{ "model_id": "x1", "model_tag": "gemma4:12b-it-q8_0", "box_class": "gpu-3080", "arm": "cpu-box", "arm_label": "cpu-box (CPU-only on the same box, size_vram == 0)", "base_url": "http://127.0.0.1:11439", "started_utc": "2026-09-10T12:51:40Z", "nproc": 20, "num_predict": 256, "prompt_path": "withheld-2026-11-10", "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c", "prompt_bytes": 2101, "scored_runs_requested": 3, "warmup_discarded_runs": 1, "num_gpu_option": null, "wall_decode_note": "decode_tok_s is ollama's own eval_count/eval_duration. decode_tok_s_wall is (eval_count - 1) / (wall_s - ttft_s): the tokens AFTER the first, over the time after the first, measured by this client's clock -- it carries the HTTP hop and the stream decode that ollama's counter does not. decode_tok_s_wall_gross is eval_count / wall_s and therefore includes prefill and any load; it is the slowest and least comparable of the three, printed so nobody has to derive it.", "loadavg_before_arm": "9.61 9.48 8.23 1/812 1008476", "free_m_before": "total used free shared buff/cache available\nMem: 47040 3675 25069 9 18828 43364\nSwap: 24575 969 23606", "runs": [ { "ttft_ms": 1294.82, "first_token_was_thinking": false, "wall_s": 85.6572, "eval_count": 256, "eval_duration_ns": 84359574000, "decode_tok_s": 3.035, "prompt_eval_count": 536, "prompt_eval_duration_ns": 831374000, "prefill_tok_s": 644.716, "load_duration_ns": 459195941, "total_duration_ns": 85654082668, "streamed_chunks": 256, "response_chars": 1042, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 3.023, "decode_tok_s_wall_gross": 2.989, "run": 1 }, { "ttft_ms": 1280.85, "first_token_was_thinking": false, "wall_s": 85.7028, "eval_count": 256, "eval_duration_ns": 84419208000, "decode_tok_s": 3.032, "prompt_eval_count": 536, "prompt_eval_duration_ns": 812593000, "prefill_tok_s": 659.617, "load_duration_ns": 464267753, "total_duration_ns": 85699804720, "streamed_chunks": 256, "response_chars": 1042, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 3.021, "decode_tok_s_wall_gross": 2.987, "run": 2 }, { "ttft_ms": 1288.06, "first_token_was_thinking": false, "wall_s": 85.7298, "eval_count": 256, "eval_duration_ns": 84439085000, "decode_tok_s": 3.032, "prompt_eval_count": 536, "prompt_eval_duration_ns": 825150000, "prefill_tok_s": 649.579, "load_duration_ns": 458633343, "total_duration_ns": 85726773087, "streamed_chunks": 256, "response_chars": 1042, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 3.02, "decode_tok_s_wall_gross": 2.986, "run": 3 } ], "errors": [], "nvidia_before": "0, , NVIDIA GeForce RTX 3080, 256, 11.01, 380.00, 25, 210", "smi_before": [ { "index": "0", "uuid": "", "name": "NVIDIA GeForce RTX 3080", "memory.used": 256.0, "power.draw": 11.01, "power.limit": 380.0, "temperature.gpu": 25.0, "clocks.sm": 210.0 } ], "ollama_version": "0.32.13", "quantization": "Q8_0", "param_size": "11.9B", "family": "gemma4", "context_length": 262144, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "ttft_ms": 1273.02, "first_token_was_thinking": false, "wall_s": 85.4071, "eval_count": 256, "eval_duration_ns": 84131410000, "decode_tok_s": 3.043, "prompt_eval_count": 536, "prompt_eval_duration_ns": 805684000, "prefill_tok_s": 665.273, "load_duration_ns": 462969664, "total_duration_ns": 85404129913, "streamed_chunks": 256, "response_chars": 1042, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 3.031, "decode_tok_s_wall_gross": 2.997 }, "ps_after_load": [ { "name": "gemma4:12b-it-q8_0", "model": "gemma4:12b-it-q8_0", "size": 