Yes, there is when the unified memory is actually accessible to both CPU and GPU, meaning you only need one copy of resources that are computed and then passed to the GPU.