Why might ICC2 report clean timing while PrimeTime signoff on the same design shows failures?
From PDVerse MMMC Interview Masterclass · pdVerse Mentor Series
Direct answer
This is almost never "the signoff tool is being pessimistic." It's much more likely a correlation gap — a mismatch somewhere between what ICC2 analyzed and what PrimeTime analyzed: netlist revision, SDC, libraries, operating condition, parasitics, SI settings, OCV/AOCV/POCV treatment, CPPR, or whether one side used graph-based analysis and the other used path-based.
Mentor explanation
It's tempting, especially under deadline pressure, to blame the signoff tool for being "too conservative" or to blame the implementation tool for being "wrong." Neither framing is productive, and both are usually incorrect. The professional response is to treat this as a correlation investigation: freeze both environments, and compare them input by input — same netlist hash, same SDC, same library revision, same parasitics, same variation methodology, same analysis mode. Almost every real ICC2-versus-PrimeTime discrepancy traces back to one of these inputs quietly differing, not to a tool defect.
Interview trap
Saying "PrimeTime is being pessimistic" or "ICC2 is wrong" as a first response. That's a conclusion, not an investigation.
Tool check
report_analysis_coverage; report_annotated_parasitics -check; report_timing -delay_type max/minCompare signoff inputs and analysis with equivalent ICC2 reports.
Key takeaways
- This is almost never "the signoff tool is being pessimistic." It's much more likely a correlation gap — a mismatch somewhere between what ICC2 analyzed and what PrimeTime analyzed: netlist revision, SDC, libraries, operating condition, parasitics, SI settings, OCV/AOCV/POCV treatment, CPPR, or whether one side used graph-based analysis and the other used path-based.
Self-check: can you answer this aloud?
Try a 45-second answer using this structure:
- State the direct answer.
- Explain the timing or physical reason.
- Name one caveat.
- Say how you would verify it in a real flow.
Continue learning free
Get a practical low-power chapter
Receive the existing “Low Power and Multivoltage Fundamentals” PDF chapter and its download link by email.
Keep connecting concepts