Back to Interview Prep

Physical Design Interview Questions - Part 3: Power & Signoff

May 28, 2026 HDL2Chips Team Physical Design

This part covers power analysis, optimization techniques, and final signoff checks before tape-out.

Q21 What is IR drop and why is it important?

IR drop is the voltage drop along the power distribution network due to wire resistance (V = I × R). Excessive IR drop reduces cell drive strength and increases delay, causing timing failures. Static IR drop is average current draw; dynamic IR drop captures peak current events and is more severe.

Q22 What is electromigration and how is it managed?

Electromigration is the gradual displacement of metal atoms due to high current density, eventually causing open circuits or shorts. Foundries provide current density limits per metal layer. Managed by: widening high-current wires, using multiple vias, limiting wire length, and running EM analysis at signoff.

Q23 What is power gating and what are its challenges?

Power gating uses sleep transistors (header/footer) to shut off power to idle blocks, reducing leakage power. Challenges: wake-up time (inrush current spike), state retention (isolation cells needed), power switch sizing, and dynamic IR drop during wake-up.

Q24 What are multi-Vt cells and how are they used for leakage optimization?

LVT (low-Vt): fast, high leakage — used on timing-critical paths. SVT (standard-Vt): balanced speed/leakage. HVT (high-Vt): slow, low leakage — used on non-critical paths. Multi-Vt optimization replaces cells on non-critical paths with HVT to reduce leakage without impacting timing.

Q25 What is the difference between pre-CTS and post-CTS timing analysis?

Pre-CTS timing uses ideal clocks (zero skew, estimated latency) and wire load models. Post-CTS timing uses propagated clocks (actual buffer delays, real skew), extracted RC parasitics, and provides accurate timing. Post-CTS results always differ from pre-CTS.

Q26 What is the difference between flat and hierarchical physical design?

Flat PD implements the entire design as one block — simpler but limited by tool capacity (~10M gates). Hierarchical PD partitions the design into blocks implemented independently with boundary constraints — scalable for large SoCs but requires more planning for timing and power at top level.

Q27 What is GDSII and what does it contain?

GDSII (Graphic Design System II) is the standard binary format for IC layout data sent to foundry. It contains: geometric shapes (polygons, paths, circles), layer numbers, structures/cells, and placement coordinates. The final GDSII is verified through DRC/LVS before tape-out.

Q28 What are the goals of power grid design?

The power grid must: deliver uniform VDD/VSS across the chip, meet IR drop targets (<3-5% of VDD), handle peak current without electromigration, minimize routing resource usage, and provide adequate decoupling capacitance. Grid analysis is iterative during PD.

Q29 What is multi-voltage design and what extra cells are needed?

Multi-voltage design uses multiple voltage domains to save power. Extra cells needed: level shifters (convert between voltage domains), isolation cells (block X-states when a domain is off), and always-on cells/buffers for power management logic.

Q30 What is included in the timing signoff checklist?

Timing signoff checks: all setup/hold violations fixed (positive slack), no false path violations, clock skew within limits, no clock gating check violations, timing across all PVT corners passes, no asynchronous clock domain violations, and STA reports clean with all constraints validated.

Power and signoff knowledge is critical for tape-out success. Best of luck!