What do positive slack and negative slack mean?
From PDVerse STA Mentor Guide, part of the pdVerse Mentor Guide
Short Answer
Positive slack means a path met its timing requirement with margin left over — the design is safe on that check. Negative slack means the path missed its requirement, which is a real timing violation that will cause incorrect behavior in silicon unless it gets fixed before tapeout.
Technical Explanation
Slack's sign is the pass/fail line the tool draws for every single path.
- Positive slack is safety margin, not waste — up to a point. It absorbs things the tool's model doesn't capture perfectly: unmodeled noise, temperature variation across the die, and gradual aging of the transistors over the product's lifetime.
- Very large positive slack is a power-saving opportunity, not just a comfort blanket. A path with hundreds of picoseconds to spare can often have its cells downsized or switched to a higher-threshold, lower-leakage variant, trading away unused speed for lower static power.
- Zero slack is a legitimate pass, not a warning sign by itself — the path meets the requirement exactly, with none of that model-uncertainty margin. Whether that's acceptable depends on how much extra margin the signoff methodology already built into the corners and derates.
- Negative setup slack means the logic is too slow for its clock. The circuit would capture the wrong value at the intended frequency.
- Negative hold slack means the logic is too fast for its own clock skew. New data overwrites the value a flop was still in the middle of capturing — a race failure that has nothing to do with frequency.
Common Mistake
- The trap: dismissing a small negative slack, like −5 ps, as "basically noise" that doesn't matter.
- The reasoning feels sound because the number is tiny compared to a nanosecond-scale clock period.
- In automated signoff, any negative slack at all — even −1 ps — fails the tapeout gate. There's no informal tolerance band; a violation is a violation until it's formally fixed or waived by the methodology.
Follow-up Question & Model Response
Can having too much positive hold slack ever be a problem on its own?
Candidate Model Response: Yes, in a roundabout way. Very large positive hold slack usually means the tool inserted more delay buffers on that path than strictly necessary to clear a hold violation. Each of those buffers costs silicon area and burns extra dynamic power every time it switches, for no timing benefit once the violation is already cleared. Reviewing paths with unusually large positive hold slack is a normal way to find buffer insertion that can be trimmed back.
Practical Example
A leakage-recovery pass targeted every path with more than +150 ps of setup slack to spare and swapped roughly 85,000 gates from a low-threshold library to a higher-threshold, lower-leakage one. Static leakage power dropped by about 28% chip-wide, and because every swapped path still kept positive slack afterward, no new timing violations were introduced.
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.
Offline PDF Bundle
Want all 1109 questions offline?
Get the complete 4-book PDF bundle (PnR, STA, MMMC, Low Power) with a clickable table of contents - no ads, no internet needed.
Continue practising