eng-0111
window recorded
citable URL: https://halobench.com/records/eng-0111/ — this address never moves; the anchor /records/#eng-0111 keeps resolving
1.66 Whmean 64.87 W − 10.1 W idle floor (box-idle-all-empty) = 54.77 W attributable
window 2026-08-16T02:29:42Z → 2026-08-16T02:31:14Z · provenance recorded
method wall-meter · aihydra energy sensor via Home Assistant (counter-difference on sensor.hardware_ai_hydra_energy, raw ~10s history samples interpolated to the run-meta window) · no contention
covers run-0307
A guard run (4 short probes: coherence/toolcall/needle/isolation), not a reward-scored capability run, so wh_per_task/pence_per_correct_answer do not apply (left null). Counter interpolated from raw ~10s-resolution HA history samples (source=history, not the 5-minute statistics bucket) since the window (92s) is shorter than statistics granularity: start 20.9887294913 kWh, end 20.9903873286 kWh (linear interpolation between bracketing raw samples at each boundary). Window spans the server-load-plus-guard span (matches run-0307's own started_at/ended_at), same convention as eng-0110 (qwen38-27b's guard backfill).
Wh, never joules — read as a cumulative-counter difference across the window; delta over the idle baseline is the only figure that means anything