The reactor and controls are fairly simple. Just a + shape with each line being control-moderator-moderator-reflector. The controls are a simple linear interpolation so that the reflected flux is at break even when the emitted flux is at a desired value, and below/above break even when above/below the target.
The main barrier to getting even more power out is actually just heat pipe throughput and the reactor's internal heat throughput. I think the recent heat capacity change is actually what made the 4GW number possible, when I was experimenting a few days ago I was having some trouble getting past ~1.5GW.