ubuntu16.04 nvidia driver + cuda8.0 + cudnn

https://developer.nvidia.com/cuda-toolkit-archive

https://developer.nvidia.com/cudnn

https://pan.baidu.com/s/1hs23HrA

https://pan.baidu.com/s/1o8mc4YI

 

ubuntu16.04

步骤：0，2.1，3，4

0.  关闭 secure boot

在BIOS中关闭Secure Boot（即关闭安全启动）！ 

 

1. 安装驱动( 可不安装）

系统设置->软件更新->附加驱动->切换到最新的NVIDIA驱动即可。应用更改->重启

 

1. sudo ./NVIDIA-Linux-x86_64-390.77.run --no-x-check --no-nouveau-check --no-opengl-files

2. 可能会出现提示The distribution-provided pre-install script failed are you sure you want to continue，没关系，继续进行

3. Would you like to register the kernel module sources with DKMS?This will allow DKMS to auomatically build a new module,if you install a different kernel later，选择No

4. Nvidia's 32-bit compatibility libraries，选择no即可

5. 在ty1下安装，  service lightgdm stop

运行nvidia-smi

卸载驱动：  Nvidia_xx.RUN --uninstall

           aptitude purge nvidia-*

2. cuda

2.0 run 不推荐

安装 cuda 9.0：

./cuda_xxx_linux.run--tmpdir=/tmp

驱动不装，其他的都要

 

2.1 推荐 驱动/cuda 一起安装

sudo dpkg -i cuda-repo-ubuntu1604-9-0-local_9.0.176-1_amd64.deb
sudo apt-key add /var/cuda-repo-<version>/7fa2af80.pub
sudo apt-get update
sudo apt-get install cuda
 

 

export CUDA_HOME=/usr/local/cuda

exportPATH=$PATH:$CUDA_HOME/bin

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CUDA_HOME/lib64

 

cd /usr/local/cuda/samples/1_Utilities/deviceQuery

sudo make

./deviceQuery

卸载cuda：./uninstall_cuda_toolkit_x.x.pl

 

3. cudnn

tar zxvfcudnn-8.0-linux-x64-v6.0.tgz

cp cuda/include/cudnn.h /usr/local/cuda/include/

cp cuda/lib64/libcudnn* /usr/local/cuda/lib64/

chmoda+r /usr/local/cuda/include/cudnn.hsudo

chmoda+r /usr/local/cuda/lib64/libcudnn*

 

4. pip install tensorflow_gpu