Active vs Passive Agent Modes: Reports and Metrics

Reports and metrics

Reports and Metrics for Active vs Passive Agent Modes focuses on stimulus/check independence score and dual-mode regression pass rate. The purpose is to turn memory observations into mechanism-backed actions with explicit owners and release-safe validation.

Reports should explain why stimulus/check independence score and dual-mode regression pass rate moved, not simply that it moved. Require evidence that links the movement to command behavior, queue policy, PHY margin, or reliability controls.

Before/after trend

BEFORE / AFTER GRAPH - Active vs Passive Agent Modes

metric quality
  ^
  |                       o target band
  |                o post-fix sweep
  |           o
  |      o baseline (failing)
  +----------------------------------------------> iteration
      evidence capture   fix applied   closure run

Use this view to prove improvement is causal, not accidental.

Evidence matrix

VIP EVIDENCE MATRIX - Active vs Passive Agent Modes

+-------------------------------+--------------------------------+--------------------------------+---------------------------+
| Evidence                      | Tells you                      | Does not prove                 | Next action               |
+-------------------------------+--------------------------------+--------------------------------+---------------------------+
| checker hit/miss + ACT/PRE mix    | locality and row-state cost    | lane-level capture integrity   | inspect training margins  |
| queue age + class breakdown   | fairness and starvation risk   | command legality details       | parse command timeline    |
| spec clause legality + bus timeline | timing-window pressure         | root cause by itself           | correlate with traffic map|
| eye / Vref / skew snapshots   | PHY margin and drift behavior  | controller policy quality      | pair with schedule logs   |
| CE/UE + scrub telemetry       | reliability trajectory         | immediate perf bottleneck only | map to hotspot addresses  |
+-------------------------------+--------------------------------+--------------------------------+---------------------------+

• Track p50/p95/p99 latency and effective bandwidth together.

• Include command and queue context alongside high-level counters.

• Tag reports with firmware, timing profile, and thermal state.

• Call out contradictory evidence instead of hiding it.

VIP deep dive

Reusable VIP layering, agent roles, monitor/scoreboard contracts, and packaging patterns that scale across protocols and projects.

Concept diagram

VIP SECTION - VIP Architecture & Packaging

testcase -> agents -> checkers -> coverage -> evidence

Metric graph

checker noise vs real violations trend

Reports and artifacts

• checker hit report

• coverage closure sheet

• compliance trace matrix

• regression health snapshot

Mini case study

A profile drift caused false checker storms until configuration hashes were locked in CI.

Debug branches

• Reproduce with locked seed and profile

• Isolate checker vs scoreboard vs DUT paths

• Map failure to spec clause and owner

Senior review question

Ask: which latency, bandwidth, and reliability evidence proves this VIP topic is closed under real traffic?

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 VIP 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.

VIP atlas notes

Active vs Passive Agent Modes should be read as an end-to-end VIP behavior, not as a single block definition. Production compliance closure reflects interactions between agents, checkers, coverage, and customer evidence before tapeout or IP release claims.

Active agents generate protocol-correct traffic while passive agents observe and check without driving the bus. Mode selection affects who owns stimulus, how monitors attach, and whether compliance suites can run against third-party DUT environments without agent contention. VIP inefficiency is multiplicative: one weak checker enable, one hollow coverage bin, or one non-reproducible failure repeated across regressions can dominate signoff risk.