eng-0252
window reconstructed
citable URL: https://halobench.com/records/eng-0252/ — this address never moves; the anchor /records/#eng-0252 keeps resolving
207.7 Whmean 180.5 W − 10.1 W idle floor (box-idle-all-empty) = 170.4 W attributable
window 2026-08-23T14:19:00Z → 2026-08-23T15:28:00Z · provenance reconstructed
method wall-meter · aihydra energy sensor via Home Assistant (counter-difference on sensor.hardware_ai_hydra_energy, raw ~10s-history samples queried by hborchestrator via the Nabu Casa API and linearly interpolated to the window; power from sensor.hardware_ai_hydra_power) · peak 216.4 W · no contention
HK-001-FULLBENCH FIT + depth-ladder engine band. Counter 34.3666 -> 34.5739 kWh over 14:19-15:28Z (207.7 Wh), the continuous active band covering the FIT c32768 cell (run-0602, pp=279.6/tg=11.73) and the retained ENGINE-path depth-ladder rungs d0/d32768 (run-0603..0604, llama-bench with the PROMPTFORGE sidecar env). Window RECONSTRUCTED from the HA power/energy history (the runner's per-run timestamps live in on-box logs not reachable from this session; retention ~10d so still joinable): I inferred the band edges (active startup 14:19, idle drop 15:28) from the sustained ~180 W draw, and attributed the one continuous engine band to the retained FIT/depth set rather than splitting per-rung. Single isolated active band (idle floor 10.1 W), no contention. cfg-0164 / TheRock 7.14 build e1da26bb8. No Wh/correct (no tau2 in this retained FIT/depth band).
Wh, never joules — read as a cumulative-counter difference across the window; delta over the idle baseline is the only figure that means anything