{ "model_id": "m1", "model_tag": "qwen3.5:9b", "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": "2026-09-10T12:29:43Z", "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": "2.58 3.52 3.85 1/808 933087", "free_m_before": "total used free shared buff/cache available\nMem: 47040 6579 1350 9 39645 40461\nSwap: 24575 991 23584", "runs": [ { "ttft_ms": 518.77, "first_token_was_thinking": false, "wall_s": 37.2799, "eval_count": 256, "eval_duration_ns": 36751248000, "decode_tok_s": 6.966, "prompt_eval_count": 537, "prompt_eval_duration_ns": 181698000, "prefill_tok_s": 2955.454, "load_duration_ns": 333609428, "total_duration_ns": 37269757224, "streamed_chunks": 235, "response_chars": 1033, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 6.937, "decode_tok_s_wall_gross": 6.867, "run": 1 }, { "ttft_ms": 544.48, "first_token_was_thinking": false, "wall_s": 37.3296, "eval_count": 256, "eval_duration_ns": 36782940000, "decode_tok_s": 6.96, "prompt_eval_count": 537, "prompt_eval_duration_ns": 196137000, "prefill_tok_s": 2737.882, "load_duration_ns": 344542893, "total_duration_ns": 37327177393, "streamed_chunks": 235, "response_chars": 1033, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 6.932, "decode_tok_s_wall_gross": 6.858, "run": 2 }, { "ttft_ms": 518.79, "first_token_was_thinking": false, "wall_s": 37.3317, "eval_count": 256, "eval_duration_ns": 36810810000, "decode_tok_s": 6.954, "prompt_eval_count": 537, "prompt_eval_duration_ns": 181174000, "prefill_tok_s": 2964.001, "load_duration_ns": 333841817, "total_duration_ns": 37329349089, "streamed_chunks": 235, "response_chars": 1033, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 6.927, "decode_tok_s_wall_gross": 6.857, "run": 3 } ], "errors": [], "nvidia_before": "0, , NVIDIA GeForce RTX 3080, 256, 11.86, 380.00, 47, 210", "smi_before": [ { "index": "0", "uuid": "", "name": "NVIDIA GeForce RTX 3080", "memory.used": 256.0, "power.draw": 11.86, "power.limit": 380.0, "temperature.gpu": 47.0, "clocks.sm": 210.0 } ], "ollama_version": "0.32.13", "quantization": "Q4_K_M", "param_size": "9.7B", "family": "qwen35", "context_length": 262144, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "ttft_ms": 524.25, "first_token_was_thinking": false, "wall_s": 37.2225, "eval_count": 256, "eval_duration_ns": 36687881000, "decode_tok_s": 6.978, "prompt_eval_count": 537, "prompt_eval_duration_ns": 180496000, "prefill_tok_s": 2975.135, "load_duration_ns": 339923030, "total_duration_ns": 37211877898, "streamed_chunks": 235, "response_chars": 1033, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 6.949, "decode_tok_s_wall_gross": 6.878 }, "ps_after_load": [ { "name": "qwen3.5:9b", "model": "qwen3.5:9b", "size": 6071601067, "digest": "6488c96fa5faab64bb65cbd30d4289e20e6130ef535a93ef9a49f42eda893ea7", "details": { "parent_model": "", "format": "gguf", "family": "qwen35", "families": [ "qwen35" ], "parameter_size": "9.7B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-09-10T12:40:36.186459678Z", "size_vram": 0, "context_length": 4096 } ], "size_total": 6071601067, "size_vram": 0, "gpu": { "arm": "cpu-box", "requested_uuid": "", "requested_index": null, "fits_on_card": false, "card_untouched": true, "size_bytes": 6071601067, "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": 11.86, "power.limit": 380.0, "temperature.gpu": 47.0, "clocks.sm": 210.0 } ], "during": { "sample_hz": 1.0, "samples": 109, "errors": [], "per_gpu": { "": { "index": "0", "name": "NVIDIA GeForce RTX 3080", "power_w": { "min": 11.29, "median": 11.71, "max": 12.02, "mean": 11.69, "n": 109 }, "power_limit_w": 380.0, "memory_used_mib": { "min": 256.0, "median": 256.0, "max": 256.0, "mean": 256.0, "n": 109 }, "temperature_c": { "min": 36.0, "median": 39.0, "max": 42.0, "mean": 38.9, "n": 109 }, "clocks_sm_mhz": { "min": 210.0, "median": 210.0, "max": 210.0, "mean": 210.0, "n": 109 }, "n": 109 } }, "target_uuid": "", "target": { "index": "0", "name": "NVIDIA GeForce RTX 3080", "power_w": { "min": 11.29, "median": 11.71, "max": 12.02, "mean": 11.69, "n": 109 }, "power_limit_w": 380.0, "memory_used_mib": { "min": 256.0, "median": 256.0, "max": 256.0, "mean": 256.0, "n": 109 }, "temperature_c": { "min": 36.0, "median": 39.0, "max": 42.0, "mean": 38.9, "n": 109 }, "clocks_sm_mhz": { "min": 210.0, "median": 210.0, "max": 210.0, "mean": 210.0, "n": 109 }, "n": 109 } }, "smi_after": [ { "index": "0", "uuid": "", "name": "NVIDIA GeForce RTX 3080", "memory.used": 256.0, "power.draw": 11.54, "power.limit": 380.0, "temperature.gpu": 36.0, "clocks.sm": 210.0 } ] }, "scored": true, "peak_runner_rss_kb": 6966036, "peak_runner_rss_gib": 6.64, "peak_runner_threads": 32, "rss_samples": 218, "loadavg_during_runs": { "min": 6.56, "median": 8.56, "max": 9.08, "mean": 8.3, "samples": 218, "nproc": 20 }, "contention_suspected": false, "nvidia_after": "0, , NVIDIA GeForce RTX 3080, 256, 11.54, 380.00, 36, 210", "smi_after": [ { "index": "0", "uuid": "", "name": "NVIDIA GeForce RTX 3080", "memory.used": 256.0, "power.draw": 11.54, "power.limit": 380.0, "temperature.gpu": 36.0, "clocks.sm": 210.0 } ], "free_m_after": "total used free shared buff/cache available\nMem: 47040 13304 402 9 33875 33736\nSwap: 24575 978 23597", "derived": { "decode_tok_s": { "min": 6.954, "median": 6.96, "max": 6.966, "mean": 6.96, "n": 3 }, "decode_tok_s_wall": { "min": 6.927, "median": 6.932, "max": 6.937, "mean": 6.932, "n": 3 }, "ttft_ms": { "min": 518.77, "median": 518.79, "max": 544.48, "mean": 527.35, "n": 3 }, "wall_s": { "min": 37.2799, "median": 37.3296, "max": 37.3317, "mean": 37.3137, "n": 3 }, "prefill_tok_s": { "min": 2737.882, "median": 2955.454, "max": 2964.001, "mean": 2885.779, "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": "2026-09-10T12:32:28Z" }