BrainModular

This seems like a very significant new feature, but how exactly does it work?
1) Does it have any influence on latency?
2) Are there any CPU/efficiency trade-offs to using it?
3) Are the old rules about Rack/signal flow/latency more or less obsolete now? (i.e. Could you build everything in a single Rack/Patch of sub-patches rather than spreading over many Racks?)
4) Is there any reason this feature can't be added to top level patches also?
Thanks!

When this option is activated Usine will execute all sub-patch voices in a separate thread.
It interesting (in CPU terms) when you have heavy sub-patches with a poly >=2.
It doesn't affect the global latency which is given by the sound card (not Usine).
Racks are still calculated in individual threads and top level patches cant have this option because they must be calculated in serie.
You can keep your actual workflow, this new option will, in some situations, reduce the CPU consumption.