Ubuntu 20.04 安装 CUDA Toolkit 的三种方式
无论采用哪一种方式，首先都需要更新 Ubuntu 软件源和升级到最新版本的软件包。由于国内从 Ubuntu 官方软件源下载速度比较慢，所以，建议采用国内 Ubuntu 镜像源，比如阿里 Ubuntu 软件源或清华大学 Ubuntu 软件源。具体的配置方式是修改配置文件 /etc/apt/sources.list，将其中的 archive.ubuntu.com 替换为 mirrors.alibaba.com 或 mirrors.tuna.tsinghua.edu.cn 。也可以在图形界面应用 "Software & Update" 中，修改 Ubuntu Software 标签页中的 Download from 后的软件源地址。
配置软件源后，采用如下命令进行软件源的更新和软件包的升级。
sudo apt update
sudo apt upgrade
下面介绍在 Ubuntu 20.04 长期支持版本中，安装 CUDA Tools 的三种方式：
方式一：采用 Ubuntu 软件源中的 CUDA Tools 软件包
这种方式安装简单，但安装的 CUDA Toolkit 版本往往不是最新版本。查询目前可安装的 CUDA Toolkit 版本的命令，如下所示
apt search nvidia-cuda-toolkit
具体安装命令如下：
sudo apt install nvidia-cuda-toolkit
方式二：先采用图形界面安装 CUDA 驱动，再安装从 NVIDIA 官网下载的 CUDA Toolkit 安装包
1）图形界面安装 CUDA 驱动
在所有应用中，选择 “Software & Update” 应用，在标签页 "Additional Drivers" 中选择 “nvidia-driver-450-server”，如下图所示：
选择后，单击 “Apply Changes” 按钮，这样就更新并切换到所选驱动。
快捷键 Ctrl + Alt + T 打开 Terminal ，运行 nvidia-smi 命令以验证切换到 CUDA 驱动是否成功。我尝试过 nvidia-driver-460 这个版本，但没有成功，因此使用稍低的版本 nvidia-driver-450-server 。
2）下载并安装 CUDA Toolkit
本机安装的 CUDA Toolkit 版本为 11.0.3，与上一步安装 CUDA 驱动 450 兼容（可以参考下载文件名的尾缀）， 具体下载命令，如下
wget https://developer.download.nvidia.com/compute/cuda/11.0.3/local_installers/cuda_11.0.3_450.51.06_linux.run
安装命令，如下
sudo sh cuda_11.0.3_450.51.06_linux.run
需要注意，安装时，选择不安装 CUDA 驱动，安装记录如下：
===========
= Summary =
===========
Driver: Not Selected
Toolkit: Installed in /usr/local/cuda-11.0/
Samples: Installed in /home/klchang/, but missing recommended libraries
Please make sure that
- PATH includes /usr/local/cuda-11.0/bin
- LD_LIBRARY_PATH includes /usr/local/cuda-11.0/lib64, or, add /usr/local/cuda-11.0/lib64 to /etc/ld.so.conf and run ldconfig as root
To uninstall the CUDA Toolkit, run cuda-uninstaller in /usr/local/cuda-11.0/bin
***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least .00 is required for CUDA 11.0 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
安装结束后，添加环境变量到 ~/.bashrc 文件的末尾，具体添加内容如下：
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64
export PATH=$PATH:/usr/local/cuda/bin
export CUDA_HOME=$CUDA_HOME:/usr/local/cuda
保存后退出。
在 Terminal 中，激活环境变量命令为 source ~/.bashrc 。
测试 CUDA Toolkit 。 通过编译自带 Samples并执行， 以验证是否安装成功。具体命令如下所示：
cd /usr/local/cuda/samples/1_Utilities/deviceQuery
sudo make
./deviceQuery
如果安装成功，则输出类似于如下信息：
./deviceQuery Starting...
CUDA Device Query (Runtime API) version (CUDART static linking)
Detected 1 CUDA Capable device(s)
Device 0: "GeForce RTX 2070 with Max-Q Design"
CUDA Driver Version / Runtime Version 11.0 / 11.0
CUDA Capability Major/Minor version number: 7.5
Total amount of global memory: 7982 MBytes (8370061312 bytes)
(36) Multiprocessors, ( 64) CUDA Cores/MP: 2304 CUDA Cores
GPU Max Clock rate: 1125 MHz (1.12 GHz)
Memory Clock rate: 5501 Mhz
Memory Bus Width: 256-bit
L2 Cache Size: 4194304 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: 1024
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 3 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 Managed Memory: 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 / 1 / 0
Compute Mode:
< Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >
deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 11.0, CUDA Runtime Version = 11.0, NumDevs = 1
Result = PASS
3）下载并安装 cuDNN
从 NVIDIA 官方网址 https://developer.nvidia.com/rdp/cudnn-download 下载 cudnn-11.0-linux-x64-v8.0.5.39.tgz 。
解压压缩包，并把相应的文件，复制到指定目录即可。如下所示：
tar zxvf cudnn-11.0-linux-x64-v8.0.5.39.tgz
sudo cp cuda/include/cudnn* /usr/local/cuda/include
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn* /usr/local/cuda/lib64/libcudnn*
方式三：CUDA 驱动和 CUDA Toolkit 都采用命令行方式安装
首先，需要卸载原有的 NVIDIA 驱动并禁用自带的驱动 nouveau；然后，重启电脑，使用 lsmod | grep nouveau 命令检查禁用自带驱动是否成功；如果禁用成功，则安装从 NVIDIA 官方地址下载的 CUDA Toolkit。其步骤则与方式二相同，差别在于这次需要安装 CUDA 驱动 。更多内容，参见 How to Install CUDA ToolKit 11.0, and Nvidia Display Driver on Ubuntu 20.04。
问题与解答
问题 1，sudo apt update 时，出现有锁无法更新的情况
$ sudo apt update
Reading package lists... Done
E: Could not get lock /var/lib/apt/lists/lock. It is held by process 1379 (packagekitd)
N: Be aware that removing the lock file is not a solution and may break your system.
E: Unable to lock directory /var/lib/apt/lists/
解决方法：
停用 packagekitd，并禁止开机启动，具体命令如下：
systemctl stop packagekitd
systemcrl disable packagekit.service
参考资料
[1] How to Install cuda on Ubuntu 20.04. https://linuxconfig.org/how-to-install-cuda-on-ubuntu-20-04-focal-fossa-linux
[2] Ubuntu16.04安装NVIDIA驱动、实现GPU加速. https://blog.csdn.net/zhang970187013/article/details/81012845
