How to install: Onboard Intel GPU (-> x-server), Dedicated Nvidia GPU (-> cuda computations only)

Finally got it working: I had to reinstall Ubuntu (couldn't make it work otherwise). In the fresh installation I first add the following lines to /etc/modprobe.d/blacklist.conf

blacklist nouveau
blacklist lbm-nouveau
alias nouveau off
alias lbm-nouveau off
options nouveau modeset=0

-> reboot

then

# Switch off X-Server
[Ctrl + Alt + F1 ]
sudo service gdm stop # "gdm" for gnome / "ligthdm" for unity
sudo ./NVIDIA-Linux-x86_64-430.40.run --no-open-gl-files --no-drm

during installation:

-> update xorg.conf?: "no"

Reboot after installation.

The important part is here the --no-open-gl-files --no-drm options. Especially drm (direct rendering manager) apparently is guilty of using the Nvidia GPU's memory.