{ "model_id": "m6", "model_tag": "nemotron-3.5-lightning:30b-a3b", "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": "2026-09-10T12:50:02Z", "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": "9.74 9.58 8.13 2/812 992200", "free_m_before": "total used free shared buff/cache available\nMem: 47040 3779 18962 9 24832 43261\nSwap: 24575 969 23606", "runs": [ { "ttft_ms": 374.67, "first_token_was_thinking": false, "wall_s": 19.9661, "eval_count": 256, "eval_duration_ns": 19590341000, "decode_tok_s": 13.068, "prompt_eval_count": 532, "prompt_eval_duration_ns": 157701000, "prefill_tok_s": 3373.473, "load_duration_ns": 214737690, "total_duration_ns": 19964763206, "streamed_chunks": 245, "response_chars": 1029, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 13.016, "decode_tok_s_wall_gross": 12.822, "run": 1 }, { "ttft_ms": 372.31, "first_token_was_thinking": false, "wall_s": 20.0151, "eval_count": 256, "eval_duration_ns": 19641515000, "decode_tok_s": 13.034, "prompt_eval_count": 532, "prompt_eval_duration_ns": 157765000, "prefill_tok_s": 3372.104, "load_duration_ns": 212257495, "total_duration_ns": 20013684808, "streamed_chunks": 245, "response_chars": 1029, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 12.982, "decode_tok_s_wall_gross": 12.79, "run": 2 }, { "ttft_ms": 375.05, "first_token_was_thinking": false, "wall_s": 19.982, "eval_count": 256, "eval_duration_ns": 19605747000, "decode_tok_s": 13.057, "prompt_eval_count": 532, "prompt_eval_duration_ns": 159671000, "prefill_tok_s": 3331.851, "load_duration_ns": 213181281, "total_duration_ns": 19980612671, "streamed_chunks": 245, "response_chars": 1029, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 13.006, "decode_tok_s_wall_gross": 12.812, "run": 3 } ], "errors": [], "nvidia_before": "0, , NVIDIA GeForce RTX 3080, 256, 11.12, 380.00, 25, 210", "smi_before": [ { "index": "0", "uuid": "", "name": "NVIDIA GeForce RTX 3080", "memory.used": 256.0, "power.draw": 11.12, "power.limit": 380.0, "temperature.gpu": 25.0, "clocks.sm": 210.0 } ], "ollama_version": "0.32.13", "quantization": "Q4_K_M", "param_size": "32.9B", "family": "nemotron_h_moe", "context_length": 1048576, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "ttft_ms": 374.8, "first_token_was_thinking": false, "wall_s": 19.6308, "eval_count": 256, "eval_duration_ns": 19254883000, "decode_tok_s": 13.295, "prompt_eval_count": 532, "prompt_eval_duration_ns": 157131000, "prefill_tok_s": 3385.71, "load_duration_ns": 215261659, "total_duration_ns": 19629317580, "streamed_chunks": 245, "response_chars": 1029, "thinking_chars": 0, "done_reason": "length", "decode_tok_s_wall": 13.243, "decode_tok_s_wall_gross": 13.041 }, "ps_after_load": [ { "name": "nemotron-3.5-lightning:30b-a3b", "model": "nemotron-3.5-lightning:30b-a3b", "size": 25509484296, "digest": "e7a64ff15fb174c42b4f463e5c888c4f2c7b9cabf9e8d65a1c0874405426c1b2", "details": { "parent_model": "", "format": "gguf", "family": "nemotron_h_moe", "families": [ "nemotron_h_moe" ], "parameter_size": "32.9B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-09-10T13:00:39.479475663Z", "size_vram": 0, "context_length": 4096 } ], "size_total": 25509484296, "size_vram": 0, "gpu": { "arm": "cpu-box", "requested_uuid": "", "requested_index": null, "fits_on_card": false, "card_untouched": true, "size_bytes": 25509484296, "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": 11.12, "power.limit": 380.0, "temperature.gpu": 25.0, "clocks.sm": 210.0 } ], "during": { "sample_hz": 1.0, "samples": 59, "errors": [], "per_gpu": { "": { "index": "0", "name": "NVIDIA GeForce RTX 3080", "power_w": { "min": 10.94, "median": 11.17, "max": 11.39, "mean": 11.17, "n": 59 }, "power_limit_w": 380.0, "memory_used_mib": { "min": 256.0, "median": 256.0, "max": 256.0, "mean": 256.0, "n": 59 }, "temperature_c": { "min": 25.0, "median": 25.0, "max": 25.0, "mean": 25.0, "n": 59 }, "clocks_sm_mhz": { "min": 210.0, "median": 210.0, "max": 210.0, "mean": 210.0, "n": 59 }, "n": 59 } }, "target_uuid": "", "target": { "index": "0", "name": "NVIDIA GeForce RTX 3080", "power_w": { "min": 10.94, "median": 11.17, "max": 11.39, "mean": 11.17, "n": 59 }, "power_limit_w": 380.0, "memory_used_mib": { "min": 256.0, "median": 256.0, "max": 256.0, "mean": 256.0, "n": 59 }, "temperature_c": { "min": 25.0, "median": 25.0, "max": 25.0, "mean": 25.0, "n": 59 }, "clocks_sm_mhz": { "min": 210.0, "median": 210.0, "max": 210.0, "mean": 210.0, "n": 59 }, "n": 59 } }, "smi_after": [ { "index": "0", "uuid": "", "name": "NVIDIA GeForce RTX 3080", "memory.used": 256.0, "power.draw": 11.0, "power.limit": 380.0, "temperature.gpu": 25.0, "clocks.sm": 210.0 } ] }, "scored": true, "peak_runner_rss_kb": 25286664, "peak_runner_rss_gib": 24.12, "peak_runner_threads": 32, "rss_samples": 117, "loadavg_during_runs": { "min": 8.95, "median": 9.36, "max": 9.61, "mean": 9.34, "samples": 117, "nproc": 20 }, "contention_suspected": false, "nvidia_after": "0, , NVIDIA GeForce RTX 3080, 256, 11.00, 380.00, 25, 210", "smi_after": [ { "index": "0", "uuid": "", "name": "NVIDIA GeForce RTX 3080", "memory.used": 256.0, "power.draw": 11.0, "power.limit": 380.0, "temperature.gpu": 25.0, "clocks.sm": 210.0 } ], "free_m_after": "total used free shared buff/cache available\nMem: 47040 28402 347 9 18824 18637\nSwap: 24575 969 23606", "derived": { "decode_tok_s": { "min": 13.034, "median": 13.057, "max": 13.068, "mean": 13.053, "n": 3 }, "decode_tok_s_wall": { "min": 12.982, "median": 13.006, "max": 13.016, "mean": 13.001, "n": 3 }, "ttft_ms": { "min": 372.31, "median": 374.67, "max": 375.05, "mean": 374.01, "n": 3 }, "wall_s": { "min": 19.9661, "median": 19.982, "max": 20.0151, "mean": 19.9877, "n": 3 }, "prefill_tok_s": { "min": 3331.851, "median": 3372.104, "max": 3373.473, "mean": 3359.143, "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": "2026-09-10T12:51:40Z" }