Ubuntu 18.04 下安装安装NVIDIA显卡驱动+CUDA-10.1+cudnn-7

1 安装显卡驱动

1.1 禁用nouveau驱动

在/etc/modprobe.d/blacklist.conf 里添加，如下内容，并执行 sudo update-initramfs -u 命令,

blacklist nouveau
options nouveau modeset=0

重启后用 lsmod | grep nouveau, 如果没有任何输出说明禁用成功。

1.2 安装NVIDIA显卡驱动

ubuntu 18.04 直接使用sudo ubuntu-drivers autoinstall安装的是390的驱动，cuda 10 显卡必须是410以上，所以要安装新一点的驱动

安装更新版本的驱动可以先添加源：

$ sudo add-apt-repository ppa:graphics-drivers/ppa
$ sudo apt update

然后执行：

$ ubuntu-drivers devices
== /sys/devices/pci0000:00/0000:00:02.0/0000:05:00.0 ==
modalias : pci:v000010DEd00001C82sv000019DAsd00002456bc03sc00i00
vendor   : NVIDIA Corporation
model    : GP107 [GeForce GTX 1050 Ti]
driver   : nvidia-driver-410 - third-party free
driver   : nvidia-driver-415 - third-party free
driver   : nvidia-driver-390 - distro non-free
driver   : nvidia-driver-418 - third-party free
driver   : nvidia-driver-396 - third-party free
driver   : nvidia-driver-430 - third-party free recommended
driver   : xserver-xorg-video-nouveau - distro free builtin
Selecting previously unselected package nvidia-dkms-430.

最后安装

$ sudo apt install nvidia-driver-430

2 安装CUDA

2.1 下载cuda安装包

下载地址如下：

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


这里下载：cuda_10.1.105_418.39_linux.run

2.2 安装

使用如下命令安装：

$ sudo sh cuda_10.1.105_418.39_linux.run --no-opengl-libs

注意第一项显卡驱动不要装，之前安装过了。其他的都装。

2.3 配置环境变量

声明一下环境变量,并将其写入到 ~/.bashrc 文件(在用户目录下)的尾部,输入内容如下

export PATH=/usr/local/cuda-10.1/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-10.1/lib64:$LD_LIBRARY_PATH

保存退出,并输入下面指令使环境变量立刻生效:

$source ~/.bashrc

2.4 设置环境变量和动态链接库

在命令行输入:

$ sudo vim /etc/profile

在打开的文件末尾加入:

export PATH=/usr/local/cuda-10.1/bin:$PATH

2.5 创建链接文件

$ sudo vim /etc/ld.so.conf.d/cuda.conf

在打开的文件中添加如下语句:

/usr/local/cuda-10.1/lib64

保存退出,然后执行

$ sudo ldconfig 

使链接立即生效。

2.6 测试 cuda 是否安装成功

切换到 CUDA 9.1 Samples 默认安装路径(即在/home/用户/ NVIDIA_CUDA-10.1_Samples 目录下), 终端下输入

$ cd ~/NVIDIA_CUDA-10.1_Samples
$ sudo make all -j8
$ cd bin/x86_64/linux/release
$ ./deviceQuery

如果 CUDA 安装成功，则显示相关信息 Result = PASS，具体如下：

$ ./deviceQuery
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 1 CUDA Capable device(s)

Device 0: "GeForce GTX 1050 Ti"
  CUDA Driver Version / Runtime Version          10.2 / 10.1
  CUDA Capability Major/Minor version number:    6.1
  Total amount of global memory:                 4039 MBytes (4234936320 bytes)
  ( 6) Multiprocessors, (128) CUDA Cores/MP:     768 CUDA Cores
  GPU Max Clock rate:                            1392 MHz (1.39 GHz)
  Memory Clock rate:                             3504 Mhz
  Memory Bus Width:                              128-bit
  L2 Cache Size:                                 1048576 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 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 2 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 supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 5 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 10.2, CUDA Runtime Version = 10.1, NumDevs = 1
Result = PASS

3 安装cudnn

3.1 下载cudnn

下载地址：https://developer.nvidia.com/rdp/cudnn-archive

3.2 安装cudnn

$ sudo dpkg -i libcudnn7_7.6.1.34-1+cuda10.1_amd64.deb
$ sudo dpkg -i libcudnn7-dev_7.6.1.34-1+cuda10.1_amd64.deb
$ sudo dpkg -i libcudnn7-doc_7.6.1.34-1+cuda10.1_amd64.deb