14429705992, "digest": "41c402fdddc2a87e40a4391caa400961e93786383fbb92a88fc764c333c25322", "details": { "parent_model": "", "format": "gguf", "family": "gemma4", "families": [ "gemma4" ], "parameter_size": "11.9B", "quantization_level": "Q8_0" }, "expires_at": "2026-09-10T13:03:30.499672795Z", "size_vram": 0, "context_length": 4096 } ], "size_total": 14429705992, "size_vram": 0, "gpu": { "arm": "cpu-box", "requested_uuid": "", "requested_index": null, "fits_on_card": false, "card_untouched": true, "size_bytes": 14429705992, "size_vram_bytes": 0, "offload_fraction": 0.0, "offload_split": "0.0% VRAM / 100.0% RAM", "verdict": "cpu-only: size_vram == 0, nothing on the card", "smi_before": [ { "index": "0", "uuid": "", "name": "NVIDIA GeForce RTX 3080", "memory.used": 256.0, "power.draw": 11.01, "power.limit": 380.0, "temperature.gpu": 25.0, "clocks.sm": 210.0 } ], "during": { "sample_hz": 1.0, "samples": 249, "errors": [], "per_gpu": { "": { "index": "0", "name": "NVIDIA GeForce RTX 3080", "power_w": { "min": 11.0, "median": 11.28, "max": 11.63, "mean": 11.29, "n": 249 }, "power_limit_w": 380.0, "memory_used_mib": { "min": 256.0, "median": 256.0, "max": 256.0, "mean": 256.0, "n": 249 }, "temperature_c": { "min": 24.0, "median": 25.0, "max": 25.0, "mean": 24.5, "n": 249 }, "clocks_sm_mhz": { "min": 210.0, "median": 210.0, "max": 210.0, "mean": 210.0, "n": 249 }, "n": 249 } }, "target_uuid": "", "target": { "index": "0", "name": "NVIDIA GeForce RTX 3080", "power_w": { "min": 11.0, "median": 11.28, "max": 11.63, "mean": 11.29, "n": 249 }, "power_limit_w": 380.0, "memory_used_mib": { "min": 256.0, "median": 256.0, "max": 256.0, "mean": 256.0, "n": 249 }, "temperature_c": { "min": 24.0, "median": 25.0, "max": 25.0, "mean": 24.5, "n": 249 }, "clocks_sm_mhz": { "min": 210.0, "median": 210.0, "max": 210.0, "mean": 210.0, "n": 249 }, "n": 249 } }, "smi_after": [ { "index": "0", "uuid": "", "name": "NVIDIA GeForce RTX 3080", "memory.used": 256.0, "power.draw": 11.14, "power.limit": 380.0, "temperature.gpu": 24.0, "clocks.sm": 210.0 } ] }, "scored": true, "peak_runner_rss_kb": 14875800, "peak_runner_rss_gib": 14.19, "peak_runner_threads": 32, "rss_samples": 498, "loadavg_during_runs": { "min": 9.0, "median": 9.92, "max": 10.09, "mean": 9.81, "samples": 498, "nproc": 20 }, "contention_suspected": false, "nvidia_after": "0, , NVIDIA GeForce RTX 3080, 256, 11.14, 380.00, 24, 210", "smi_after": [ { "index": "0", "uuid": "", "name": "NVIDIA GeForce RTX 3080", "memory.used": 256.0, "power.draw": 11.14, "power.limit": 380.0, "temperature.gpu": 24.0, "clocks.sm": 210.0 } ], "free_m_after": "total used free shared buff/cache available\nMem: 47040 18257 392 9 28925 28783\nSwap: 24575 969 23606", "derived": { "decode_tok_s": { "min": 3.032, "median": 3.032, "max": 3.035, "mean": 3.033, "n": 3 }, "decode_tok_s_wall": { "min": 3.02, "median": 3.021, "max": 3.023, "mean": 3.021, "n": 3 }, "ttft_ms": { "min": 1280.85, "median": 1288.06, "max": 1294.82, "mean": 1287.91, "n": 3 }, "wall_s": { "min": 85.6572, "median": 85.7028, "max": 85.7298, "mean": 85.6966, "n": 3 }, "prefill_tok_s": { "min": 644.716, "median": 649.579, "max": 659.617, "mean": 651.304, "n": 3 }, "eval_count": { "min": 256, "median": 256, "max": 256, "mean": 256.0, "n": 3 }, "thinking_chars_total": 0 }, "ttft_note": "Every scored run sends the SAME frozen prompt, so from the second request the KV prefix is cached and derived.ttft_ms is a CACHED-PREFIX first-token time, not a fresh-prompt one. LEDGER.md, 'THE HEADLINE TRAP'.", "ps_final": [], "finished_utc": "2026-09-10T12:57:48Z" }