Why must post-ECO timing be re-extracted with fresh parasitics instead of reusing the pre-ECO SPEF?

Short Answer

A SPEF file records the resistance and capacitance PrimeTime uses to compute net delay, and it describes one specific physical layout. Once an ECO adds a buffer, resizes a cell, or reroutes a net, the physical layout has changed, so the old SPEF no longer describes the real chip, and any timing report built from it is describing a design that no longer exists.

Technical Explanation

• SPEF, Standard Parasitic Exchange Format, is extracted from the actual routed layout, capturing real wire resistance and capacitance for every net.
• A cell resize or a newly inserted buffer changes the physical cells and wires on that net and its neighbors, which means the parasitics on and around it are genuinely different after the ECO, not just relabeled.
• Reusing the pre-ECO SPEF after applying size_cell or insert_buffer (PT) lets PrimeTime's timing engine keep computing delay from a physical description that predates the fix, producing a slack number that does not reflect the real, ECO'd layout.
• A physical design tool must legalize and re-route the ECO change in the actual layout before fresh parasitic extraction can even be generated, so re-extraction is a separate, necessary step after the netlist edit, not just a formality.
• Because re-extraction and re-routing both take real time, teams are sometimes tempted to sign off on the pre-extraction, PrimeTime-only estimate, acceptable as a quick sanity check, but never as the final signoff number.
• The ECO and timing signoff process on the implementation side exists specifically to close this loop: legalize the edit, re-route it, extract fresh SPEF, and hand a genuinely updated timing picture back to PrimeTime for the real signoff report.

Follow-up Question & Model Response

Could a fix that looks clean immediately after an ECO command turn out to still violate once the design is re-extracted?

Candidate Model Response: Yes, and this happens often enough that no ECO is considered closed until after re-extraction. PrimeTime's post-ECO estimate uses its own model of the new cell's load and the surrounding parasitics as they were before the physical edit, a reasonable approximation but not the real routed result. Once the physical tool legalizes the edit, it may route the new buffer's connections through a longer or more coupled path than PrimeTime assumed, changing the real delay from the estimate. This is why the fix is only considered verified after a fresh SPEF-based timing run confirms the same slack improvement holds with real, extracted parasitics.