eng-0141
window recorded
citable URL: https://halobench.com/records/eng-0141/ — this address never moves; the anchor /records/#eng-0141 keeps resolving
1446.6 Whmean 180.82 W − 10.1 W idle floor (box-idle-all-empty) = 170.72 W attributable
window 2026-08-16T16:14:57Z → 2026-08-17T00:14:57Z · provenance recorded
method wall-meter · aihydra energy sensor via Home Assistant (counter-difference on sensor.hardware_ai_hydra_energy, raw ~10s-resolution history samples, linearly interpolated to the run-meta window) · no contention
per task 103.3278 Wh · 3.1308 p at 30.3 p/kWh
covers run-0358
aihydra qwen36-27b-mtp-fullbench tau2 arm1 window (16:14:57Z-00:14:57Z, 28800s, the full 8h TAU2_TIMEOUT safety ceiling — a BOUND-CUT arm, 20/26 tasks scored, 14 passed). wh_per_task is Wh-per-CORRECT-ANSWER for THIS ARM ONLY: 1446.5888 / 14 = 103.3278 Wh, 3.1308 pence at 30.3 p/kWh. WHOLE-SESSION energy (includes simulator wait time end to end), same denominator convention as every other board entry. See clm-0076 for the COMBINED figure across this arm and the continuation (run-0360 / eng-0143): 2055.19 Wh / 16 correct = 128.45 Wh per correct answer — the figure that belongs on the model page, since 26 scored/25-scored is the comparable unit, not this arm alone. Counter interpolated from raw ~10s-resolution HA history samples pulled fresh this session: start 22.4315377590 kWh, end 23.8781265856 kWh.
Wh, never joules — read as a cumulative-counter difference across the window; delta over the idle baseline is the only figure that means anything
Cited by — computed at build time, never stored
claims clm-0076