Any new RMs can be implemented within Abstract Shells. Each RM can be implemented in parallel in separate Vivado session as each RP is managed independently. The implementation flow is no different than the standard DFX flow starting with the full static design image.
Use the same methodology in the Abstract Shell run as you used in the run that created the original implementation. For example, if the parent implementation uses the add_files / link_design approach (used in project mode), use the same approach for the Abstract Shell child implementation runs. If you used open_checkpoint and read_checkpoint -cell to build the initial design, continue that approach for the Abstract Shell implementation run.
The flow through the implementation tools follows the same steps from
route_design, and like the standard flow, actions performed
by the Vivado tools focus only on the target RM to be implemented. Any new constraints –
such as placement directives, floorplanning, and timing goals – can be applied and
scoped to the target RM.
route_design is complete, call
save the entire Abstract Shell with the implemented RM and call and
write_checkpoint -cell to save just the implemented RM alone. The
RM checkpoint alone can be read back into the full static design checkpoint (along with
other RM checkpoints for other RPs if necessary) to assemble a complete design.