The AXI Register Slice IP core is typically used for adding pipeline registers between memory mapped or streaming AXI interfaces to help close timing. For larger devices, adding the right amount of pipelining without overly increasing the register utilization and the application latency is a common challenge. To simplify the pipeline insertion task and allow the Vivado placer more flexibility, you can use the auto-pipeline optimization feature for the AXI Register Slice IP core. When this feature is enabled, a special physical synthesis phase (between the floorplanning and global placer phases) inserts and places the additional pipeline stages based on setup timing slack and SLR distance. The AXI Register Slice IP core remains compliant with the AXI handshake protocol despite the increased latency due to the use of small FIFOs.
You can enable this feature in the IP Configuration Wizard. Set the Register Slice Options (REG_*) to Multi SLR Crossing. In addition, set the Use timing-driven pipeline insertion for all Multi-SLR channels option to 1 to enable auto-pipelining. The following figure shows an example.