File size: 1,188 Bytes
4f03424 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 | {
"power": "AC",
"low_power_mode": false,
"thermal_warning_reported": false,
"other_model_service": "llama-server present, observed 0.1% CPU, approximately 15.6 GiB physical footprint",
"caveat": "Desktop-session benchmark; other applications remain open. Not an isolated laboratory run.",
"idle_sleep": "Temporarily inhibited for benchmark process only",
"phase_offload": true,
"swap_before_formal_run_gib": 19.60015869140625,
"swap_note": "Swap allocated during earlier resident diagnostics; per-image swap deltas are recorded separately.",
"other_model_services": "Other pre-existing Python and llama-server services left running; earlier physical footprints approximately 25.5 and 15.6 GiB.",
"mid_run_observation": "At approximately 23:35 PDT both pre-existing model services were still present and around 0.1% CPU; pmset reported no thermal or performance warning. GPU contention and thermal state were not continuously instrumented.",
"timing_variability": "Q8 texture was substantially faster than its preceding Q8 cases despite unchanged runtime; do not interpret the aggregate latency difference as a controlled causal estimate of quantization speed."
}
|