 UBUNTU_PKG_NAME = "nvidia-367" 换成UBUNTU_PKG_NAME = "nvidia-375"**

       `$sudo sed -i "s/nvidia-367/nvidia-375/g" `grep nvidia-367 -rl .``
       `$sudo make`
```
 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_60,code=sm_60 -gencode arch=compute_60,code=compute_60 -o nvgraph_SemiRingSpmv nvgraph_SemiRingSpmv.o  -lnvgraph
mkdir -p ../../bin/x86_64/linux/release
cp nvgraph_SemiRingSpmv ../../bin/x86_64/linux/release
make[1]: Leaving directory `/usr/local/cuda-8.0/samples/7_CUDALibraries/nvgraph_SemiRingSpMV'
Finished building CUDA samples
```

全部编译完成后， 进入 samples/bin/x86_64/Linux/release, sudo下运行deviceQuery
sudo ./deviceQuery

```
  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 / 3 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >
> Peer access from Graphics Device (GPU0) -> Graphics Device (GPU1) : Yes
> Peer access from Graphics Device (GPU1) -> Graphics Device (GPU0) : Yes

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 8.0, CUDA Runtime Version = 8.0, NumDevs = 2, Device0 = Graphics Device, Device1 = Graphics Device
Result = PASS
```
*******************************************************************************************************************

5.2 test example

  $5_../
  $./nbody

...
	-hostmem          (stores simulation data in host memory)
	-benchmark        (run benchmark to measure performance)
	-numbodies=<N>    (number of bodies (>= 1) to run in simulation)
	-device=<d>       (where d=0,1,2.... for the CUDA device to use)
	-numdevices=<i>   (where i=(number of CUDA devices > 0) to use for simulation)
	-compare          (compares simulation results running once on the default GPU and once on the CPU)
	-cpu              (run n-body simulation on the CPU)
	-tipsy=<file.bin> (load a tipsy model file for simulation)

NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.

> Windowed mode
> Simulation data stored in video memory
> Single precision floating point simulation
> 1 Devices used for simulation
> Compute 6.1 CUDA device: [Graphics Device]


********************************************************************************************************************
6.
环境：
ubuntu14.04.5
nvidia 375.39
问题描述：
在编译caffe runtest时候，出现的问题
/sbin/ldconfig.real: /usr/lib/nvidia-375/libEGL.so.1 不是符号连接
/sbin/ldconfig.real: /usr/lib32/nvidia-375/libEGL.so.1 不是符号连接
原因：
系统找的是一个符号连接，而不是一个文件。这应该是个bug....
解决方法：
1.对这两个文件更名
2.重新建立符号连接

sudo mv /usr/lib/nvidia-375/libEGL.so.1 /usr/lib/nvidia-375/libEGL.so.1.org
sudo mv /usr/lib32/nvidia-375/libEGL.so.1 /usr/lib32/nvidia-375/libEGL.so.1.org
sudo ln -s /usr/lib/nvidia-375/libEGL.so.375.39 /usr/lib/nvidia-375/libEGL.so.1
sudo ln -s /usr/lib32/nvidia-375/libEGL.so.375.39 /usr/lib32/nvidia-375/libEGL.so.1