How do you decide whether a timing violation needs an ECO fix or a full place-and-route re-run?

Short Answer

A violation is a good candidate for an ECO when it's small in magnitude, localized to a few paths, and fixable by resizing, buffering, or reconnecting a handful of cells without exceeding what spare capacity or small layout gaps can absorb. A violation calls for a full place-and-route re-run instead when it's large, widespread across many paths sharing a systemic cause -- a congested region, a poor floorplan decision, an under-resourced clock tree -- because an ECO can only patch individual paths one at a time and cannot fix the systemic condition creating many of them at once.

Technical Explanation

The first signal to check is scale: a handful of paths with a few tens of picoseconds of violation each is squarely within what fix_eco_timing (PT) is built to close; hundreds or thousands of violations, or ones with hundreds of picoseconds each, usually indicate a structural problem an ECO cannot patch path by path in reasonable time.

• The second signal is root-cause concentration: if many violating paths trace back to one shared cause -- a congested placement region, an unbalanced clock tree, a corner assumption that changed globally -- fixing them individually treats symptoms while the shared cause remains, often reappearing as new violations elsewhere.
• Grouping violations by shared physical or logical cause, rather than one path at a time, is the practical way to tell the two situations apart: if grouping collapses a thousand violations into three or four root causes, each is likely fixable at its source, which may itself be an ECO, rather than a full re-run.
• Schedule and risk also weigh in: an ECO is faster and touches less of the design, which matters most close to tapeout, but a full re-run gives placement and routing algorithms freedom to solve the underlying problem properly when schedule allows it.
• A hybrid approach is common: fix the isolated violations with a scoped ECO, and address the systemic contributor with a targeted incremental place-and-route pass limited to that specific area, rather than either a blanket ECO attempt or a complete re-run.
• The decision has to account for remaining schedule, since an ECO that turns out insufficient wastes time a re-run would have used productively, while an unnecessary re-run wastes schedule a scoped ECO would have preserved.

Follow-up Question & Model Response

If violations are grouped by root cause and one group turns out to be a clock-tree imbalance affecting 200 paths, is that still fixable with an ECO?

Candidate Model Response: It can be, but the fix should target the shared cause rather than the 200 individual paths -- resizing or adding delay to the specific clock buffers responsible corrects timing for every downstream path fed by that buffer in one change instead of 200 separate path-level edits. This works because the 200 violations are a symptom of one structural cause, not 200 independent problems, so fixing the cause once is faster and more consistent. Whether this stays ECO territory or edges into a small, targeted CTS re-run depends on how much of the clock tree needs to change -- a single buffer resize is squarely ECO territory, while rebalancing several branches looks more like a scoped CTS re-run. The grouping step reveals which of those two a given case actually is, before any fix is attempted.