Beyond the basic congestion-driven effort setting, what other specific ICC2 app-options target congestion reduction, and what does each one specifically address?

Short Answer

Several targeted options exist beyond the general effort knob: congestion_layer_aware (per-layer congestion instead of combined layers), increased_cell_expansion (expand virtual cell area per local routing need), congestion_expansion_direction=both (default is horizontal only), ndr_area_aware (accounts for clock NDR area specifically), seq_array_icg_aware (reduces congestion from sequential-array clock nets), spread_repeater_paths (spreads repeaters orthogonally, avoiding clumping at macro/blockage edges), and wide_cell_use_model (wide-cell density modeling for advanced nodes).

Technical Explanation

• congestion_layer_aware models congestion PER LAYER instead of combining all layers into one congestion estimate -- more precise, since different layers can have genuinely different congestion profiles.
• increased_cell_expansion expands the virtual cell area used for congestion estimation based on actual local routing need, rather than a fixed expansion.
• congestion_expansion_direction=both expands congestion consideration in BOTH directions -- the default is horizontal only, so this is a real behavioral change, not a minor tweak.
• ndr_area_aware specifically accounts for clock NDR (non-default rule) area in congestion estimation -- clock nets with wider NDRs consume more routing resource than a naive estimate would assume.
• seq_array_icg_aware reduces congestion specifically arising from sequential-array clock nets -- a targeted fix for a specific congestion source pattern.
• spread_repeater_paths spreads repeaters ORTHOGONALLY, specifically to avoid clumping at macro or blockage edges -- a real, documented failure pattern this option targets.
• wide_cell_use_model provides wide-cell density modeling specifically for advanced nodes, where a cell too wide to straddle vertical power straps (wider than half the strap pitch) can only have ONE such cell placed between straps -- a real physical constraint this option accounts for.

Formula Or Decision Rule

Wide-cell rationale: if a cell's width exceeds half the vertical power strap pitch and it cannot straddle straps, only ONE such cell fits between adjacent straps -- a hard physical placement constraint at advanced nodes.

What To Check

• Warning sign: general congestion-effort tuning doesn't resolve a persistent, specific congestion pattern (e.g. concentrated near macro edges, or from sequential-array clock nets).
• Inspect: identify the specific pattern and check whether a targeted option (spread_repeater_paths for edge clumping, seq_array_icg_aware for sequential-array clocks) matches it.
• Correct: enable the specific targeted option rather than continuing to raise the general congestion effort, which may not address a pattern-specific root cause.

Healthy, Suspicious & Hard-stop Results

• Expected: a specific, identified congestion pattern is addressed with the correspondingly specific targeted option, not just general effort tuning.
• Investigate: general congestion effort has been raised repeatedly with limited improvement on a specific, identifiable congestion pattern -- worth checking whether a targeted option matches that pattern instead.
• Stop: repeated general-effort tuning continues with no diagnosis of the actual congestion pattern's root cause, when several specific, documented options exist for exactly the common patterns (macro-edge clumping, sequential-array clocks, wide-cell strap constraints).

What The Interviewer Is Testing

Whether you know several of these SPECIFIC, targeted congestion options (not just the general effort knob) and can match a real congestion symptom to the right targeted fix.