What does set_dont_touch_network actually protect by default, and how does -clock_only narrow that scope?

Short Answer

By default, set_dont_touch_network [get_clocks CLK] protects BOTH clock paths AND clock-as-data paths -- meaning a clock signal used somewhere as an ordinary data input is also protected. -clock_only narrows that to just the clock paths, leaving clock-as-data usage open to optimization. -clear removes the protection entirely.

Technical Explanation

• By default, set_dont_touch_network [get_clocks CLK] protects BOTH clock paths AND clock-as-data paths -- a broader scope than just "the clock tree."
• Clock-as-data paths means a clock signal used somewhere in the design as an ordinary data input, not as a clock -- the default behavior protects that usage too.
• -clock_only narrows the protection to just the genuine clock paths, leaving any clock-as-data usage open to normal optimization.
• -clear removes the dont_touch_network protection entirely, returning the clock network to normal optimization eligibility.

Follow-up Question & Model Response

"Why might a design legitimately want to optimize a clock-as-data path even while keeping the true clock network protected?"

Candidate Model Response: Because a clock signal used as ordinary data (e.g. feeding a frequency-detection circuit) is functionally just a data signal at that point, and forcing it to stay unoptimized purely because its SOURCE happens to be a clock can leave real timing or area improvement unclaimed on that specific path.