Protocol whiteboard flow

1. Draw the layers: software, transaction, link/channel, physical/timing.
2. Draw the transaction identity: ID, tag, address, endpoint, lane, or cache line.
3. Mark flow control: ready, grant, credit, FIFO, retry, or link state.
4. Mark ordering: same ID, different ID, barrier, fence, completion, or shareability.
5. State the metric: latency, throughput, error rate, retry, or training margin.
6. State the artifact: waveform, VIP log, analyzer trace, firmware log, or counter.
7. Propose one reduced experiment and one product regression.

Key takeaways

• The best protocol answers are layered, measurable, and tied to a specific transaction.

• Never answer with only a block diagram; carry the failing transaction through the diagram.

Protocol deep dive

Protocol correctness is a layered contract enforced at every boundary.

Concept diagram

intent -> transaction -> channel -> physical

Metric graph

baseline -> experiment -> target

Metrics and artifacts to collect

• latency

• throughput

• error rate

• compliance status

Mini case study

Always anchor debug to the first bad transaction.

Debug branches

• Name layer

• Name ID

• Name artifact

Senior review question

Ask: what is the first transaction that deviates, and which spec rule does it test?

Key takeaways

• Connect every protocol claim to a transaction identity and measurable metric.

• Store the artifact (waveform, log, counter) next to every signoff decision.

Common pitfalls

• Debugging timeouts without finding the first bad transaction.

• Quoting peak bus width without payload efficiency and retry overhead.

• Treating VIP compliance as a substitute for system integration replay.