Home › Evidence › Records › eng-0097
⚠ This notebook has stopped taking notes: newest record is 23 days old (eng-0280), against an expected cadence of 14 days.

eng-0097

window recorded
citable URL: https://halobench.com/records/eng-0097/ — this address never moves; the anchor /records/#eng-0097 keeps resolving
10.99 Whmean 144.4 W − 10.1 W idle floor (box-idle-all-empty) = 134.3 W attributable
window 2026-08-15T14:21:51Z → 2026-08-15T14:26:25Z · provenance recorded
method wall-meter · aihydra energy sensor via Home Assistant (5-minute statistics on sensor.hardware_ai_hydra_energy, linearly interpolated to the run window) · no contention
covers run-0288 run-0289
aihydra deepseek-v4-flash-fullbench throughput matrix cell. Wall-metered via HA sensor.hardware_ai_hydra_energy, joined from the permanent 5-minute statistics series (linearly interpolated to the exact run-meta window boundaries), not raw 10s history -- the matrix's ~4h span exceeded the history query's row cap by the time of ingest. Baseline 10.1 W is the repo's dated idle-aihydra-2026-08 box-idle-all-empty record.

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

model pages deepseek-v4-flash