There’s no difference in using Buffers and Buses when using supernova, in my experience.
The only difference is that you can use ParGroup instead of Group to have synths run in parallel on multiple cores of your CPU. For synths inside a parallel group you can’t depend on their order of execution (i.e. you can’t have sound routed from one synth to another - nothing prevents you from doing that but the result is undefined).
You can still do that between synths in a regular Group, or between synths running in different ParGroups inside a regular Group, for example
// (not an actual code)
1000.do (SourceSynth) //sound sources
1000.do (DestinationSynth) //receive sound from SoundSource synths