Skip to content

LabLecture01.md

Latest commit

 

History

History
136 lines (96 loc) · 4.83 KB

LabLecture01.md

File metadata and controls

136 lines (96 loc) · 4.83 KB

Tutorial for the newcomers to ImagingLab@ZJU - 001

为什么是Linux？

Linux 能够满足我们实验室 多人 同时 使用服务器的需求。更多的信息可以参见 9 reasons Linux is a popular choice for servers。

Linux 系统操作和基础命令教程

  • 首先推荐各位新同学先在自己电脑上配置一个Linux系统熟悉环境，可以使用双系统或虚拟机，但我们更推荐使用 Docker 在Windows上搭建Linux环境，具体操作参见 中文教程。

  • 构建好Linux系统（在你自己的电脑中）后，下面1-5步是超级账户建立新用户的操作（服务器端这些操作会由管理员完成）。

Please PAY ATTENTION that the ${} represents an variable, and it shouldn't be typed in your command.

  1. Log in server with the ROOT user
ssh zj_servers2@10.11.64.218 # zj_servers2 是超级账户名，10.11.64.218 是服务器的IP地址
  1. Enter the passwd of sudoer (ask the manager for the passwd of the root)
passwd:xxxxxxxxxx
  1. Add a new user, such as chensq
sudo useradd ${username} # such as `sudo useradd chensq`

sudo passwd ${username}  # such as `sudo passwd chensq`
  1. Set the default shell for the user
# 声明登陆后的shell
sudo usermod -s /bin/bash ${username} # such as `sudo usermod -s /bin/bash chensq`
  1. Set your default folder. PLEASE put it on HDD.
cd ${DIR of HDD} # such as `cd /raid` 进入某一个文件夹路径

sudo mkdir ${username} # such as `sudo mkdir chensq` 建立你自己的文件夹

#声明你这个用户每一次登录以后初始化的文件夹
sudo usermod -d ${DIR of HDD}/${username} ${username} # such as `sudo usermod -d /raid/chensq chensq` 

# 声明这个文件夹的所有权归于于你这个用户，使用这个命令的时候必须小心，不能输错！往年有很多同学都会搞错这个命令，将所有人的文件都归属到自己账户下！
sudo chown -R ${username} ${DIR of HDD}/${username} # such as `sudo chown -R chensq /raid/chensq`
  • 下面的操作就是在你的个人账户中建立可运行的计算环境（Python，CUDA...）
  1. Set your System Environment, copy .bashrc into your folder

bashrc需要仔细介绍，具体内容：什么是.bashrc，有什么用?

# log in as your own user
su ${username} # such as `su chensq`

# copy the bashrc to your folder
cp /home/zj_servers2/.bashrc ~/ # such as `cp /home/zj_servers2/.bashrc ~/`

# activate the system parameters in .bashrc
source .bashrc

# confirm the environment is activate (including cuda, cudnn, and so on)
nvcc -V
  1. Set up your Python Environment
# copy the anaconda into your folder
cp /raid/readme/Anaconda3-2023.03-1-Linux-x86_64.sh ~/

# install the anaconda into your environment
bash Anaconda3-2023.03-1-Linux-x86_64.sh

# check if the path of anaconda is in your bashrc
vim .bashrc

# if not add the line below in the bottom
export PATH=${Anaconda}/bin:$PATH # such as `export PATH=/raid/chensq/anaconda3/bin:$PATH`

# reactivate the environment
source .bashrc

# confirm the python and conda
which python
conda list
  1. Set the Mirror of your anaconda and pip
# Mirror of anaconda, please check the detail of this link
https://mirror.tuna.tsinghua.edu.cn/help/anaconda/

# Mirror of pip, please check the detail of this link
https://mirror.tuna.tsinghua.edu.cn/help/pypi/
  1. Set up the virtual environment of Python
# create a virtual environment with the env_name and the specific python version
conda create -n ${env_name} python=${__version__} # such as `conda create -n pytorch python=3.9.6`

# download the package of python
pip install numpy

# ATTENTION: for pytorch, tensorflow, and JAX... We recommend you to first download these frameworks.
# Because other packages, such as numpy..., have the dependence on these frameworks. SO CHECK IT OUT FIRST!
  1. Other tool for you to use
# tmux (a tool to multiple your terminal, Use tmux, turn off your PC!) 
[tmux tutorial](see http://louiszhai.github.io/2017/09/30/tmux/)

# htop (check the cpu and the pid of processes)
htop

# watch
watch -n ${refresh_time} nvidia-smi # such as watch -n 1 nvidia-smi (means every 1s print the GPU-info)

# check the storage of the disk
df -h

# check the data I/O of the disk
iostat -x ${refresh_time} # such as iostat -x 1 (means every 1s print the data I/O)

# this doc is write in markdown
[markdown tutorial](see https://github.com/guodongxiaren/README)

Create by Chen Shiqi at 2023.06.28

Edit by Chen Shiqi at 2023.09.18