When would you run route_global, route_track and route_detail separately instead of route_auto?

Short Answer

Run the steps separately when you want to look at the result of global routing before paying for detail routing. route_auto runs all three engines back to back, which is fine on a stable block. On a new or congested block, stopping after route_global lets you read the congestion map and fix a hot spot in minutes instead of discovering it hours later as thousands of DRCs.

Technical Explanation

• route_auto runs global routing, track assignment and detail routing in one command. It runs the same three engines as the separate commands.
• Global routing is the cheap part. Detail routing is where most of the runtime goes, often many times longer than global routing on a large block.
• Stopping after route_global gives you the overflow report and the congestion map from real global routes. If one channel is badly overflowing, you can change placement, add a routing guide or adjust layer usage before any detail routing time is spent.
• For an even cheaper look, route_global -congestion_map_only true builds the congestion map without keeping any global routes. This is what report_congestion falls back to when no map exists.
• Running route_track on its own is less common but useful for checking wire length and via counts, or for timing and crosstalk-driven experiments on track assignment.
• Once a block has been through a few iterations and congestion is known to be fine, route_auto is simpler to maintain in scripts and runs the same engines.

What To Check

• Overflow reported by route_global, per layer and in total.
• The congestion map in the GUI, looking for clusters rather than scattered single cells.
• Wire length and via count summary after route_track.
• DRC count per detail routing iteration, and whether it is still falling.

Healthy, Suspicious & Hard-stop Results

• Healthy (illustrative): Global route overflow at 0 on every layer, so the flow continues straight into track assignment and detail routing.
• Suspicious (illustrative): Overflow below 0.5% of global routing cells, spread thinly; worth one look at the map, then continue.
• Hard stop: Overflow concentrated in one channel or next to one macro. Stop, fix placement or add a routing guide, and rerun route_global before any detail routing.

What The Interviewer Is Testing

• Whether you understand where routing runtime is spent.
• Do you use global routing as a decision point rather than a formality?
• A strong answer shows you know the congestion-map-only option exists and when it is enough.

Follow-up Question & Model Response

"Is the congestion map from -congestion_map_only true as good as the one from a full route_global?"

Candidate Model Response: It is built by the same global router, so the overflow picture is the same kind of data, but no global routes are kept afterwards. That is fine when you only want to know where the crowding is. If you want to continue into track assignment, or run timing on global route parasitics, you need real global routes. So use the map-only run for quick what-if checks and a real route_global when you intend to carry on.