Ubuntu20.04 配置记录

Just for record Ubuntu20.04 Setup

0x00 apt换cn源

1 2	sudo apt update sudo apt upgrade -y

0x01 apt install

sudo apt install \
    git git-lfs wget curl vim \
    language-pack-zh-hans tree \
    ncdu htop tmux nnn nload \
    ca-certificates gpg \
    net-tools make gcc g++

0x02 fonts

Optional
I use Font:‘Meslo LG L Regular for Powerline’

1 2	git clone git@github.com:powerline/fonts.git sh fonts/install.sh

0x03 install zsh & ohmyzsh

install zsh

1	sudo apt install zsh

install ohmyzsh

1	sh -c "$(curl -fsSL https://raw.githubusercontent.com/ohmyzsh/ohmyzsh/master/tools/install.sh)"

ohmyzsh plugins

pushd .oh-my-zsh
for plug in {zsh-autosuggestions,zsh-syntax-highlighting,zsh-completions}; do
	git clone git@github.com:zsh-users/$plug.git ~/.oh-my-zsh/custom/plugins/$plug
done;
popd

manually change default $SHELL
1
chsh -s /usr/bin/zsh
重启终端后可以看到
1
2
$ echo $SHELL
/usr/bin/zsh

clipboard

Optional GUI required
install dependency, xclip on ubuntu

pbcopy, pbpaste (macOS)
cygwin (Windows running Cygwin)
wl-copy, wl-paste (if $WAYLAND_DISPLAY is set)
xsel (if $DISPLAY is set)
xclip (if $DISPLAY is set)
lemonade (for SSH) https://github.com/pocke/lemonade
doitclient (for SSH) http://www.chiark.greenend.org.uk/~sgtatham/doit/
win32yank (Windows)
tmux (if $TMUX is set)

Usage:

<command> | clipcopy  # copy from stdout
clipcopy <file>       # copy file content

clippaste             # paste to stdout, same to CTRL+C
clippaste | <command> # paste to pipe
clippaste > <file>    # paste contents to a file

0x04 LLVM stuff

LLVM download package

wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add -

sudo apt update

sudo apt install clangd-17 clang-format-17 libomp-17-dev

0x05 CMake

Install from apt repo

wget -O - https://apt.kitware.com/keys/kitware-archive-latest.asc 2>/dev/null \
 | gpg --dearmor - \
 | sudo tee /usr/share/keyrings/kitware-archive-keyring.gpg >/dev/null

echo 'deb [signed-by=/usr/share/keyrings/kitware-archive-keyring.gpg] https://apt.kitware.com/ubuntu/ focal main' \
 | sudo tee /etc/apt/sources.list.d/kitware.list >/dev/null

sudo apt-get update

# install 3.25.2 (CLion not support 3.26 yet)
sudo apt install cmake{,-data,-qt-gui}=3.25.2-0kitware1ubuntu20.04.1

sudo apt-mark hold cmake cmake-data cmake-qt-gui

0x06 NVIDIA stuff

Driver

driver装最新的就行，高版本兼容低版本
skip this section on WSL, nvidia driver on windows already done this

download a xxx.run file
install NVIDIA driver bash NVIDIA-Linux-x86_64-530.41.03.run
reboot
install (cuda)[https://developer.nvidia.com/cuda-toolkit-archive]

NVIDIA Docker

Work on Ubuntu,Debian, WSL as well.

First install Docker

Install Docker Engine

WSL

install Windows Docker-Desktop

Linux

Example for Ubuntu>18.04:

install from apt repo

Install using the repository

sudo install -m 0755 -d /etc/apt/keyrings
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /etc/apt/keyrings/docker.gpg
sudo chmod a+r /etc/apt/keyrings/docker.gpg

echo \
  "deb [arch="$(dpkg --print-architecture)" signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linux/ubuntu \
  "$(. /etc/os-release && echo "$VERSION_CODENAME")" stable" | \
  sudo tee /etc/apt/sources.list.d/docker.list > /dev/null

sudo apt-get update

sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin

Install nvidia-docker2

official doc

Note that in some cases the downloaded list file may contain URLs that do not seem to match the expected value of distribution which is expected as packages may be used for all compatible distributions. As an examples:

For distribution values of ubuntu20.04 or ubuntu22.04 the file will contain ubuntu18.04 URLs

For a distribution value of debian11 the file will contain debian10 URLs

distribution=$(. /etc/os-release;echo $ID$VERSION_ID)

curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-docker-keyring.gpg

curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-docker-keyring.gpg] https://#g' | sudo tee /etc/apt/sources.list.d/nvidia-docker.list

sudo apt-get update

sudo apt-get install -y nvidia-docker2

Update daemon.json

Edit docker config

1	sudo vim /etc/docker/daemon.json

set default-runtime to nvidia.
set data root of docker contents.

