I'll think about it! Precomputing combinational chips should be a fairly easy optimization to implement, but it's not something I've done yet.
