Top GLS Interview Questions

Chapter 15 · Section 15.1 · Interview & Signoff Review Preparation

Project thread — this finale synthesizes the whole book (single flip-flop → mini-SoC) into interview readiness. 15.2 does scenarios, 15.3 drills, 15.4 the signoff checklist, 15.5 a readiness self-assessment.

1. Why Should I Learn This?

Interviews test whether you understand GLS's role — and the differentiator is boundary awareness.

• The top questions cluster into nine themes across the book.
• Strong answers are structured and name what GLS is not (STA/LEC/CDC/coverage).
• Knowing the boundaries is what marks a senior GLS engineer.

This is the conceptual half of interview prep (15.2 scenarios, 15.3 drills).

2. Real Silicon Story — the candidate who knew the boundaries

Two candidates answered "why run GLS if you have STA?" One said "GLS checks timing too" — a shallow, wrong-boundary answer. The other said "STA is the exhaustive timing signoff; GLS is a dynamic run that catches what STA can't — X/reset, uninitialized state, timing-dependent functional behavior, low-power, DFT — and it's not a timing signoff."

The second candidate got the offer. Same topic, but one knew the boundary (GLS ≠ STA) and could name what each tool owns. Boundary awareness signaled real understanding.

Lesson: the top GLS questions are boundary questions in disguise. A strong answer names what GLS is for and what it isn't — that's what interviewers listen for.

3. Concept — the nine question themes

(1) Why GLS / vs other tools:

• What does GLS catch that RTL/STA/LEC/CDC can't? (X/reset/init, timing-dependent functional, low-power, DFT). Boundaries: STA = timing signoff, LEC = equivalence, static CDC/MTBF = CDC, RTL/UVM = coverage.

(2) Artifacts:

• Netlist, .lib (synth/STA) vs Verilog cell model (GLS runs), UDP, specify arcs, SDF (timing, verify annotation), SPEF (parasitics → delay calc → SDF).

(3) X-propagation:

• X sources; pessimism vs optimism; RTL hides / GLS exposes (synthesis doesn't preserve X semantics); X-flow (controlling values); xprop.

(4) Timing:

• Setup (slow corner) vs hold (fast corner, period-independent); MIN:TYP:MAX; glitches (inertial); real-vs-artifact.

(5) Reset:

• Async (immediate) vs sync (at edge); recovery/removal (release timing); reset signoff (every state flop reset, no X escapes); not every flop needs reset.

(6) CDC/metastability:

• Simulators can't model metastability (X flags risk); synchronizers under real delays; gray-code FIFOs; GLS complements static CDC/MTBF.

(7) Low-power:

• UPF (power intent); power-aware GLS (corruption); isolation/retention; sequencing; functional not power-integrity.

(8) DFT:

• Scan flops/chains; shift/capture; ATPG pattern GLS (simulate, not coverage); test-mode reset/X.

(9) Methodology:

• The funnel (real-vs-artifact first); first divergence; force/deposit (debug aids not fixes); common mistakes.

The meta-trait (accuracy):

• Boundary awareness — every strong answer names what GLS is and what it isn't. GLS is dynamic; STA/LEC/static-CDC/ATPG own their signoffs (0.3/9.1/11.1).

4. Mental Model — answer like a map, not a fact

A strong GLS answer is a map (where this tool sits), not an isolated fact.

• A fact answer: "GLS checks the netlist." (True but shallow — no boundaries.)
• A map answer: "GLS is the dynamic gate-level check for what RTL/STA/LEC/CDC can't see — X/reset, timing-dependent functional, low-power, DFT — run alongside STA (timing signoff), LEC (equivalence), static CDC/MTBF (CDC), and RTL/UVM (coverage)."
• The map answer shows you understand the terrain — where GLS is, where the neighbors are, and the borders.
• Interviewers hire the cartographer, not the fact-reciter.

Answer with the map: what GLS is, what it isn't, and where the borders are.

5. Working Example — a curated top-questions list

A curated set (with the one-line boundary each answer must hit):

# TOP GLS INTERVIEW QUESTIONS - REPRESENTATIVE (each answer must name the BOUNDARY):
# WHY / vs tools:
#   Q: Why run GLS if you have STA?     -> A: GLS=dynamic, catches X/reset/functional-timing/LP/DFT; STA=timing signoff
#   Q: GLS vs LEC?                       -> A: LEC=exhaustive equivalence (formal); GLS=dynamic sim (X/timing/etc.)
#   Q: Does GLS verify CDC?              -> A: complements static CDC/MTBF; can't model metastability (flags risk)
# ARTIFACTS:
#   Q: .lib vs Verilog cell model?       -> A: .lib=synth/STA; GLS runs the Verilog model; SDF bridges timing
#   Q: What does SDF carry?              -> A: timing only (IOPATH/INTERCONNECT/checks, MIN:TYP:MAX); not logic
# X:
#   Q: Why does RTL hide bugs GLS finds? -> A: RTL X-optimism; synthesis doesn't preserve X semantics
# TIMING / RESET:
#   Q: Fix a hold violation?             -> A: add delay / fix skew (period-independent); NOT slow the clock
#   Q: Why sync async reset RELEASE?     -> A: recovery/removal metastability at deassertion
# CDC / LP / DFT / METHOD:
#   Q: Can sim model metastability?      -> A: NO; X flags the risk; MTBF verifies survival
#   Q: What does power-aware GLS check?   -> A: power intent FUNCTIONALLY (corruption/iso/retention), not power integrity
#   Q: Pattern GLS vs ATPG coverage?      -> A: GLS confirms patterns SIMULATE; ATPG measures fault COVERAGE
#   Q: First step debugging gate X?       -> A: real bug or artifact? then first divergence -> classify -> trace

Practical context (representative, tool-neutral):

# How to answer a GLS question (tool-neutral):
#   1) state what GLS DOES for this topic (the dynamic gate-level value)
#   2) name the BOUNDARY -- what it ISN'T (STA=timing, LEC=equivalence, static-CDC/MTBF=CDC, ATPG=coverage)
#   3) give a crisp example (a waveform/behavior)
#   4) close with the takeaway
# The differentiator is (2): boundary awareness. Shallow answers skip it.