3.3 测试 cuDNN

6.1复制 cuDNN sample 到一个目录下,这里复制到 HOME 下

$ cp -r /usr/src/cudnn_samples_v7 /$HOME

6.2进入 HOME 目录

$ cd $HOME/cudnn_samples_v7/mnistCUDNN/

6.3 编译 mnistCUDNN sample

$ make clean && make all –j8

6.4 运行 mnistCUDNN sample

$ ./mnistCUDNN

如果出现 Test passed!表明 cuDNN 已成功安装

$ ./mnistCUDNN
cudnnGetVersion() : 7601 , CUDNN_VERSION from cudnn.h : 7601 (7.6.1)
Host compiler version : GCC 5.5.0
There are 1 CUDA capable devices on your machine :
device 0 : sms  6  Capabilities 6.1, SmClock 1392.0 Mhz, MemSize (Mb) 4038, MemClock 3504.0 Mhz, Ecc=0, boardGroupID=0
Using device 0

Testing single precision
Loading image data/one_28x28.pgm
Performing forward propagation ...
Testing cudnnGetConvolutionForwardAlgorithm ...
Fastest algorithm is Algo 1
Testing cudnnFindConvolutionForwardAlgorithm ...
^^^^ CUDNN_STATUS_SUCCESS for Algo 0: 0.020288 time requiring 0 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 1: 0.038912 time requiring 3464 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 2: 0.044032 time requiring 57600 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 5: 0.101376 time requiring 203008 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 4: 0.121856 time requiring 207360 memory
Resulting weights from Softmax:
0.0000000 0.9999399 0.0000000 0.0000000 0.0000561 0.0000000 0.0000012 0.0000017 0.0000010 0.0000000
Loading image data/three_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0000000 0.0000000 0.0000000 0.9999288 0.0000000 0.0000711 0.0000000 0.0000000 0.0000000 0.0000000
Loading image data/five_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0000000 0.0000008 0.0000000 0.0000002 0.0000000 0.9999820 0.0000154 0.0000000 0.0000012 0.0000006

Result of classification: 1 3 5

Test passed!

Testing half precision (math in single precision)
Loading image data/one_28x28.pgm
Performing forward propagation ...
Testing cudnnGetConvolutionForwardAlgorithm ...
Fastest algorithm is Algo 1
Testing cudnnFindConvolutionForwardAlgorithm ...
^^^^ CUDNN_STATUS_SUCCESS for Algo 0: 0.019360 time requiring 0 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 1: 0.032768 time requiring 3464 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 2: 0.040960 time requiring 28800 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 5: 0.100992 time requiring 203008 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 4: 0.116416 time requiring 207360 memory
Resulting weights from Softmax:
0.0000001 1.0000000 0.0000001 0.0000000 0.0000563 0.0000001 0.0000012 0.0000017 0.0000010 0.0000001
Loading image data/three_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0000000 0.0000000 0.0000000 1.0000000 0.0000000 0.0000714 0.0000000 0.0000000 0.0000000 0.0000000
Loading image data/five_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0000000 0.0000008 0.0000000 0.0000002 0.0000000 1.0000000 0.0000154 0.0000000 0.0000012 0.0000006

Result of classification: 1 3 5

Test passed!

至此，Ubuntu 18.04下安装cuda10 和cudnn7的过程就完成了！！！

4 问题

运行tensorflow时提示如下错误：

2019-07-24 13:01:26.926999: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Could not dlopen library 'libcudart.so.10.0'; dlerror: libcudart.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64:
2019-07-24 13:01:26.927165: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Could not dlopen library 'libcublas.so.10.0'; dlerror: libcublas.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64:
2019-07-24 13:01:26.927349: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Could not dlopen library 'libcufft.so.10.0'; dlerror: libcufft.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64:
2019-07-24 13:01:26.927516: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Could not dlopen library 'libcurand.so.10.0'; dlerror: libcurand.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64:
2019-07-24 13:01:26.927688: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Could not dlopen library 'libcusolver.so.10.0'; dlerror: libcusolver.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64:
2019-07-24 13:01:26.927847: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Could not dlopen library 'libcusparse.so.10.0'; dlerror: libcusparse.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64:

解决安装如下内容：

$ conda install cudatoolkit
$ conda install cudnn

参考：https://github.com/tensorflow/tensorflow/issues/26182

或者重新安装 keras-gpu：

$ conda install keras-gpu