I've not found NVLink to be very useful in Octane (yet) In scenes where I could max out physical vRam, I would get regular Cuda/Optix errors when rendering with NVLink enabled.
In the latest 2020.1.x builds the developers have acknowledged this issue and have marked it as "Currently Under Investigation" It is really helpful for them to communicate this in my opinion:
In Standalone there is a useful blue Peer to Peer usage bar in the bottom right hand corner of the render window. In Octane for C4D there is no such user interface in the live viewer which also makes it less useful in production.
I'm sure Ahmet will add this when we gets time though.
In case you don't already know NVLink allows you to use your second second GPUs memory as an Out of Core source. Due to the transfer rates of PCI Express, it is much more desirable to do this instead of using system memory because there will be less performance degradation.
But at present, unfortunately my NVLink adaptor remains unused due to software still looking for the best way to utilize the technology.