The configuration, workload, conditions, performed work and checked result, compared with the failing replay. Otherwise the result may not exercise the original failure.

Common trap: Giving a generic fix without identifying the boundary and unmet obligation in this example.

Define the measurement boundary, then compare accepted useful bytes, latency, queue occupancy, command mix, row/bank state, refresh recovery, direction turnarounds and per-traffic-class service against a known-good replay. Diff firmware and operating conditions. Use controlled changes to distinguish scheduler policy, mapping, traffic shape, downstream capacity and interface width; more than one limit may coexist.

Common trap: Quoting peak bandwidth or requesting more bus width.