ubuntu20.04安装nvidia显卡驱动/CUDA/cuDNN

1 安装显卡驱动
- 1.1 禁止secure boot
- 1.2 安装闭源驱动
- 1.3 重启查看驱动
2 安装CUDA
- 2.1 下载
- 2.2 安装CUDA
- 2.3 配置环境
- 2.4 查看CUDA
3 安装cuDNN
- 3.1 下载
- 3.2 安装cuDNN
- 3.3 查看cuDNN
4 安装gpu版Pytorch

1 安装显卡驱动

1.1 禁止secure boot

这一步很重要，如果bios中为enable，驱动将安装失败。

windows每次开机有可能自动开启secure boot，所以安装前要确认为禁用状态。

1.2 安装闭源驱动

图形模式安装

命令行模式安装

1
2
3
4
5
6
7
8
9
10
11
12
13

zjy@zjy-HP-ENVY-Laptop-13-ad0xx:~$ ubuntu-drivers devices

== /sys/devices/pci0000:00/0000:00:1c.0/0000:01:00.0 ==
modalias : pci:v000010DEd00001D12sv0000103Csd0000834Cbc03sc02i00
vendor : NVIDIA Corporation
model : GP108M [GeForce MX150]
driver : nvidia-driver-390 - distro non-free
driver : nvidia-driver-418-server - distro non-free
driver : nvidia-driver-450-server - distro non-free
driver : nvidia-driver-450 - distro non-free
driver : nvidia-driver-455 - distro non-free recommended
driver : nvidia-driver-440-server - distro non-free
driver : xserver-xorg-video-nouveau - distro free builtin

这里选择 recommended，即：

1	sudo apt install nvidia-driver-455

1.3 重启查看驱动

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

如上图所示，显卡安装成功。

如果不放心可以用下面的指令查看：

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

zjy@zjy-HP-ENVY-Laptop-13-ad0xx:~/桌面$ neofetch

.-/+oossssoo+/-. zjy@zjy-HP-ENVY-Laptop-13-ad0xx
`:+ssssssssssssssssss+:` -------------------------------
-+ssssssssssssssssssyyssss+- OS: Ubuntu 20.04.1 LTS x86_64
.ossssssssssssssssssdMMMNysssso. Host: HP ENVY Laptop 13-ad0xx
/ssssssssssshdmmNNmmyNMMMMhssssss/ Kernel: 5.4.0-54-generic
+ssssssssshmydMMMMMMMNddddyssssssss+ Uptime: 1 min
/sssssssshNMMMyhhyyyyhmNMMMNhssssssss/ Packages: 1741 (dpkg), 6 (snap)
.ssssssssdMMMNhsssssssssshNMMMdssssssss. Shell: bash 5.0.17
+sssshhhyNMMNyssssssssssssyNMMMysssssss+ Resolution: 2560x1440
ossyNMMMNyMMhsssssssssssssshmmmhssssssso DE: GNOME
ossyNMMMNyMMhsssssssssssssshmmmhssssssso WM: Mutter
+sssshhhyNMMNyssssssssssssyNMMMysssssss+ WM Theme: Adwaita
.ssssssssdMMMNhsssssssssshNMMMdssssssss. Theme: Yaru [GTK2/3]
/sssssssshNMMMyhhyyyyhdNMMMNhssssssss/ Icons: Yaru [GTK2/3]
+sssssssssdmydMMMMMMMMddddyssssssss+ Terminal: gnome-terminal
/ssssssssssshdmNNNNmyNMMMMhssssss/ CPU: Intel i5-7200U (4) @ 3.100GHz
.ossssssssssssssssssdMMMNysssso. GPU: Intel HD Graphics 620
-+sssssssssssssssssyyyssss+- GPU: NVIDIA GeForce MX150
`:+ssssssssssssssssss+:` Memory: 840MiB / 7715MiB
.-/+oossssoo+/-.

