After routing you have 300 setup violations, 50 DRCs and SI noise failures. What order do you attack them in, and why?

Short Answer

Connectivity first, then DRCs, then timing with signal integrity included, then hold, and re-check DRC and antenna after every round of fixes. Opens and shorts make every other number meaningless. DRC fixes move wires and change parasitics, so timing fixed before DRCs is timing you will fix again.

Technical Explanation

• Start with check_lvs and check_routes for opens and shorts. An open net has no real delay and a short merges two nets, so any timing number on those nets is meaningless.
• Fix routing DRCs next. Clearing a spacing or short violation reroutes wires, and rerouting changes length, layers, vias and neighbours. Every one of those changes the parasitics timing depends on.
• Then fix setup with signal integrity turned on. If noise failures and crosstalk delta delay are part of the problem, the timing you optimize against must include them, or route_opt fixes the wrong paths.
• Treat noise failures as their own category. Glitch violations on quiet nets are fixed by upsizing victim drivers, spacing or shielding, which can also improve the setup picture on nearby paths.
• Fix hold after setup. Setup fixes, especially upsizing, speed paths up and can create hold violations; hold fixes add delay and can create small setup violations, so hold goes last with a setup re-check.
• After each wave of timing fixes, route_opt legalizes and ECO routes, which can bring DRCs or antenna violations back. Re-run check_routes each time.
• The loop ends when a full pass changes nothing: connectivity clean, DRCs at zero, setup, hold and noise passing in every scenario.

What To Check

• Opens and shorts from check_lvs -max_errors 0 and check_routes.
• DRC count and types after each step.
• Setup and hold per scenario with SI enabled, from report_qor.
• Noise violations from report_noise.
• Antenna violations after every rerouting step.

Healthy, Suspicious & Hard-stop Results

• Healthy (illustrative): Each step leaves the earlier ones clean, and the final pass changes nothing.
• Suspicious (illustrative): DRCs returning after each route_opt pass, usually in the same congested area.
• Hard stop: Timing fixed on a database with opens or shorts, or with SI turned off.

What The Interviewer Is Testing

• Whether you order by dependency: what changes the inputs to the next check.
• Do you keep SI on while fixing timing?
• A strong answer shows you plan re-checks rather than treating each category as done once.

Follow-up Question & Model Response

"Would you ever fix timing before DRCs?"

Candidate Model Response: Only if the DRCs are confined to an area that no critical path goes through, and fixing them will not reroute nets outside it. For example, a handful of violations inside a macro channel carrying only slow control signals. Even then, you re-check timing after the DRC fix. The rule exists because DRC fixes move wires; where you can show they will not move the wires that matter, the order matters less.