Home › Evidence › Records › eng-0166
⚠ This notebook has stopped taking notes: newest record is 23 days old (eng-0280), against an expected cadence of 14 days.

eng-0166

window recorded
citable URL: https://halobench.com/records/eng-0166/ — this address never moves; the anchor /records/#eng-0166 keeps resolving
77.45 Whmean 160.7 W − 10.1 W idle floor (box-idle-all-empty) = 150.6 W attributable
window 2026-08-17T18:36:55Z → 2026-08-17T19:05:50Z · 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
covers run-0424 run-0425
aihydra qwen3-coder-next-80b-fullbench-matrix throughput cell, vulkan d262144 (1 rep(s), 2 runs: prefill+decode pairs), window 18:36:55-19:05:50Z (1735s). Counter interpolated from raw ~10s-resolution HA history samples: start 26.186384 kWh, end 26.263834 kWh. Cell-window energy (server load + all reps together), not per-rep split.

Wh, never joules — read as a cumulative-counter difference across the window; delta over the idle baseline is the only figure that means anything