【Linux】Linux安装navida驱动和cuda开发包

查看硬件版本

   lspci | grep -i nvidia

   NVIDIA Corporation GP100GL [Tesla P100 PCIe 16GB] (rev a1)

根据版本下载对应驱动

   https://www.nvidia.cn/Download/index.aspx?lang=cn
   http://cn.download.nvidia.com/tesla/410.104/NVIDIA-Linux-x86_64-410.104.run

本例为选择了cuda_toolkit_9.0;

驱动版本文件：NVIDIA-Linux-x86_64-384.183.run

安装驱动

   chmod +x ./NVIDIA-Linux-x86_64-384.183.run
   ./NVIDIA-Linux-x86_64-384.183.run

Q:如果提示”The Nouveau kernel driver is currently in use by your system…………”

需要禁用第三方驱动

apt purge nvidia*
vim /etc/modprobe.d/blacklist-nouveau.conf—>
```
blacklist nouveau
options nouveau modeset=0
```
update-initramfs -u
reboot
检查是否禁用lsmod grep nouveau

Q:Verifying archive integrity… Error in check sums 1237657955 1463329162

文件校验错误，需要重新下载

查看驱动信息

nvidia-smi

下载并安装CUDA_toolkit

wget https://developer.nvidia.com/compute/cuda/9.0/Prod/local_installers/cuda_9.0.176_384.81_linux-run

./cuda_9.0.176_384.81_linux-run

Q：unsupported compiler: 7.3.0

g++、gcc的版本问题

降低 g++、gcc的版本.

   gcc --version;
   gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0
   g++ --version;
   gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0
   apt install gcc-6 g++-6;
   ln -s /usr/bin/gcc-6 /usr/local/bin/gcc
   ln -s /usr/bin/g++-6 /usr/local/bin/g++

再执行

./cuda_9.0.176_384.81_linux-run

环境变量配置

添加可执行文件路径

vim ~/.bashrc
export PATH=/usr/local/cuda-9.0/bin:$PATH
source ~/.bashrc

添加动态库鲁健

vim  /etc/ld.so.conf

添加

/usr/local/cuda-9.0/lib64
ldconfig

下载ffmpeg并在编译

git clone  https://git.ffmpeg.org/ffmpeg.git
git checkout n(tag)3.4.5
./configure --enable-cuda --enable-cuvid --enable-nvenc --enable-nonfree --enable-libnpp --extra-cflags=-I/usr/local/cuda/include --extra-ldflags=-L/usr/local/cuda/lib64

make -j 10

Q:nasm/yasm not found or too old

apt-get install yasm

Q:ERROR: cuda requested, but not all dependencies are satisfied: ffnvcodec

git clone https://git.videolan.org/git/ffmpeg/nv-codec-headers.git
cd nv-codec-headers
make
sudo make install
export PKG_CONFIG_PATH=/usr/local/lib/pkgconfig:$PKG_CONFIG_PATH

如何还有问题：检查pkgconf是否安装；apt-get install pkgconf

Q：The minimum required Nvidia driver for nvenc is 390.25 or newer

ffmpeg版本过高，或者驱动版本过低，适当调整

测试： https://developer.nvidia.com/ffmpeg Transcode a single video file FFMPEG: ffmpeg -hwaccel cuvid -c:v h264_cuvid -i -vf scale_npp=1280:720 -c:v h264_nvenc

查看cuda版本

cuda 版本

cat /usr/local/cuda/version.txt

cudnn 版本

cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2

ubuntu_18_04_64

安装tensorflow

检查python开发环境 python3 –version pip3 –version virtualenv –version
安琥在那个cuDNN https://docs.nvidia.com/deeplearning/sdk/cudnn-install/index.html

virtualenv

virtualenv ENV

source /path/to/ENV/bin/activate

deactivate