Using the Timing Report to Determine the Impact of Power Optimization - 2020.2 English

UltraFast Design Methodology Guide for Xilinx FPGAs and SoCs

Document ID
Release Date
2020.2 English

Power optimization works to minimize the impact on timing while maximizing power savings. However, in certain cases, if timing degrades after power optimization, you can employ a few techniques to offset this impact.

Where possible, identify and apply power optimizations only on non-timing critical clock domains or modules using the set_power_opt XDC command. If the most critical clock domain happens to cover a large portion of the design or consumes the most power, review critical paths to see if any cells in the critical path have the IS_CLOCK_GATED property with value TRUE, indicating that the paths are the result of a power optimization. To improve timing at the expense of increased power in a subsequent implementation, use the set_power_opt XDC constraint to disable power optimization on the power-optimized cells in the critical path. Then rerun implementation with the set_power_opt XDC constraints or Tcl commands.

The following Tcl example disables power optimization on cells in the top 100 failing paths:

set pwr_critical_cells [get_cells -of [get_timing_paths -slack_lesser_than 0 -max_paths 100] -filter {IS_CLOCK_GATED}]
set_power_opt -exclude_cells $pwr_critical_cells