Sample:

{
    "runtimes": {
        "nvidia": {
            "path": "nvidia-container-runtime",
            "runtimeArgs": []
        }
    },
    "default-runtime":"nvidia",
    "data-root":"/opt/docker_root"
}

Then, restart docker

1	systemctl restart docker # or daemon reload

Test Nvidia Docker

Test with official docker images:

docker run -it --rm nvcr.io/nvidia/k8s/cuda-sample:devicequery-cuda11.7.1-ubuntu20.04

docker run -it --rm nvcr.io/nvidia/k8s/cuda-sample:vectoradd-cuda11.7.1-ubi8

docker run nvcr.io/nvidia/k8s/cuda-sample:nbody -benchmark -numbodies=1024

CUDA toolkit

Linux

official doc

Cuda install .deb file contains a driver, so choose a specific package to install meta package

CUDNN

download from tar file

doc

download files from https://developer.download.nvidia.com/compute/cudnn/redist/cudnn/linux-x86_64/
copy files into the CUDA directory

tar -xvf cudnn-linux-x86_64-8.x.x.x_cudaX.Y-archive.tar.xz
sudo cp cudnn-*-archive/include/cudnn*.h /usr/local/cuda/include 
sudo cp -P cudnn-*-archive/lib/libcudnn* /usr/local/cuda/lib64 
sudo chmod a+r /usr/local/cuda/include/cudnn*.h /usr/local/cuda/lib64/libcudnn*

other

libcudnn不会和cudatoolkit一起安装，nvidia提供下载页面，下载后copy到cuda的lib目录下。对应docker label:nvidia/cuda:xxxx-cudnn-xxx。

有一个trick:pip 安装的torch中带一个libcudnn，在torch的安装位置的lib里，如/usr/lib/python3.10/site-packages/torch/lib。不然torch的gpu版的体积也不用这么大。

还可以使用pip安装。如pip install nvidia-cudnn-xxxxx。

WSL

WSL CUDA

修复 libcuda.so.1 is not a symbolic link

原因：
WSL nvidia-container-cli 需要用 /usr/lib/wsl/lib下的东西，而且都是copy，不可以是link
WSL 内ldconfig 需要用 {xxx}.so link to file {xxx}.so.1.1

issue and solution

WSL内/usr/lib/wsl/lib是auto mount%SYSTEMROOT%\System32\lxss\lib的，把这个automount关掉，修改/etc/wsl.conf

1 2	[automount] ldconfig = false

创建link：

1 2	mkdir /usr/lib/wsl/lib2 ln -s /usr/lib/wsl/lib/* /usr/lib/wsl/lib2

然后修改/etc/ld.so.conf.d/ld.wsl.conf，让ldconfig全用link的

1	/usr/lib/wsl/lib2

最后重新启动wsl。

Post-installation Actions

doc

set path

1
2
3

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

Verify the Installation

check version

1	cat /proc/driver/nvidia/version

run samples from https://github.com/nvidia/cuda-samples

0x07 Setup python env with conda

see Install MiniConda

0x08 OpenCL

WSL不支持Opencl No OpenCL platforms reported

sudo apt install ocl-icd-opencl-dev \
                    libx11-dev \
                    libxcb1-dev \
                    libgles-dev \
                    libglx-dev \
                    libomp-dev \
                    libglfw3-dev \
                    libglew-dev \
                    libx264-dev \
                    libx265-dev \
                    nasm \
                    clinfo

Check status:

clinfo

0x09 OpenCV

git clone -b 4.5.5 --depth 1 git@github.com:opencv/opencv.git
pushd opencv
mkdir -p build
pushd build
cmake ..
make -j16
sudo make install
popd
popd

0x0a Gnome Extention

简单记录几个我在用的extension

Install

Documentation

1	sudo apt-get install chrome-gnome-shell gnome-tweaks gnome-shell-extensions

Chrome extension: Gnome shell integration

0x00 apt换cn源

0x01 apt install

0x02 fonts

0x03 install zsh & ohmyzsh

install zsh

install ohmyzsh

ohmyzsh plugins

clipboard

0x04 LLVM stuff

0x05 CMake

0x06 NVIDIA stuff

Driver

NVIDIA Docker

First install Docker

WSL

Linux

Install nvidia-docker2

Update daemon.json

Test Nvidia Docker

CUDA toolkit

Linux

CUDNN

download from tar file

other

WSL

修复 libcuda.so.1 is not a symbolic link

Post-installation Actions

set path

Verify the Installation

0x07 Setup python env with conda

0x08 OpenCL

0x09 OpenCV

0x0a Gnome Extention

Install

My Extensions: