Offloading is correctly set up, just
xrandr --setprovideroffloadsink NVIDIA-G0 modesetting
doesn’t have to be used, that’s for mesa based offloading. With nvidia, just start applications on the nvidia gpu by prepending
__NV_PRIME_RENDER_OFFLOAD=1 __GLX_VENDOR_LIBRARY_NAME=nvidia
to the application, e.g.
__NV_PRIME_RENDER_OFFLOAD=1 __GLX_VENDOR_LIBRARY_NAME=nvidia glxgears
See: https://download.nvidia.com/XFree86/Linux-x86_64/440.64/README/primerenderoffload.html