Lacksan community

Lacksan HP Performance Forum

Workstation tuning, measured and written down.

Read-only preview

Forum Rules

Posting standard: how to report a result so it survives review

Pinned Locked Started · 1 replies · 903 views

  1. Documented fact

    If you are posting a number, the post needs to answer these questions before anyone can use it.

    • Exact machine: model, generation, and where relevant the SKU.
    • Windows build number, not the marketing version.
    • Power source and power plan during the run.
    • Thermal starting condition. A cold first run and a fourth consecutive run are different experiments.
    • What changed, expressed as the specific setting, key or firmware option.
    • How many runs, and the median rather than the best one.
    • How to put it back.

    The last point is not optional. A change you cannot reverse is not a tuning change, it is damage with a pleasant side effect.

    Machine:      HP ZBook Firefly 14 G8
    Windows:      build 26200
    Power:        AC, plan not recorded
    Change:       <setting>
    Runs:         n = 5, median reported
    Before/After: <value> -> <value>
    Rollback:     <exact steps>
  2. Documented fact

    Two further rules that come up constantly.

    First: do not post a registry key without saying what it does and which component reads it. A key with no documented consumer is folklore.

    Second: a single sample is a single sample. It can direct the next experiment. It cannot establish that anything got faster. This applies to Lacksan's own published results as much as to anyone else's.

This thread is locked. It is reference material rather than an open discussion.