This image shows two configurations of a Nested DFX design. Both designs have the same static logic (Top) and floorplan for RP A. RP A is further divided into two more RPs, W and X for RM A1, or Y and Z for RM A2. RP A could have more RM versions (A3, A4, etc.) and these could have any number, size or shape sub-RPs, even none at all. The lower level RPs W, X, Y, and Z will each have their own collection of RMs (W1, W2, W3, etc.), and must each be implemented with all implementation results above them locked.
In the Nested DFX implementation flow, users implement each RM in the context of the static design above it. The first RM for any RP establishes the static implementation results for the RM level immediately above it (A1 or A2 in this case), then all remaining RMs are implemented into this context. This is the flow regardless of where in the hierarchy the current target RP is.