Why is a per-cell IR-drop annotation treated as uncorrelated with the main rail, and what's the consequence?
From PDVerse PrimeTime STA Interview Guide · pdVerse Mentor Guide
Ten-second Interview Answer
Because a local IR drop is a physically independent droop - one instance's grid sagging says nothing about another's - so PrimeTime compares the full worst-case difference across the annotation instead of assuming driver and load track together.
Complete Technical Explanation
Correlation is a physical claim, not a convenience. When two cells share a UPF supply net, their voltages genuinely move together and PrimeTime can safely ignore the difference. A per-cell IR-drop annotation is the opposite situation: it describes a local droop on one instance's grid, and there is no physical reason U3's voltage sagging implies that U1's does. The two values are independent, so for signal-level checking the tool deliberately compares the full worst-case difference across the annotation. Take a crossing where the annotation produces a driver at 0.8 against a load at 1.1 in one direction and 1.2 against 0.7 in the other: both comparisons are taken at their worst, and both are reported as violations. The alternative - assuming driver and load track each other - would be optimistic in exactly the case the annotation was created to model, and would hide a real electrical mismatch between two cells that can, in fact, be at different voltages at the same instant. The consequence for you is that annotating IR drop can legitimately introduce signal-level violations on crossings that were clean before, and those violations are not artifacts to be filtered away. Contrast the two rules once more: a shared UPF net is correlated and stays quiet, a per-cell annotation is uncorrelated and speaks up.
In Pd / Signoff
This shapes how you read signal-level results after annotating IR drop with set_voltage -cell -pg_pin_name . New violations on previously clean domain crossings are the expected behaviour, and each one is a real worst-case pairing to assess - either the grid needs work at that instance or the crossing needs a level shifter, not a threshold adjustment to make the report quiet.
Common Trap
Treating violations that appear only after IR-drop annotation as pessimism and widening thresholds to suppress them. The worst-case comparison is intentional, because the two cells' droops are independent; suppressing it hides a mismatch that can genuinely occur.
Expect Next
Why does the same nominal voltage difference stay quiet when both cells share one UPF supply net?
Topic And Primetime Commands
Multivoltage / UPF & Power Domains. set_voltage
Keep building interview depth
Continue practising STA
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.
Continue practising