• machine configuration
  • Linux
  • python-package
  • oh-my-zsh
  • Git
  • brew
  • spacevim
  • bashrc
  • ssh
  • IDE/Editor
  • Java
  • Special track
    • ROS in ubuntu
    • openCV
    • blender
    • meshlab
    • PCL
  • Computing
    • Anaconda
    • GPU
  • source channel
  • Shadowsocks

machine configuration

Linux

• df -hl
• du -h --max-depth=1
• tar -zcvf dirname.tar.gz dirname1 dirname1
• tar -zcvf dirname.zip dirname and tar zxvf dirname.zip
• tar -zcvf dirname.zip dirname and tar zxvf dirname.tar.gz

python-package

• pip install -U scikit-learn
• python -m pip show scikit-learn # to see which version and where scikit-learn is installed
• python -m pip freeze # to see all packages installed in the active virtualenv
• python -c "import sklearn; sklearn.show_versions()" or
• conda install scikit-learn
• conda list scikit-learn
• conda list
• python -c "import sklearn; sklearn.show_versions()"

oh-my-zsh

• sh -c "$(curl -fsSL https://raw.githubusercontent.com/robbyrussell/oh-my-zsh/master/tools/install.sh)"
• run chsh-s/bin/zsh to change default Shell.
• then reboot.

Git

  git config --global user.name "qihao-huang"
  git config --global user.email "qihao.huang@outlook.com"

brew

  /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install.sh)"
  # CN
  /bin/zsh -c "$(curl -fsSL https://gitee.com/cunkai/HomebrewCN/raw/master/Homebrew.sh)"
  sudo chown -R $(whoami) /usr/local/share/doc /usr/local/share/man /usr/local/share/man/man1
  chmod u+w /usr/local/share/doc /usr/local/share/man /usr/local/share/man/man1
  brew install node

spacevim

  curl -sLf https://spacevim.org/install.sh | bash

bashrc

• Check my configs

    echo ${PATH}
    # add new variables in bashrc/bash_profile
    echo 'alias cc-v="conda activate cc-v"' >> ~/.zshrc
    echo 'export PATH="/Users/quiescence/anaconda3/bin:$PATH"' >> ~/.zshrc
    source ~/.zhsrc
    # or
    rezsh
  

ssh

• ssh-keygen -t rsa -C "qihao.huang@outlook.com"
• then run pbcopy < ~/.ssh/id_rsa.pub in mac or cat ~/.ssh/id_rsa.pub | xsel in ubuntu,
• and copy it into Github settings.
• vs code ssh remote config

IDE/Editor

• VS Code

Java

• openJDK
  • brew tap AdoptOpenJDK/openjdk; brew cask install adoptopenjdk11
  • test with java -version and javac -version.
• mac java version
  • echo $(/usr/libexec/java_home)
  • export JAVA_HOME=$(/usr/libexec/java_home) in ~/.bash_profile
    • e.g. echo 'export JAVA_HOME=$(/usr/libexec/java_home)' >> ~/.zshrc

Special track

ROS in ubuntu

• installment

openCV

• Version: 3.3.1 embedded in ROS Kinetic (recommend) or

  pip install opencv-python==3.3.1
  

blender

• download and extract *.tar.bz2
• run */bin/python3.5m for python API.
• set blender Python API if needed.

meshlab

• sudo apt install meshlab.

PCL

• CPU version

  sudo add-apt-repository ppa:v-launchpad-jochen-sprickerhof-de/pcl
  sudo apt-get update
  sudo apt-get install libpcl-dev
  

  param:

  • libpcl-dev for 16.04 or later
  • libpcl-all for 14.04
• For GPU version, compile from source.

Computing

Anaconda

• download anaconda.sh from anaconda
• follow shell/.bash_profile
• conda create -n qihao-dev python=3.7

GPU

• follow scripts/GPU_install.sh
• PyTorch 1.4 with CUDA 9.0/10.0: install PyTorch using conda, DONT install it using system’s python.

  conda install pytorch torchvision -c pytorch # anaconda's conda
  # or
  pip install torch torchvision  # anaconda's pip
  

• test with:

  import torch
  print(torch.__version__) # Version
  print(torch.cuda.is_available()) # GPU is available
  print(torch.cuda.device_count())
  print(torch.cuda.get_device_name(0))
  print(torch.cuda.current_device())
  

• test multi-GPU:
  • tensorflow-mgpu-cifar10
  • pytorch-mgpu-cifar10

  https://github.com/qihao-huang/tensorflow-mgpu-cifar10.git
  cd pytorch-mgpu-cifar10
  export CUDA_VISIBLE_DEVICES=0,1 # parallel training with GPUs 0 and 1.
  nohup python train_cifar10.py > log.txt &
  watch -n 0.5 nvidia-smi
    
  # Tensorflow
  import tensorflow.compat.v1 as tf
  tf.disable_v2_behavior()
  
  # import tensorflow as tf
  print(tf.__version__)
  
  # https://www.tensorflow.org/install/source#gpu
  # export CUDA_VISIBLE_DEVICES=2
  # conda activate tf-1-14
  # export PATH=/usr/local/cuda-10.1/bin:$PATH
  # export LD_LIBRARY_PATH=/usr/local/cuda-10.1/lib64:$LD_LIBRARY_PATH
  # export LIBRARY_PATH=/usr/local/cuda-10.1/lib64:$LIBRARY_PATH
  # nvcc --vesion
    
  os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
  os.environ["CUDA_VISIBLE_DEVICES"] = "2"
  
  from tensorflow.python.client import device_lib
  print(device_lib.list_local_devices())
  
  tf.test.is_gpu_available()
  tf.test.gpu_device_name()
  

• tmux: tutorial

  tmux new -s session_name
  nvidia-smi
  # running training
  # ctrl+b d
  # next time
  tmux attach -t session_name
  

• Tensorboard/Jupyter:

  # Jupyter 
  jupyter kernelspec list
  # python
  import sys
  print(sys.executable)
  print(sys.version)
  print(sys.version_info)
    
  

  hostname -I 
  # connect Cisco Annyconnect to CS/Univ's VPN
  # open local terminal
  # for Tensorboard
  ssh -L 6006:localhost:6006 <your_gpu_acct_username>@10.XXX.XXX.XXX
  # for Jupyter
    
  # remote:
  gpu-interactive
  hostname -I 
  jupyter-lab --no-browser
  # local:
  ssh -L 8888:localhost:8888 <your_gpu_acct_username>@10.XXX.XXX.XXX
  

source channel

• follow scripts/source_channel.sh

Shadowsocks

• download desktop client
• ubuntu:
  1. download shadowsocks client

      sudo add-apt-repository ppa:hzwhuang/ss-qt5
      sudo apt-get update
      sudo apt-get install shadowsocks-qt5
     

  2. download chromium from software center.
  3. set network proxy with auto mode using file:///path/to/configs/autoproxy.pac
• mac: download ShadowsocksX-NG from GitHub.