How does LVF (Liberty Variation Format) distance-based derating let a library model POCV variation without needing a separate AOCV side file?
From PDVerse STA Mentor Guide ยท pdVerse Mentor Guide
Short Answer
LVF lets a cell's own Liberty timing arcs carry distance-based derating factors directly in the library, alongside the ordinary POCV sigma tables, so the tool can look up a derate that depends on how far apart two timing points are in the clock network without loading a separate AOCV side file at all. Because the data lives in the library and can vary with slew and load like any other LVF table, it is functionally equivalent to a side file's distance-based table, but automatically tied to the cell characterization it came from.
Technical Explanation
- POCV normally uses LVF's
ocv_std_dev_cell_riseand similar tables to model the standard deviation of a delay arc's variation, letting the tool compute a per-arc sigma instead of a single flat derate percentage. - LVF distance-based derating extends this by letting the library provide the same kind of standard-deviation data as a function of distance between two points in the design, not just as a function of slew and load.
- This distance-dependent variation data is arc-dependent and can be modeled for delay timing arcs as well as setup and hold constraints โ data of this shape cannot be represented in a side file, only directly in an LVF library.
- The unit of POCV LVF distance data is the library's own time unit, so the tool applies it consistently with every other timing number already coming from the same characterization, without a unit-conversion step a hand-built side file would need.
- Functionally, LVF distance-based derating produces the same kind of result a hand-authored AOCV side file would โ less pessimism for cells or paths that are physically close together, more for cells that are far apart โ but the data originates from the same silicon characterization as the rest of the library instead of a separately-maintained file.
- Because the data lives inside the library, updating the technology's process characterization automatically updates the distance-based derating along with every other timing number, whereas a side file has to be regenerated and requalified on its own schedule.
- What breaks: mixing an old AOCV side file with a newer library that also carries LVF distance-based data risks double-applying distance-based derating from two independent sources, unless the flow explicitly chooses one source and disables the other.
Common Mistake
The Trap: Continuing to load a legacy AOCV side file alongside a newer library that already carries LVF distance-based derating, on the assumption that more derating data can only be more accurate.
- Both sources apply a distance-dependent adjustment to the same arcs, and unless the flow explicitly disables one, the effective derate is a compounded, over-conservative combination of the two, not simply "extra safety margin."
- The resulting slack numbers are pessimistic in a way that looks plausible, since nothing crashes or errors, so the double-derating can persist across several signoff cycles before anyone questions why margins look unusually tight compared to a sister design using the library alone.
Follow-up Question & Model Response
"If a library now provides LVF distance-based data, is there ever a reason to still keep the old AOCV side file active alongside it?"
Candidate Model Response: Generally no, once the library's own distance-based LVF data is validated to cover the same cells and arcs the side file used to cover โ carrying both is redundant and risks the double-derating described above. A side file might still earn a temporary place if it covers a specific structure the LVF characterization does not yet model, such as a newly added custom cell not yet re-characterized, but that should be a deliberate, scoped exception with the LVF data disabled for exactly those cells, not a blanket "keep both for safety" policy. Confirming coverage with report_timing_derate (PT) before deciding is the safest way to know exactly which arcs each source is touching.
Practical Example
A 7nm library update adds LVF distance-based derating covering every standard cell arc, replacing a legacy AOCV side file the design had used for three prior tapeouts. A team that forgot to retire the old side file saw setup slack on long, physically-spread reg-to-reg paths drop by an unexplained 20-30 ps compared to the previous tapeout on an otherwise unchanged block. Disabling the legacy side file and relying solely on the library's own LVF distance data restored the expected slack, confirming the two sources had been stacking derate on the same arcs.
Complete STA Handbook
Master Signoff-Ready Static Timing Analysis
Get the complete 10-chapter STA handbook covering setup/hold margins, clock modeling, OCV/POCV, crosstalk noise, and PrimeTime closure.

Continue practising