Design context and scope
This lab note reworks the original placement article into a debug-oriented sequence. It keeps the same scope (stages, algorithms, optimization, and validation) while focusing on the practical questions asked during repeated placement runs.
Coverage map for this lab note
- Placement goals and strategic priorities: timing, routability, density balance, and legal manufacturable placement.
- Inputs required by placement and the pre-placement checks that prevent misleading optimization results.
- Special and pre-placed cells (macros, endcaps, tap cells, and related structures) and why they shape the search space.
- Global placement as coarse optimization for wirelength, timing, and routability before exact legal coordinates.
- Legalization stage and the rules it enforces while preserving as much QoR as possible.
- Detailed placement for local improvements, spacing compliance, and timing and routability refinement.
- Placement optimization toolkit (buffering, resizing, HFNS, restructuring) and pre-CTS tradeoffs.
- Post-placement qualification using congestion, timing, density, and report-based review.
- How placement quality affects CTS and routing closure, especially at advanced nodes.
What you should be able to explain
- Describe the purpose and expected outputs of global placement, legalization, and detailed placement.
- Interpret placement reports to separate timing problems from congestion and floorplan root causes.
- Build a run-to-run comparison method using a small set of consistent placement metrics.
- Explain why a placement that looks timing-good can still fail in CTS or routing.
Review checklist before moving ahead
- Verify pre-placement setup (libraries, constraints, floorplan, special cells) before comparing QoR.
- Review congestion map, density hotspots, and worst path groups after global placement.
- Check legalization side effects on displacement and local density.
- Inspect pre-CTS optimization actions (HFNS, buffering, resizing) for intended impact.
- Record placement risks that must be tracked in CTS and routing, not just current WNS or TNS.
Common watchouts
- Do not compare placement runs using only WNS; congestion and density often predict later failures better.
- Legalization fixes can silently degrade timing or increase route pressure if not reviewed.
- HFNS and buffering can solve one metric while increasing power or CTS complexity.
Self-check prompts
- What metrics do you use to decide whether a placement run is worth continuing?
- When does a congestion issue point back to floorplanning instead of placement settings?
- How would you explain legalization to someone who only looks at timing reports?