准备工作:
ubuntu16.04启动盘
pycharm
python3.6.4
NVIDIA驱动—对应主机英特尔显卡下载!!!注意:英特尔!!!
CUDA11.0
CUDNN8.0.5
启动盘制作+Ubuntu系统安装: https://blog.csdn.net/m0_46254797/article/details/121641045
--------------------------------开始--------------------------------
pycharm官网:https://www.jetbrains.com/pycharm/download/#section=linux
右键解压缩,放在Downloads下
进去 bin目录下,运行命令
sh ./pycharm.sh
pycahrm快捷方式:/usr/share/applications
--------------------------pycahrm安装完毕-------------------
----------------------------软件和更新--------------------------
Ubuntu16.04提示软件和更新时,是在默认python3.5下更新,如果更新到python3.6,需要降到3.5
sudo -su root
rm /usr/bin/python3
ln -s /usr/bin/python3.5 /usr/bin/python3
再进行软件更新
---------------------开始安装python3.6----------------------------
安装python3.6.13
sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys 6AF0E1940624A220
sudo apt update
sudo apt install python3.6
安装pip pip3 pip3.6
sudo apt install curl
sudo curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py
sudo python36(没啥用这一步哈哈哈哈哈)
sudo python3.6 get-pip.py
//现在python指向2.7
//python3指向3.5
//python3.6指向3.6.13
//pip、pip3、pip3.6 全部指向python3.6.13
which python3.6 一般是在/usr/bin/python3.6
pycharm设置python3.6.13: file–>settings–>project:xxx–>python interpreter–>system interpreter–>/usr/bin/python3.6
sudo pip install numpy 安装numpy
sudo pip install h5py 安装h5py
sudo pip install sklearn 安装sklearn
sudo pip install scipy 安装scipy
sudo pip install opencv-python 安装cv2
sudo pip install tensorflow-gpu==1.4 安装TensorFlow1.4(先不要安装!!继续往下看)
------------------------python3.6.13安装完成---------------------
-------------------------安装NVIDIA驱动----------------------------
1.重启系统–>BIOS(惠普按F10)–>关闭Secure boot(必须关闭!!!)
2.安装相关依赖:
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
sudo apt-get install libopenblas-dev liblapack-dev libatlas-base-dev
sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-dev
3.下载nvidia驱动(对应自己主机的英特尔显卡)
https://www.nvidia.com/Download/index.aspx?lang=en-us%EF%BC%8C%E6%9F%A5%E7%9C%8B%E9%80%82%E5%90%88%E8%87%AA%E5%B7%B1%E6%98%BE%E5%8D%A1%E7%9A%84%E9%A9%B1%E5%8A%A8%E5%B9%B6%E4%B8%8B%E8%BD%BD%EF%BC%9A
4.安装驱动
(1).run文件,路径不要出现中文!
(2)在终端输入:
sudo gedit /etc/modprobe.d/blacklist.conf
(3)输入密码,然后在打开的文件末尾加上:
blacklist nouveau
options nouveau modeset=0
保存文件后再退出
(4)在终端输入:(这里是更新系统):
sudo update-initramfs -u
更新完系统后,要重启电脑。
(5)电脑重启后,在终端的输入:(验证nouveau是否已禁用)
lsmod | grep nouveau
如果输入该命令,回车,没有任何信息显示,说明nouveau已经被禁用
(6)按Ctrl+Alt+F1,进入文字界面,输入用户名及密码,登录。
(7) 输入:(关闭图形界面,这个命令必须执行,否则会出错)
sudo service lightdm stop
(8)如果你之前有其他版本的nvidia驱动,在这就要先卸载之前的驱动再安装新的驱动
sudo apt-get remove nvidia-*
如果没有nvidia驱动,这条可以忽略
(9)用cd命令进入到.run文件所在目录,在终端输入:(给驱动文件赋予执行权限)
sudo chmod a+x xxxx.run
(10)安装(执行该命令也是要在.run文件目录下)
sudo sh ./xxxx.run -no-x-check -no-nouveau-check -no-opengl-files
//只有禁用opengl这样安装才不会出现循环登陆的问题
// -no-opengl-files:只安装驱动文件,不安装OpenGL文件
(11)The distribution-provided pre-install script failed! Are you sure you want to continue?
选择 yes 继续。
Would you like to register the kernel module souces with DKMS? This will allow DKMS to automatically build a new module, if you install a different kernel later?
