A feedthrough improves top-level timing but harms the block interface. How do you decide?
From PDVerse PnR Interview Handbook · pdVerse Mentor Guide
Short Answer
Compare top-level delay saved with block pin count, alignment, route capacity, buffering, voltage-area compatibility, reuse, and ECO flexibility. The system owner and block owner must agree.
Technical Explanation
A faster shortcut can still be a bad interface contract. Test a detour, pure feedthrough, buffered feedthrough, or reused path with the same timing and congestion method. Identify which resource becomes limiting.
Formula Or Decision Rule
Decision rule: choose the lower total project risk, not the shortest path in isolation.
What To Check
• Warning sign: The top level claims the feedthrough without block evidence. • Inspect: choose one affected region, macro, row, pin, path, or net and trace the physical cause. • Correct: Document both alternatives, assign ownership, implement the chosen constraint, and rerun block and top-level checks.
Command Checks & Actions
• check_feedthroughs -include_original_feedthroughs -include_buffered -pure -mixed: checks targeted feedthrough classes and constraints
• report_feedthroughs -reporting_style block_based: reports feedthrough nets and traversed blocks
Run only the commands needed for this question. Save the report with the floorplan version and analysis context.Healthy, Suspicious & Hard-stop Results
• Healthy: Compare top-level delay saved with block pin count, alignment, route capacity, buffering, voltage-area compatibility, reuse, and ECO flexibility. The system owner and block owner must agree. • Suspicious: The top level claims the feedthrough without block evidence. • Hard stop: required legality or physical feasibility is missing or unexplained. Document both alternatives, assign ownership, implement the chosen constraint, and rerun block and top-level checks.
Common Mistake
Do not hide the symptom with an arbitrary utilisation, halo, channel, blockage, pin move, or die-size change. The top level claims the feedthrough without block evidence.
What The Interviewer Is Testing
The interviewer is testing whether you can connect “A feedthrough improves top-level timing but harms the block interface. How do you decide?” to measurable physical evidence and an owned correction.
Follow-up Question & Model Response
Follow-up probe: What would make you change your decision? Model response: “I would change it if the same controlled rerun shows that the top level claims the feedthrough without block evidence. I would use the named objects and reports to decide whether to iterate, waive, or return the issue.”
Practical Example
Symptom: The top level claims the feedthrough without block evidence. Corrective path: Document both alternatives, assign ownership, implement the chosen constraint, and rerun block and top-level checks.
Continue practising