PCIe/CXL Deep Dive · All levels

Debug, Compliance, and Performance: Tricky Q&A

Senior interview and review questions for Debug, Compliance, and Performance.

Section Q&A bank

Use these drills after completing all topics in Debug, Compliance, and Performance. Answer with workload context, mechanism proof, artifact, owner, and release decision.

What makes a protocol analyzer capture actionable versus overwhelming?

diagram
[INT][PCIE-CXL][DEBUG-COMPLIANCE]

Q: What makes a protocol analyzer capture actionable versus overwhelming?

A:
Actionable captures have precise triggers on error events, include pre/post windows, and correlate with register snapshots. Raw multi-hour traces without filters delay triage and miss first-failure transitions.

FOLLOW-UP TRAP: Collecting full-speed traces without error-qualified triggers.

How do you distinguish retimer issues from endpoint PHY issues in LTSSM debug?

diagram
[INT][PCIE-CXL][DEBUG-COMPLIANCE]

Q: How do you distinguish retimer issues from endpoint PHY issues in LTSSM debug?

A:
Retimer problems often appear at segment boundaries with repeated recovery on specific lanes; endpoint issues persist across retimer bypass tests. A/B topology experiments and retimer FW logs isolate the segment.

FOLLOW-UP TRAP: Replacing endpoint cards without testing retimer firmware or cable plant.

What belongs in a compliance regression gate versus one-time signoff?

diagram
[INT][PCIE-CXL][DEBUG-COMPLIANCE]

Q: What belongs in a compliance regression gate versus one-time signoff?

A:
Electrical and protocol regressions for speed changes, EQ updates, and firmware releases belong in CI-like gates. One-time structural tests may sign off architecture, but any change to PHY/FW needs targeted replay.

FOLLOW-UP TRAP: Treating initial compliance pass as permanent without change-triggered replay.

Why can increasing MPS hurt real workload performance?

diagram
[INT][PCIE-CXL][DEBUG-COMPLIANCE]

Q: Why can increasing MPS hurt real workload performance?

A:
Larger payloads improve peak bandwidth but increase completion splitting complexity, buffer pressure, and latency for small transactions. Mixed workloads may regress when MPS favors bulk DMA over latency-sensitive traffic.

FOLLOW-UP TRAP: Maximizing MPS because benchmark tools report higher GB/s only.

Q&A drill guide

diagram
WORKLOAD -> PCIe/CXL SYMPTOM -> TIMING/QUEUE METRIC -> ROOT CAUSE -> FIX -> REGRESSION

Sketch while answering

diagram
DEBUG CLOSURE LOOP

trigger capture -> hypothesis -> bounded fix -> compliance/perf replay

Key takeaways

  • Always tie controller and PHY counter shifts to application latency and throughput outcomes.

  • Lock firmware timing profile, thermal condition, and DIMM state before comparing PCIe/CXL captures.

Common pitfalls

  • Chasing peak bandwidth while ignoring p99 latency and fairness tails.

  • Changing timing guardbands without separating SI noise from scheduling issues.

  • Declaring closure without reliability gates, fault injection, and regression replay.