Average utilisation is good, but one region is locally impossible. How do you prove it?

Short Answer

Chip-average utilization is a useless number here — it's like saying a city has plenty of parking on average while one block is gridlocked. You have to zoom into the specific region and measure its usable area, not the die's. Usable area in that region = the physical area of standard-cell rows minus everything that eats into it: macro keepout halos, hard/soft placement blockages, voltage-area guardbands, fixed cells, and row fragments left by macro edges. Report that number with report_design/effective-utilization math, not the naive rectangle area.

Technical Explanation

• Compare that usable area against the actual demand assigned there — the cell area (and any growth margin) that hierarchy/pin assignment or partitioning pushed into that region.
• If demand > usable area, it's mathematically impossible no matter how much global whitespace exists elsewhere — logic trapped behind a macro wall, in a narrow channel, or inside an over-restrictive blockage can't borrow space from a quiet part of the die.
• Isolate one root cause at a time: is it hierarchy/pin-affinity assignment putting too many cells there, is it macro placement carving up the rows, is it row fragmentation, or is it a blockage that's wider than it needs to be?
• Fix by rebalancing: shrink or soften the blockage, move a macro, rebalance the RTL/hierarchy assignment, or grow the local voltage area — then re-derive the effective-utilization number to prove the fix actually closed the gap.

Formula Or Decision Rule

Local utilisation = counted local cell area / usable local placeable area.

What To Check

• Warning sign: The team lowers global target utilisation without opening the trapped region.
• Inspect: choose one affected region, macro, row, pin, path, or net and trace the physical cause.
• Correct: Change the local geometry or assignment, then recompute and rerun congestion.

Healthy, Suspicious & Hard-stop Results

• Healthy: Measure usable local area after rows, halos, blockages, voltage areas, and fixed objects, then compare with the cell and routing demand assigned there.
• Hard stop: required legality or physical feasibility is missing or unexplained.

What The Interviewer Is Testing

The interviewer is testing whether you can connect “Average utilisation is good, but one region is locally impossible. ” to measurable physical evidence and an owned correction.

Follow-up Question & Model Response

"* Candidate Model Response: Model response: “I would change it if the same controlled rerun shows that the team lowers global target utilisation without opening the trapped region. ”