eng-0133
window recorded
citable URL: https://halobench.com/records/eng-0133/ — this address never moves; the anchor /records/#eng-0133 keeps resolving
17.44 Whmean 123.85 W − 10.1 W idle floor (box-idle-all-empty) = 113.75 W attributable
window 2026-08-16T11:22:45Z → 2026-08-16T11:31:12Z · 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 3.4886 Wh · 0.1057 p at 30.3 p/kWh
covers run-0331
Backfilled batch join for the 2026-08-16 ornith-35b screen's 5-task tau2 SMOKE (ROCm, cfg-0093) -- this was flagged energy:null in the candidate record and clm-0068, joined within the ~10-day HA retention window during this job's follow-up fullbench session. wh_per_task is Wh-per-correct-answer: wh_total / tasks_passed (5 of 5) = 3.4886 Wh, 0.1057 pence at the standing 30.3 p/kWh tariff -- a 5-task SMOKE window, not comparable to the full 26-task standard (see eng-0134). Counter interpolated from raw ~10s-resolution HA history samples: start 21.9845072601 kWh, end 22.0019500298 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