Matched owner observation

Same rig. Two sounds.
One fair check.

Same exact card, room, meter, distance, and method. Record what changed.

Runs here. Nothing is uploaded. Owner observation only, not a product noise rating.

Quiet Rig Compare

Measure twice.

Record background first. Keep the exact board, shroud, room, case, load, airflow, power, microphone position, and meter method unchanged.

Fixed method: LAeq dBA, 0.5 m, A weighting, slow response, at least 30 seconds.

Exact rig

Exact GPU model Board or OEM number Shroud revision Meter, app, and microphone Calibration note Room note Background LAeq dBA

Setup A

Fan model Fan setting Observed RPM LAeq dBA Duration seconds

Setup B

Fan model Fan setting Observed RPM LAeq dBA Duration seconds

Microphone fixed at 0.5 m A weighting confirmed Slow response confirmed Same case, load, airflow, power, room, and microphone position

Mark any stop observed Fan stopped or surged Unexpected vibration or movement Lead or connector became warm Temperature or physical warning appeared

Use owner-entered observations. This page does not operate the GPU or fan.

Method before winner

The room counts.

Background

Measure the room first. If background reaches either setup reading, recheck.

Position

Keep the microphone at 0.5 m and away from your body. Do not stand between it and the rig.

Instrument

Record the meter, app, microphone, calibration note, A weighting, and slow response.

NIOSH recommends A weighting, slow response, instrument calibration, and at least 30 seconds of averaging for changing noise. Its iOS Sound Level Meter app reports LAeq, but this Quarkpad receipt is still a shop comparison, not compliance or lab certification. Read the NIOSH method. Read the app limits.

Trying to quiet a home AI rig?

Send the exact target.

Quarkpad uses the board label and matched observations to choose what deserves a real bench test.

Open the Fit Desk

Same rig. Two sounds.One fair check.