选择 No 继续。
问题没记住,选项是:install without signing
问题大概是:Nvidia’s 32-bit compatibility libraries?
选择 No 继续。
Would you like to run the nvidia-xconfigutility to automatically update your x configuration so that the NVIDIA x driver will be used when you restart x? Any pre-existing x confile will be backed up.
选择 Yes 继续
(12)验证是否安装成功 :
nvidia-smi
(13)恢复图形界面
sudo service lightdm start
(14)重启系统
--------------------------------NVIDIA驱动安装完成-------------------------------
------------------------------------安装cuda-----------------------------------------
cuda网址: https://developer.nvidia.com/cuda-toolkit-archive
cudnn网址:https://developer.nvidia.com/rdp/cudnn-archive
下载cudnn前需要注册一下!!!
按照自己电脑的显卡驱动版本去下载cuda和cudnn
---------------------cuda11.0安装-------------------------------
(1)wget命令下载cuda
wget http://developer.download.nvidia.com/compute/cuda/11.0.2/local_installers/cuda_11.0.2_450.51.05_linux.run
(2)安装.run文件
sudo sh cuda_11.0.2_450.51.05_linux.run
(3)配置环境变量:
gedit ~/.bashrc
(4)文件末尾添加:
export PATH="/usr/local/cuda-11.0/bin:$PATH"
export LD_LIBRARY_PATH="/usr/lcoal/cuda-11.0/lib64:$LD_LIBRARY_PATH"
export CUDA_HOME="/usr/local/cuda-11.0"
(5)环境变量生效:
source ~/.bashrc
(6)cuda版本信息:
nvcc -V
(7)nvidia驱动信息查询:
nvidia-smi
(8)测试cuda是否安装成功:
cd /usr/local/cuda-11.0/samples/1_Utilities/deviceQuery
sudo make
./deviceQuery
结果如下表明安装成功:deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 11.4, CUDA Runtime Version = 11.0, NumDevs = 1
Result = PASS
------------------------文件安装--------------------------------(可选)//如果你的cuda支持的gcc版本过高,可进行gcc降级
1.检查gpu是否安装cuda:
lspci | grep -i nvidia
2.检查linux内核版本:
uname -m && cat /etc/*release
3.查看gcc版本:
gcc --version (cuda8.0最高支持gcc5.3.1)
4.gcc降级:
sudo apt-get install g++-4.9
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-4.9 20
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-5 10
sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-4.9 20
sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-5 10
sudo update-alternatives --install /usr/bin/cc cc /usr/bin/gcc 30
sudo update-alternatives --set cc /usr/bin/gcc
sudo update-alternatives --install /usr/bin/c++ c++ /usr/bin/g++ 30
sudo update-alternatives --set c++ /usr/bin/g++
5.检查系统内核版本并安装相应kernels headers和开发包
uname -r
sudo apt-get install linux-headers-$(uname -r)
---------------------------------安装cudnn8.0.5--------------------------------------
(1)进入下载目录,开始解压:
tar -zxvf cudnn-11.0-linux-x64-v8.0.5.39.tgz
(2)然后需要把cudnn移动到cuda中:
sudo cp -P cuda/lib64/libcudnn* /usr/local/cuda/lib64/
sudo cp cuda/include/cudnn.h /usr/local/cuda/include/
(3)为所有用户设置读取权限
sudo chmod a+r /usr/local/cuda/include/cudnn.h
sudo chmod a+r /usr/local/cuda/lib64/libcudnn*
(4)在终端查看cudnn版本信息:
sudo cp cuda/include/cudnn_version.h /usr/local/cuda/include/
cat /usr/local/cuda/include/cudnn_version.h | grep CUDNN_MAJOR -A 2
(5)结果如下:显示版本信息说明成功
#define CUDNN_MAJOR 8
#define CUDNN_MINOR 0
#define CUDNN_PATCHLEVEL 5
#define CUDNN_VERSION (CUDNN_MAJOR * 1000 + CUDNN_MINOR * 100 + CUDNN_PATCHLEVEL)
#endif /* CUDNN_VERSION_H */
-----------------------------------cudnn8.0.5安装完毕---------------------------------
此时发现cuda11.0对应的tensorflow版本是tensorflow-2.4.0版本(如果你代码中是1.x版本,可能会出现不兼容问题,
不过问题大,我们重新装一下nvidia驱动,cuda和cudnn就好,不需要重装系统)
sudo pip install tensorflow==2.4.0