Signal-integrity timing
Signal-integrity timing is STA that accounts for the fact that a real on-chip signal isn't a clean, isolated pulse — it's a wiggling voltage on a wire surrounded by other switching wires, sagging power rails, and coupling capacitance. Plain STA treats each net in a vacuum; SI-aware STA adds back the messy reality: crosstalk that speeds or slows edges, supply noise that weakens drive, and glitches that can sneak through logic. It's the difference between timing a runner on an empty track versus in a crowded, jostling pack.
Practically, SI timing runs after detailed routing and parasitic extraction, when the tool finally knows which wires are physically neighbours and how strongly they couple. It iterates — because an aggressor's own timing window determines whether it can even hit a victim at the worst moment — and reports SI-adjusted slack alongside both delay degradation and noise (glitch) violations. As nodes shrink and wires crowd together, SI is no longer a corner-case afterthought but a first-class part of sign-off; ignoring it is how a chip passes 'clean' STA and still fails on silicon.
SI timing folds crosstalk and supply noise back into slack.
SI analysis depends critically on accurate parasitic extraction (the coupling caps) and on IR-drop data (the local supply voltage) — feed it optimistic parasitics or ignore power-grid sag and your SI slack is a comforting lie.