A canonical waveform an interviewer might show (power-up X → reset), as a real waveform:

6. Debugging Session — a shallow answer vs a boundary-aware one

7. Common Mistakes

• Answering with facts, not boundaries. Name what GLS isn't (STA/LEC/CDC/coverage).
• "GLS checks timing too." GLS is a dynamic spot-check; STA is timing signoff (0.3).
• "RTL is always golden." RTL X-optimism — a GLS X can be right (6.4).
• "No X = CDC/metastability fine." Simulators can't model metastability (9.2).
• Treating an expected X (corruption/notifier) as a bug in an answer (12.6).

8. Industry Best Practices

• Answer structured (does / isn't / example / takeaway).
• Always name the boundary — GLS vs STA/LEC/static-CDC/coverage.
• Recognize expected X (corruption/notifier/crossing) vs real bugs (12.6).
• Use crisp canonical examples (power-up X, hold-is-period-independent, RTL-hides-X).
• Practice the nine themes — you'll be asked across them.

Senior Engineer Thinking

• Beginner: "GLS simulates the netlist to check it works."
• Senior: "GLS is the dynamic gate-level check for what RTL/STA/LEC/CDC can't see — X/reset, timing-dependent functional, low-power, DFT — run alongside them, each owning its signoff. The boundary is the answer."

The senior answers with the map (what GLS is and isn't), not an isolated fact.

Silicon Impact

The reason interviewers probe boundaries is that misplacing GLS's role ships bugs — the same mistakes as the 12.6 catalog, in verbal form: treating GLS as timing signoff (skipping STA, 0.3), dismissing a real X as "RTL is golden" (6.4), trusting a clean sim for metastability (9.2), calling an expected corruption X a bug (12.6). A candidate who knows the boundaries in an interview is the engineer who won't make those mistakes on the job — which is exactly why boundary awareness is the strongest hiring signal. This lesson makes the top questions explicit as boundary questions, so preparing for the interview and preparing to do the job well are the same preparation. The best GLS answer and the best GLS practice are both: name what GLS is, and what it isn't.

Engineering Checklist

• Can answer the nine themes structured (does / isn't / example / takeaway).
• Always name the boundary (GLS vs STA/LEC/static-CDC/coverage).
• Recognize expected X vs real bugs (12.6).
• Have crisp canonical examples ready (power-up X, hold period-independence, RTL-hides-X).
• Practiced the flagship questions (why-vs-STA, hold fix, metastability, RTL-golden).

Try Yourself

1. Take each of the nine themes and write a structured, boundary-aware answer to its flagship question.
2. Observe: every strong answer names what GLS isn't (STA/LEC/CDC/coverage) — that's the differentiator.
3. Change: for a question you answered with a bare fact, add the boundary and a crisp example.
4. Expect: the answer now sounds senior — a map, not a fact. Practice out loud until the boundary is reflexive.

Interview prep is a discipline, tool-independent. No paid tool required.

Interview Perspective

• Weak: "GLS checks the gate-level netlist works."
• Good: "GLS catches X/reset, timing-dependent behavior, low-power, and DFT that RTL and STA miss."
• Senior: "GLS is the dynamic gate-level check for what nothing else sees — X/reset/init, timing-dependent functional behavior, low-power, DFT — run alongside STA (timing signoff), LEC (equivalence), static CDC/MTBF (CDC), and RTL/UVM (coverage). Every strong GLS answer names what it is and what it isn't; boundary awareness is the whole game."

9. Interview / Review Questions

10. Key Takeaways

• The top GLS interview questions cluster into nine themes: why/vs-tools, artifacts, X-propagation, timing, reset, CDC/metastability, low-power, DFT, and methodology — spanning the whole book.
• The single trait that marks a strong answer is boundary awareness — naming what GLS is and what it isn't (STA = timing signoff, LEC = equivalence, static CDC/MTBF = CDC, RTL/UVM = coverage).
• Answer structured: state what GLS does, name the boundary, give a crisp example, close with the takeaway.
• The top questions are boundary questions in disguise — a shallow answer ("GLS checks timing too," "RTL is always golden," "no X = CDC fine") is wrong on the border, exactly where interviewers probe.
• Boundary awareness in the interview = the engineer who won't misplace GLS's role on the job — so interview prep and doing the job well are the same preparation; GLS is the dynamic gate-level input (0.3). Next: 15.2 — scenario-based GLS questions.

Quick Revision

Top GLS interview Qs = nine themes: why/vs-tools · artifacts · X-propagation · timing · reset · CDC/metastability · low-power · DFT · methodology. The differentiator: BOUNDARY AWARENESS — name what GLS is AND isn't (STA=timing signoff, LEC=equivalence, static-CDC/MTBF=CDC, ATPG/UVM=coverage). Answer structured: does / isn't / example / takeaway. Shallow answers miss the border; senior answers name it. Next: 15.2 — scenario-based GLS questions.