任何问题，请在网页底部留言，或者联系 inksci@qq.com

update: 2021-04-30

inksci@inksci-ThinkStation-P340:~$ nvcc –V

Command 'nvcc' not found, but can be installed with:

sudo apt install nvidia-cuda-toolkit

inksci@inksci-ThinkStation-P340:~$ sudo apt install nvidia-cuda-toolkit
[sudo] password for inksci: 

$ nvcc –V
gcc: No such file or directory
inksci@inksci-ThinkStation-P340:~$ sudo apt install nvidia-cuda-toolkit^C
inksci@inksci-ThinkStation-P340:~$ cd /usr/local
inksci@inksci-ThinkStation-P340:/usr/local$ ls
bin  etc  games  include  lib  man  sbin  share  src

inksci@inksci-ThinkStation-P340:~$ prime-select query

Command 'prime-select' not found, but can be installed with:

sudo apt install nvidia-prime

inksci@inksci-ThinkStation-P340:~$ sudo apt install nvidia-prime
[sudo] password for inksci: 

inksci@inksci-ThinkStation-P340:~$ prime-select query
nvidia

https://developer.nvidia.com/cuda-downloads

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pinsudo mv cuda-ubuntu2004.pin /etc/apt/preferences.d/cuda-repository-pin-600wget https://developer.download.nvidia.com/compute/cuda/11.3.0/local_installers/cuda-repo-ubuntu2004-11-3-local_11.3.0-465.19.01-1_amd64.debsudo dpkg -i cuda-repo-ubuntu2004-11-3-local_11.3.0-465.19.01-1_amd64.debsudo apt-key add /var/cuda-repo-ubuntu2004-11-3-local/7fa2af80.pubsudo apt-get updatesudo apt-get -y install cuda

/usr/local$ ls
bin  cuda  cuda-11  cuda-11.3  etc  games  include  lib  man  sbin  share  src

https://developer.nvidia.com/rdp/cudnn-archive

'libcudnn8_8.1.1.33-1+cuda11.2_amd64(1).deb'
'libcudnn8-dev_8.1.1.33-1+cuda11.2_amd64(1).deb'
 libcudnn8-samples_8.1.1.33-1+cuda11.2_amd64.deb

安装后

$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Sun_Jul_28_19:07:16_PDT_2019
Cuda compilation tools, release 10.1, V10.1.243

~$ nvidia-smi
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.

reboot

nvidia-smi 正常

============================================

cudnn6 + cuda8 for tensorflow gpu 1.4

cuda8 不支持 tensorflow gpu 1.5

首先，你应该运行

查看电脑显卡信息命令：
lspci | grep VGA

检查你的电脑有哪些显卡：
lspci -k | grep -A 2 -i "VGA"

来查看是否使用了nvidia显卡

对于双显卡电脑，intel和nvidia显卡，应该到boost中将集成显卡调整为switchable显卡

然后可以开始安装 nvidia 驱动了。

参考：Ubuntu系列切换Intel和NVIDIA显卡（我在Linux显卡上遇到的坑）

https://segmentfault.com/a/1190000009269284

看一下系统推荐了什么驱动

sudo ubuntu-drivers devices

按照推荐的安装，然后重启系统

reboot

然后应该是黑屏

乖乖，查看下当前使用的显卡：

prime-select query

应该显示为nvidia

先切换成 intel

sudo prime-select intel

然后重启桌面

sudo service lightdm stop

sudo service lightdm start

应该就进入桌面了，哈哈！

此时，你的附加驱动里应该有了很多选项，在附加驱动里，更改为 340（原理：试试看）

哈哈，可以了！

关于 cudnn 的安装，

到 nvidia 下载 deb 包点击安装即可

libcudnn6_6.0.21-1+cuda8.0_amd64


libcudnn6-dev_6.0.21-1+cuda8.0_amd64


libcudnn6-doc_6.0.21-1+cuda8.0_amd64

此时仍然不能使用：

ImportError: libcublas.so.8.0: cannot open shared object file: No such file or directory

貌似 cuda 没有安装

查看

nvcc –V

根据提示使用 apt-get 安装
在 nvidia 官网下载 cuda-8.0

https://developer.nvidia.com/cuda-80-ga2-download-archive

官方的指示没有更新，应当使用（否则会安装最新的cuda-10.0）：

sudo apt-get install cuda-8.0

export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}:/usr/lib/nvidia-410/

那些我没提到过的 ... 大概不重要。