正确识别到了核显和独显。

2 安装CUDA

2.1 下载

下载地址：https://developer.nvidia.com/zh-cn/cuda-downloads
安装指南：https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html
在这里插入图片描述

2.2 安装CUDA

执行：

1	sudo sh cuda_11.1.0_455.23.05_linux.run

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

x End User License Agreement x
x - x
x NVIDIA Software License Agreement and CUDA Supplement to x
x Software License Agreement. x
x x
x Preface x
x - x
x The Software License Agreement in Chapter 1 and the Supplement x
x in Chapter 2 contain license terms and conditions that govern x
x the use of NVIDIA software. By accepting this agreement, you x
x agree to comply with all the terms and conditions applicable x
x to the product(s) included herein. x
x x
x NVIDIA Driver x
x
x Do you accept the above EULA? (accept/decline/quit):

输入accept，回车。

1
2
3
4
5
6
7
8
9

│ CUDA Installer
│ - [ ] Driver
│ [ ] 455.23.05
│ + [X] CUDA Toolkit 11.1
│ [X] CUDA Samples 11.1
│ [X] CUDA Demo Suite 11.1
│ [X] CUDA Documentation 11.1
│ Options
│ Install

在对应选项上按空格（space）取消勾选Driver，再向下选中Install，回车。
在这里插入图片描述
安装完成后显示：

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

===========
= Summary =
===========

Driver: Not Selected
Toolkit: Installed in /usr/local/cuda-11.1/
Samples: Installed in /home/zjy/, but missing recommended libraries

Please make sure that
- PATH includes /usr/local/cuda-11.1/bin
- LD_LIBRARY_PATH includes /usr/local/cuda-11.1/lib64, or, add /usr/local/cuda-11.1/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run cuda-uninstaller in /usr/local/cuda-11.1/bin
***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least .00 is required for CUDA 11.1 functionality to work.
To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:
sudo <CudaInstaller>.run --silent --driver

Logfile is /var/log/cuda-installer.log

2.3 配置环境

装完成后，需要配置环境变量，编辑~/.bashrc文件，如果其他用户需要使用cuda，再按照上述步骤添加环境变量并更新即可：

1	source ~/.bashrc

在末尾添加：

1
2
3

export CUDA_HOME=/usr/local/cuda
export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:${CUDA_HOME}/lib64
export PATH=${CUDA_HOME}/bin:${PATH}

添加后更新环境变量：

1	source ~/.bashrc

cuda安装目录为/usr/local/cuda-xxx，xxx为版本号，同时，cuda还会建立一个/usr/local/cuda同步链接，所以可以直接将该路径添加至环境变量，之后更换cuda版本，也不需要修改环境变量。

参考：Ubuntu18.04 LTS 使用CUDA11.1编译TensoFlow-GPU版本

2.4 查看CUDA

1
2
3
4
5
6
7

zjy@zjy-HP-ENVY-Laptop-13-ad0xx:~/下载$ nvcc -V

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2020 NVIDIA Corporation
Built on Tue_Sep_15_19:10:02_PDT_2020
Cuda compilation tools, release 11.1, V11.1.74
Build cuda_11.1.TC455_06.29069683_0

3 安装cuDNN

3.1 下载

下载地址：https://developer.nvidia.com/rdp/cudnn-download
在这里插入图片描述
这里选择cuDNN Library for Linux (x86_64)

安装指南：https://docs.nvidia.com/deeplearning/cudnn/install-guide/index.html

3.2 安装cuDNN

Navigate to your directory containing the cuDNN tar file.
Unzip the cuDNN package.

1
tar -xzvf cudnn-x.x-linux-x64-v8.x.x.x.tgz
Copy the following files into the CUDA Toolkit directory.

1
2
3
sudo cp cuda/include/cudnn*.h /usr/local/cuda/include
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn*.h /usr/local/cuda/lib64/libcudnn*

3.3 查看cuDNN

1
2
3
4
5
6
7
8
9

zjy@zjy-HP-ENVY-Laptop-13-ad0xx:~$ cat /usr/local/cuda/include/cudnn_version.h | grep CUDNN_MAJOR -A 2

#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 */

安装成功。

4 安装gpu版Pytorch

选择对应版本的pytorch：https://pytorch.org/
在这里插入图片描述
在新环境中安装，并测试：

1
2
3
4
5
6
7
8

(pytorch_gpu) zjy@zjy-HP-ENVY-Laptop-13-ad0xx:~$ python
Python 3.7.9 (default, Aug 31 2020, 12:42:55)
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.

>>> import torch
>>> torch.cuda.is_available()
True #成功