Cell vs Net Delay Budgets: Debug Playbook

Debug playbook

Debug Playbook for Cell vs Net Delay Budgets focuses on cell delay %, net delay %, high fanout net count, transition violations. The goal is to connect the observable symptom to timing mechanism, ownership, and regression risk.

STA debug is a search for the FIRST deviation, not the loudest downstream symptom. Hold failures after setup ECO are a classic example of fixing the wrong layer first.

Root-cause tree

ROOT-CAUSE TREE — Cell vs Net Delay Budgets

cell delay %, net delay %, high fanout net count, transition violations looks wrong
        |
   same netlist tag?
     /        \
   no          yes
   |            |
 view/SDC     real path or
 mismatch     constraint bug
 /    \           |
lib   SPEF    delay breakdown
diff  diff    cell vs net

1. Freeze the failing netlist tag, SDC revision, and SPEF corner.

2. Find the first constraint or delay anomaly on the critical path.

3. Map the path to launch/capture clocks and path group.

4. Classify the failure: constraint bug, view mismatch, cell delay, net delay, or SI.

5. Prove the mechanism with one reduced path or isolated scenario.

6. Patch the smallest owner-controlled fix and rerun setup, hold, and MMMC.

Review memo template

STAFF STA REVIEW MEMO — Delay Graph & Path Analysis / Cell vs Net Delay Budgets

1. Symptom
   - Watched metric: cell delay %, net delay %, high fanout net count, transition violations
   - Failing scenario: <corner/mode/view>
   - Path group: <reg2reg / in2reg / reg2out / async>
   - Database tags: <netlist, SDC, SPEF, lib>

2. Mechanism hypothesis
   - Primary mechanism: Path delay splits between library cell arcs and interconnect parasitics; implementation fixes target the dominant contributor.
   - Competing hypothesis: <constraint bug, view mismatch, cell delay, net delay, SI>
   - Missing evidence: <report_timing, clock report, constraint trace>

3. Proposed action
   - Minimal reversible change: <SDC fix, sizing, VT swap, route, useful skew>
   - Expected metric movement: <WNS/TNS delta>
   - Regression risk: hold, other corners, SI, leakage

4. Signoff
   - Re-run artifact: report_timing -cap -tran -nets, delay breakdown report, RC extraction log
   - Required owners: PD owner, STA owner, extraction owner
   - Final decision: ECO, waive, constraint fix, or escalate

STA deep dive

Slack is a path story — read the graph, not just the number.

Concept diagram

TIMING GRAPH

FF --cell-- logic --net-- logic --cell-- FF
launch                              capture

Metric graph

SLACK TAIL

paths |    ***
      |  *******
      |*********  <- fix WNS first

Reports and artifacts

• WNS/TNS

• path slack histogram

• cell vs net %

• path group summary

Mini case study

WNS -90ps but 80% net delay on one congested route — PD route fix beat upsizing three cells.

Debug branches

• Cell-heavy -> sizing/VT

• Net-heavy -> route/layer

• Check launch/capture clocks

Senior review question

Ask: what corner/mode/view proves this topic is closed or failing?

Key takeaways

• State corner, mode, view, and database tag with every slack claim.

• Run setup and hold plus MMMC regression after every ECO.

Common pitfalls

• Comparing STA runs with different SPEF or SDC tags.

• Broad false_path to green-wash violations.

• Setup-only ECO without hold check.

Principal STA review addendum

Path delay splits between library cell arcs and interconnect parasitics; implementation fixes target the dominant contributor.

Metric: cell delay %, net delay %, high fanout net count, transition violations