eng-0079
window recorded
citable URL: https://halobench.com/records/eng-0079/ — this address never moves; the anchor /records/#eng-0079 keeps resolving
81.26 Whmean 147.2 W − 10.1 W idle floor (box-idle-all-empty) = 137.1 W attributable
window 2026-08-14T18:27:11Z → 2026-08-14T19:00:18Z · provenance recorded
method wall-meter · aihydra energy sensor via Home Assistant (counter-difference on sensor.hardware_ai_hydra_energy) · no contention
covers run-0269
aihydra run-meta.jsonl entries mtp-vulkan-Q8_0-nmaxoff-c32768 (18:27:11-18:43:44) + mtp-vulkan-Q8_0-nmax2-c32768 (18:43:47-18:52:12) + mtp-vulkan-Q8_0-nmax3-c32768 (18:52:15-19:00:18), suite qwen38-screen, cfg-0068 — the first sweep pass, matching run-0269s stated span floor..n3. The n_max>=4 cells that follow in the raw sweep are EXCLUDED from this window: they are the degenerate EOS-cliff cells (clm-0055) and are not part of the published run. See clm-0055 for the per-n_max Wh/1000-tokens breakdown computed from the nmaxoff/nmax3 sub-windows. Retroactive join computed 2026-08-15.
Wh, never joules — read as a cumulative-counter difference across the window; delta over the idle baseline is the only figure that means anything