Multiple GPUs (not the same model)

Hi There !

I’m currently working on Style Transfer oriented projects, processes running on cuDNN and Tensoflow, on Ubuntu.
I’m planning on building a new rig dedicated to this project, with a RTX Titan.
I’m currently working with a 1080Ti.
So I was wondering if it would be possible to keep the 1080Ti in the new rig, aside the RTX Titan, not as a SLI obviously, as the models are different, but if it would be possible to use it simply as VRAM supply for the Titan (adding 11 GB of VRAM to the 24 of the Titan, if cuDNN processes fill them up) ?

Edit : I found the answer to my question, sorry for the inconvenience. In fact It depend on the process itself and how it was written.