GLX Commands

commands.sh

# On local MacOS machine...

  1) Download and Install TurboVCN

    Only source for binaries was sourceforge: https://sourceforge.net/projects/turbovnc/files/2.1.2/
    `TurboVNC-2.1.2.dmg` failed to run on MacOS Siera, but `TurboVNC-2.1.2-AppleJava.dmg` was sucessful.
    Only after installing an old version of Java: https://support.apple.com/kb/DL1572?locale=en_GB

  2) Start AWS p3.2xlarge instance

  3) Add to ~/.ssh/config

    ```
    Host glx.com
        HostName [instance-ip].amazonaws.com
        User ubuntu
        IdentityFile ~/.ssh/[keypair].pem
        LocalForward 5901 localhost:5901
    ```

  4) SSH onto instance with `ssh glx.com`

# On remote AWS EC2 instance...

  1) Get GLX test utils

    ```
    sudo apt-get update
    sudo apt-get install -y mesa-utils
    glxinfo
    glxgears
    ```

  2) Download NVIDIA drivers, for LibGL

    ```
      wget http://us.download.nvidia.com/tesla/396.44/nvidia-diag-driver-local-repo-ubuntu1604-396.44_1.0-1_amd64.deb
      sudo dpkg -i nvidia-diag-driver-local-repo-ubuntu1604-396.44_1.0-1_amd64.deb
      sudo apt-key add /var/nvidia-diag-driver-local-repo-396.44/7fa2af80.pub
      sudo apt-get install cuda-drivers
      sudo reboot
    ```

  3) Add NVidia LibGL to path

    * Find Location of NVidia `libGL.so`: sudo find / -name 'libGL.so'
    * Add NVidia libs path (via `.bashrc`): `export LD_LIBRARY_PATH=/usr/lib/nvidia-396/:$LD_LIBRARY_PATH`
    * `source .bashrc`

  4) Confirm NVidia LibGL is used

    ```
    ldd `which glxgears`
    ```

    And look for something like: `libGL.so.1 => /usr/lib/nvidia-396/libGL.so.1 (0x00007f336ff83000)`

### LOG HISTORY

# Get CUDA version
   16  nvcc --version

# Get GLX test
  109  sudo apt-get update
   29  sudo apt-get install -y mesa-utils
   30  glxinfo
   glxgears

# Install xserver
   49  sudo apt-get -y install xserver-xorg

# Install mesa drivers (as backup)
  sudo apt-get install -y libglu1-mesa-dev freeglut3-dev mesa-common-dev libxmu-dev libxi-dev

# Find BUS of NVIDIA
   69  lspci
   79  nvidia-xconfig --query-gpu-info

# Create config
   80  sudo nvidia-xconfig --busid=PCI:0:30:0 --use-display-device=none
   50  sudo nvidia-xconfig -a --use-display-device=None --virtual=1280x1024

# Adjust config
  sudo nano /etc/X11/xorg.conf
  Comment out the server layout, and screen blocks

# Start x server
   51  sudo /usr/bin/X :0 &

# Find location on nvidia driver libs
  111  sudo find / -name 'libGL.so'

# Add to path  
  112  export LD_LIBRARY_PATH=/usr/lib/nvidia-396/:$LD_LIBRARY_PATH

# Check glx is using correct lib
   90  ldd `which glxgears`

# Install virtualgl
  171  wget https://sourceforge.net/projects/virtualgl/files/2.5.2/virtualgl_2.5.2_amd64.deb/download -O virtualgl_2.5.2_amd64.deb
  172  sudo dpkg -i virtualgl*.deb
  173  rm virtualgl*.deb

# Install turboVCN  
  174  wget https://sourceforge.net/projects/turbovnc/files/2.1.1/turbovnc_2.1.1_amd64.deb/download -O turbovnc_2.1.1_amd64.deb
  175  sudo dpkg -i turbovnc*.deb
  176  rm turbovnc*.deb

# Unsorted
  177  sudo service lightdm stop
  178  sudo /opt/VirtualGL/bin/vglserver_config -config +s +f -t
  179  sudo reboot
  180  export LD_LIBRARY_PATH=/usr/lib/nvidia-396/:$LD_LIBRARY_PATH
  181  glxgears
  182  xauth merge /etc/opt/VirtualGL/vgl_xauth_key
  183  xdpyinfo -display :0
  184  /opt/VirtualGL/bin/glxinfo -display :0 -c
  185  exit
  186  export LD_LIBRARY_PATH=/usr/lib/nvidia-396/:$LD_LIBRARY_PATH
  187  ldd `which glxgears`
  188  /opt/TurboVNC/bin/vncserver
  189  export DISPLAY=:1
  190  DISPLAY=1 vglrun glxgears
  191  ls
  192  source activate coach_env
  193  coach -l
  194  vglrun coach -r -p CartPole_PG
  195  /opt/TurboVNC/bin/vncserver
  196  vglrun glxgears
  197  DISPLAY:=1 vglrun glxgears
  198  DISPLAY=1 vglrun glxgears