Correctness: protocol violations, ordering escapes, stale data, ECC or CRC errors.
Performance: throughput, p99 latency, payload efficiency, retry rate, credit stalls.
Bring-up: enumeration success, link width/speed, training margin, boot failure rate.
Verification: compliance pass rate, coverage closure, waiver count, false fail rate.
METRIC GRAPH — latency / throughput / error rate
throughput / success
^
| target
| - - - - - - -
| o after bounded fix
| o
| o baseline
| o failing run
+--------------------------------------> experiment
config A isolated root cause accepted change
Readout:
- compare identical payload, clock, reset, traffic seed, and firmware setup
- separate headline bandwidth from useful payload bandwidth
- explain why the protocol mechanism moved the metric