IntermediateQuestion 151 of 162Source: Synopsys ICC2 Implementation User Guide: Routing

Single-layer vs accumulated antenna ratio: which is stricter, and why?

From PDVerse PnR Interview Handbook · pdVerse Mentor Guide

By Tabish Iqbal · 9+ Years VLSI Physical Design Experience

Short Answer

Accumulated modes are stricter than single-layer, because they count metal on the current layer plus every layer below it, while single-layer counts only the current layer. ICC2 has six antenna modes, combining how area is measured (surface or sidewall) with how layers are counted (single-layer, accumulated-ratio or accumulated-area). The foundry rule decides which mode you must use.

Table of antenna modes 1 to 6 combining surface or sidewall area with single-layer, accumulated-ratio or accumulated-area recognition, beside a layer stack showing which layers count in each mode. — Technical Reference DiagramSingle-layer vs accumulated antenna ratio: which is stricter, and why?

Technical Explanation

The antenna check compares the metal connected to a gate with the gate area, and the result must stay below the maximum ratio in the rule.
Surface area mode measures metal as W × L. Sidewall area mode measures (W + L) × 2 × thickness and needs the layer thickness attributes in the technology file.
Single-layer recognition counts only the metal on the layer being checked. It is the most forgiving and gives the best routability.
Accumulated-ratio adds up the single-layer ratios of the current layer and all layers below. Accumulated-area adds up the metal area of the current and lower layers, then divides by gate area. Both count more metal than single-layer, so both are stricter.
A global rule is required with define_antenna_rule, which needs -mode and -diode_mode. Layer rules with define_antenna_layer_rule are optional and override the global rule for that layer.
The diode mode says how protection diodes relax the limit. There are many diode modes, for example mode 0 for no protection and mode 1 for unlimited protection.
Layer hopping works less well against accumulated modes, because lower-layer metal still counts after the hop. Diodes are more often needed.

Formula Or Decision Rule

The check: antenna_area / gate_area < max_antenna_ratio for each input pin.
Single-layer: ratio = metal area on the current layer connected to the gate / gate area.
Accumulated-ratio: ratio = sum of the single-layer ratios of the current layer and every layer below.
Accumulated-area: ratio = metal area on the current and all lower layers / gate area.
Mode number: 1-3 use surface area (W × L), 4-6 use sidewall area ((W + L) × 2 × thickness); within each group, 1 or 4 is single-layer, 2 or 5 accumulated-ratio, 3 or 6 accumulated-area.

What To Check

Which mode the foundry rule specifies, per layer if layer rules exist.
That sidewall modes have the thickness attributes defined.
Antenna violations by layer from check_routes -antenna true.
Whether violations cluster on the upper layers, where accumulation is largest.

Command Checks & Actions

ICC2define_antenna_rule -mode 1 -diode_mode 2 -metal_ratio 400 -cut_ratio 20

Defines the required global rule: surface, single-layer mode with limited diode protection (illustrative ratios).

ICC2define_antenna_layer_rule -mode 1 -layer M4 -ratio 1000 -diode_ratio {0 0 1 0 0}

Adds a layer-specific ratio for M4 under the same mode as the global rule (illustrative values).

ICC2check_routes -antenna true

Runs the routing checks, including antenna against the defined rules (antenna checking is on by default).

Healthy, Suspicious & Hard-stop Results

Healthy (illustrative): The rules match the foundry deck exactly and routing ends with zero antenna violations.
Suspicious (illustrative): Violations only on top layers with accumulated modes, typically needing diodes rather than hops.
Hard stop: A mode that does not match the foundry rule, which makes both the check and the fixes meaningless.

Common Mistake

The Trap: Using single-layer mode because it gives fewer violations, when the foundry rule is accumulated-area. Routing looks clean, but the signoff antenna deck, which counts all lower layers, reports hundreds of violations at the end.

What The Interviewer Is Testing

Whether you can explain why accumulated counting is stricter.
Do you know the mode is dictated by the foundry, not chosen for convenience?
A strong answer shows you connect the mode to which fix will work.

Follow-up Question & Model Response

"Why does layer hopping help less under an accumulated mode?"

Candidate Model Response: Layer hopping works by reducing how much metal is attached to the gate on the layer being etched. Under single-layer counting that is all that matters. Under accumulated counting, metal on the lower layers still counts when the upper layer is checked, so moving part of a wire up a layer does not remove it from the total. The hop reduces the ratio less, and a diode, which gives the charge somewhere to go, is often the better fix.

Practical Example

Tapeout Scenario: A gate of 0.1 um² connects through 3 um² of M1, 6 um² of M2 and 12 um² of M3 before the driver joins on M4 (illustrative). In single-layer mode, the M3 check sees 12/0.1 = 120. In accumulated-area mode, it sees (3 + 6 + 12)/0.1 = 210. With a limit of 200, the net passes single-layer and fails accumulated-area. A hop on M3 near the gate cuts the M3 share to 1 um², giving (3 + 6 + 1)/0.1 = 100 in accumulated-area mode, which passes. Under accumulated-ratio the per-layer ratios 30 + 60 + 10 give the same 100, so the fix works for both.

PnR Flow Mentor Guide

Master the Physical Design Implementation Flow

Read the complete 8-chapter PnR Flow Mentor Guide free on the web — library setup through placement, clock tree synthesis, routing, chip finishing, hierarchical implementation, and ECO, all the way to stream-out.

Explore PnR Flow Mentor Guide →