What does fix_eco_timing do by default for setup vs hold?

Short Answer

fix_eco_timing (PT) will not run without -type setup or -type hold, because the two problems need opposite changes. By default, setup fixing uses cell sizing alone to cut data path delay, and hold fixing uses sizing plus buffer insertion to add delay. Both work only on data paths, not clock networks. Setup fixing avoids new DRC violations but may create hold violations, while hold fixing avoids creating either setup or DRC violations.

Technical Explanation

• The type is mandatory on every call because setup needs less delay and hold needs more. There is no combined mode.
• The default methods differ. Setup uses cell sizing only; hold uses sizing and buffer insertion. To buffer for setup, add -methods insert_buffer with -buffer_list. Side-load sizing (-methods size_cell_side_load) works only with -type setup and -cell_type combinational.
• Precedence is asymmetric. Setup fixing honours DRC but may alter hold slack, because the PT UG treats setup as the harder problem. Hold fixing honours both setup slack and DRC.
• Only data paths change by default. Clock-network fixing with -cell_type clock_network needs the physically aware flow, and TNS-driven targeting with -target_violation_type tns also needs a PrimeTime-ADV license.
• By default the command does not fix timing on paths that carry DRC violations. fix_eco_timing -ignore_drc (PT) reverses that and may degrade max transition, max capacitance and max fanout.
• It iterates from the worst violation until everything is fixed or further work is not worth the runtime, minimizing area and power. It then prints the remaining and unfixable violations; -verbose or -estimate_unfixable_reasons explains why, once eco_report_unfixed_reason_max_endpoints (PT) sets how many to report.
• You can fix against recalculated slack with -pba_mode path or -pba_mode exhaustive, which avoids spending cells on graph-based pessimism.
• Before any ECO command, set timing_save_pin_arrival_and_slack (PT) to true and run a full timing update, as the ECO setup steps require.

What To Check

• The -type used and the method list reported in the log.
• The Final ECO Summary: number of size_cell and insert_buffer commands and the area increase.
• Unfixable reason codes, for example W (fixing might degrade DRC) or B (delay improvement too small).
• Hold after the setup pass, and setup after the hold pass.

Healthy, Suspicious & Hard-stop Results

• Healthy (illustrative): A setup pass made only of size_cell changes with a small area increase and no new DRC violations.
• Suspicious (illustrative): Setup clears but hold violations jump from 60 to 150 because upsized cells sped up short paths.
• Hard stop: Paths left unfixed with reason W on every stage. DRC violations are blocking timing fixing and must be fixed first.

What The Interviewer Is Testing

• Knowing the default method sets and precedence rules without looking them up.
• Can you explain why setup may spend hold margin but hold may not spend setup?

Follow-up Question & Model Response

"Why may setup fixing create hold violations when hold fixing may not create setup violations?"

Candidate Model Response: The PT UG says it directly: setup is the harder problem. A setup fix needs delay removed, and there are only so many faster cells and shorter wires. A hold fix needs delay added, which is almost always possible somewhere on the path with a buffer or load cell. So the tool protects the scarce resource, setup slack, and lets hold absorb the cost. That is also why setup runs before hold in the recommended order.